National Academies Press: OpenBook
« Previous: Overview and Summary
Page 34
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 34
Page 35
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 35
Page 36
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 36
Page 37
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 37
Page 38
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 38
Page 39
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 39
Page 40
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 40
Page 41
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 41
Page 42
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 42
Page 43
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 43
Page 44
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 44
Page 45
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 45
Page 46
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 46
Page 47
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 47
Page 48
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 48
Page 49
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 49
Page 50
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 50
Page 51
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 51
Page 52
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 52
Page 53
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 53
Page 54
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 54
Page 55
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 55
Page 56
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 56
Page 57
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 57
Page 58
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 58
Page 59
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 59
Page 60
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 60
Page 61
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 61
Page 62
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 62
Page 63
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 63
Page 64
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 64
Page 65
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 65
Page 66
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 66
Page 67
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 67
Page 68
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 68
Page 69
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 69
Page 70
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 70
Page 71
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 71
Page 72
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 72
Page 73
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 73
Page 74
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 74
Page 75
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 75
Page 76
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 76
Page 77
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 77
Page 78
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 78
Page 79
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 79
Page 80
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 80
Page 81
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 81
Page 82
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 82
Page 83
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 83
Page 84
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 84
Page 85
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 85
Page 86
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 86
Page 87
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 87
Page 88
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 88
Page 89
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 89
Page 90
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 90
Page 91
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 91
Page 92
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 92
Page 93
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 93
Page 94
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 94
Page 95
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 95
Page 96
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 96
Page 97
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 97
Page 98
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 98
Page 99
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 99
Page 100
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 100
Page 101
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 101
Page 102
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 102
Page 103
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 103
Page 104
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 104
Page 105
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 105
Page 106
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 106
Page 107
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 107
Page 108
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 108
Page 109
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 109
Page 110
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 110
Page 111
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 111
Page 112
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 112
Page 113
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 113
Page 114
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 114
Page 115
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 115
Page 116
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 116
Page 117
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 117
Page 118
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 118
Page 119
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 119
Page 120
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 120
Page 121
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 121
Page 122
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 122
Page 123
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 123
Page 124
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 124
Page 125
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 125
Page 126
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 126
Page 127
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 127
Page 128
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 128
Page 129
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 129
Page 130
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 130
Page 131
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 131
Page 132
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 132
Page 133
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 133
Page 134
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 134
Page 135
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 135
Page 136
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 136
Page 137
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 137
Page 138
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 138
Page 139
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 139
Page 140
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 140
Page 141
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 141
Page 142
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 142
Page 143
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 143
Page 144
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 144
Page 145
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 145
Page 146
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 146
Page 147
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 147
Page 148
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 148
Page 149
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 149
Page 150
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 150
Page 151
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 151
Page 152
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 152
Page 153
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 153
Page 154
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 154
Page 155
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 155
Page 156
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 156
Page 157
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 157
Page 158
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 158
Page 159
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 159
Page 160
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 160
Page 161
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 161
Page 162
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 162
Page 163
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 163
Page 164
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 164
Page 165
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 165
Page 166
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 166
Page 167
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 167
Page 168
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 168
Page 169
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 169
Page 170
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 170
Page 171
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 171
Page 172
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 172
Page 173
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 173
Page 174
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 174
Page 175
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 175
Page 176
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 176
Page 177
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 177
Page 178
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 178
Page 179
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 179
Page 180
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 180
Page 181
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 181
Page 182
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 182
Page 183
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 183
Page 184
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 184
Page 185
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 185
Page 186
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 186
Page 187
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 187
Page 188
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 188
Page 189
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 189
Page 190
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 190
Page 191
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 191
Page 192
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 192
Page 193
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 193
Page 194
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 194
Page 195
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 195
Page 196
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 196
Page 197
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 197
Page 198
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 198
Page 199
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 199
Page 200
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 200
Page 201
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 201
Page 202
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 202
Page 203
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 203
Page 204
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 204
Page 205
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 205
Page 206
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 206
Page 207
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 207
Page 208
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 208
Page 209
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 209
Page 210
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 210
Page 211
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 211
Page 212
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 212
Page 213
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 213
Page 214
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 214
Page 215
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 215
Page 216
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 216
Page 217
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 217
Page 218
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 218
Page 219
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 219
Page 220
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 220
Page 221
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 221
Page 222
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 222
Page 223
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 223
Page 224
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 224
Page 225
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 225
Page 226
Suggested Citation:"Response by Type of NMT Strategy." National Academies of Sciences, Engineering, and Medicine. 2012. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities. Washington, DC: The National Academies Press. doi: 10.17226/22791.
×
Page 226

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

also saw a 7 percent gain in target area walk counts. Counts obtained for Portland’s 2007 and 2008 programs showed target area bicycle volume net increases of 8 to 14 percent after adjustment for substantial citywide bicycling increases. Target area transit ridership data in Seattle, Washington, and Cambridge, Australia, show 11 and 16 percent 9-month and first-year increases, respectively, in parallel with individualized marketing. Seeking to increase physical activity is a major thrust of contemporary public health policy, and public-health-based interventions to promote active transportation have been tried. Although vaguely similar in concept to transportation-oriented individualized marketing, applications to date have been research-oriented and of much smaller scale. Self-reported walking increases of 1/2 to 1 hour per week were documented in a typical example, but long-term impacts beyond a few weeks or months are in doubt except for one intensive-support example. Additional summariza- tion is provided in the previously cross-referenced “Adult and Child Public Health Relationships Summary” that concludes the “Public Health Issues and Relationships” subsection within the “Related Information and Impacts” section. RESPONSE BY TYPE OF NMT STRATEGY This section focuses on the response of urban travelers making utilitarian trips, along with candi- dates for recreation and exercise, to a wide variety of pedestrian and bicycle improvements and strategies. The facilities and improvements addressed include sidewalks; at-grade and grade- separated crossings; pedestrian zones, malls, and skywalks; bicycle lanes and routes; shared use paths and trails; and system interconnections. Also covered are linkages to transit; point-of-destination provisions; pedestrian- and bicycle-friendly neighborhood design; policies and programs; and promotions and information. Some closely related user and usage characteristics data is provided. Most such information is located either within the “Underlying Traveler Response Factors” section, to the extent that it helps illuminate walking and cycling choice mechanisms, or within the “Related Information and Impacts” section. The latter section includes global data on Non-Motorized-Transportation (NMT) use overall and representative facility-specific pedestrian and bicycle volume information. Reference should also be made to the “Related Information and Impacts” section for additional information and interpretations from public health and other diverse perspectives. Sidewalks and Along-Street Walking Changes in volumes of walkers or walk activity levels in direct response to specific sidewalk improvements are documented here to the extent they are available. This information is bolstered with research on the effects, on overall walking levels, of sidewalk availability and street traffic intensity. Findings on prevalence of walking in pedestrian-friendly versus less attractive walking environments are examined both in this subsection—which focuses on sidewalk availability, traf- fic, and street characteristics issues—and later in the “Pedestrian/Bicycle Friendly Neighborhoods” subsection, where a broader view is taken. Pedestrian Volumes Overview It is useful, before examining traveler response data for sidewalks, to understand the nature and scale of NMT volumes encountered. Walking, the primary candidate for sidewalk use, can of 16-34

course occur with or without paved sidewalks or paths. People also walk on roads, shoulders, and unpaved areas. A majority of walkers do, however, use paved sidewalks or paths. A survey of Florida residents found that 67 percent of walking trips were on sidewalks or dedicated footpaths (NuStats International, 1998). A national survey found that 45 percent of respondents used mostly sidewalks while about 6 percent used mostly bicycle/walking paths or trails in their foot travels. Another 33 percent walked primarily on paved streets, roads, and shoulders (NHTSA and BTS, 2002). More background information on where people walk and in what numbers is provided in the “Related Information and Impacts” section under “Facility Usage and User Characteristics,” starting with the “Frequency of Facility Usage by Facility Type” discussion. One of the tabulations presented in the “Facility Usage and User Characteristics” subsection (see Table 16-98) presents volumes from illustrative intersection counts in San Francisco Bay Area counties. Total 2-hour AM plus 2-hour PM intersection pedestrian volumes covering all cross- walks, on both streets, range from 4,925 (roughly 12,000/day) at a “South o’Market” San Francisco central business district (CBD) intersection, to 900 (some 2,200/day) at a Santa Clara County (Silicon Valley) intersection with low-rise apartments near ethnic gathering spots, to 135 (300–350/day) at an exurban Napa County intersection with a town hall and dwellings, and on down to nine pedes- trians (20 to 25/day) at partially developed office/commercial intersections in suburban Santa Clara County and Napa County locations (see Table 16-98 for sources). A rough conversion of these representative intersection pedestrian volumes to average individual sidewalk volumes, for pur- poses of understanding typical magnitudes, may be accomplished—where sidewalks on both sides exist—by dividing by four. Observed pedestrian volumes run even higher than the one San Francisco CBD example, but volumes in the lower ranges are much more prevalent. Toward the fringes of any sidewalk sys- tem, pedestrian volumes are usually diminishingly small. Neighborhood sidewalks are in one sense like local roads, providing a land service function. Thus, most urban and a number of sub- urban jurisdictions require sidewalks as part of any new street construction. Portland, Oregon, requires sidewalks on any new street, excepting only cul-de-sacs with less than 5 dwellings and streets with severe natural constraints (Federal Highway Administration, 2004). Seattle require- ments call for sidewalks in connection with platting of any new street. In addition, the city requires that sidewalks be provided in connection with platting or developing six to 10 or more units, depending on zoning, or any units at all in the case of designated areas or streets (City of Seattle, 2008). Individual Sidewalk Provision Examples Availability of pedestrian counts before and after sidewalk improvements has improved, but remains limited. Virtually all examples are descriptive analyses with no statistical tests, no control- area counts, and too-frequent reliance on informal reporting. Table 16-1 summarizes examples encountered with apparently solid before and after observations, along with some less formal accountings. The final entry is from Safe Routes to School studies. 16-35

16-36 Study (Date) Process (Limitations) Key Findings 1. Aboelata et al. (2004) A badly degraded 1.5-mile sidewalk encircling Evergreen Cemetery in the Latino community of Boyle Heights in Los Angeles was conver- ted into a rubberized jogging path. (Analysis approach not reported.) Daily use rose from roughly 200 to over 1,000 people using the path for jogging, walking, and socializing. The increase probably includes some diversion from other facilities — doctors are said to advise use of the soft path surface. 2. Investigation by the Handbook Authors, 2002-06 (see Montgomery Co. case study — “Results- Sidewalk Improvements”) A rural-section 2-lane state highway in suburban Garrett Park, MD, with a badly degraded sidewalk on 1 side for 4 blocks and none for 1 block, was rebuilt with curbs, street trees, and sidewalks on both sides for 4 blocks and 1 side for the 5th block. (3-hour winter counts, 1 day each.) The AM peak period child pedestrian count 1 block from a school crossing decreased from 11 to 6, perhaps because of safety patrol termination. The adult and teenager count increased from 5 to 21 persons (up 320%). Thus the total 3-hour pedestrian count increased from 16 to 27 (up 69%). 3. Harkey and Zegeer (2004) One mile of partly commercial arterial in University Place, WA, was rebuilt from 5 lanes, gravel shoulders and no sidewalks to 4 traffic lanes with bike lanes, wide sidewalks, a median, and 2 mid- block crosswalks. (No sidewalk counts.) Few pedestrians walked or crossed the arterial without sidewalks or cross- walks. Usage after improvement is suggested by the 3,200 monthly pedes- trians on the midblock crosswalks. Crashes decreased 60% (no change in ped. crashes despite walking increase). 4. Painter – 1996 as summarized by Cao, Mokhtarian, and Handy (2007) and Heath et al. (2006) Before and 6-weeks-after study of pedestrian volume changes seen with street lighting improvements along three poorly lit streets and a footpath in London, with descriptive analysis. (“Fair execution.”) Volume increases (presumably after dark): Site 1 (footpath), males +50%, females +64%; Site 2, males +44%, females +45%; Site 3, males +34%, females +48%; Site 4, males +101%, females +71%; overall increase of 51%. 5. Boarnet et al. (2005a and b) (for more see “NMT Policies and Programs” — “Schoolchild-Fo- cused Programs” in this “Reponse by Type of NMT Strategy” section) Of 10 CA schools surveyed to ascer- tain 2002-03 SRTS impacts, 5 had received sidewalk improvements. Parents were asked retrospective questions about changes in walking and cycling to school. Counts were made 2 days running of child pedestrians at project sites, before and after improvement. (Survey obtained parent perceptions, not expressed in numbers. Count dates relative to school year not reported.) Walk/bike increases were more likely to be reported for children passing via sidewalk improvements (17%) than for study control subjects (3%). The increases were higher than for traffic controls (16% vs. 4%) or other crossing improvements (12% vs. 6%). Before- and-after-improvement counts showed a weighted average 5-site 46% increase in child pedestrians, with a ±82% reduction in the proportion walking in the roadway or on the shoulder. Note: Where substantial additional information on individual studies is provided in text and tables or figures, this is noted — and the location within the chapter is given — in the first column. Sources: As indicated in the first column. Table 16-1 Summary of Before and After Studies of Individual Sidewalk Provision or Improvement Examples in Various Locations

Four out of the five cases in Table 16-1 (all except the 3rd entry) offer quantified observations of pedestrian usage shifts. All cases show increased pedestrian activity accompanying the sidewalk improvements. Using overall averages (of all pedestrians studied, or all examples, within each individual study) pedestrian count increases ranged from 46 percent to 400 percent. The four quan- tified cases have a median pedestrian volume increase, among case study averages, of roughly 60 percent. The 400 percent increase in the 1st Table 16-1 entry cannot be disregarded as an anom- aly, given that the adult and teenager increase in the 2nd entry was 320 percent. The 5th and final entry in Table 16-1, one of the four cases with quantified observations, pertains to situations where there was no sidewalk at all in the “before” condition. The increase observed is slightly below the median sidewalk improvement increase, but probably not significantly so, especially considering that the counted population (children only) differed from other cases (see Table 16-1, column one, for citations and cross-referencing). Despite analysis limitations, it is fairly obvious that improved and new sidewalks do attract and serve more pedestrians. What is not known from the before-and-after counts alone is whether the added pedestrian volumes represent additional walking in the form of new walk trips, more fre- quent walk trips, or lengthened walk trips, or whether and to what extent the added volumes come from walk trips diverted from other routes or destinations. Diverted walk trips are not associated with either shifts in travel mode or additional walking. Nevertheless, even diverted trips repre- sent some benefit gained by the users, whether added safety, a more pleasant walk, or greater con- venience. Where there are increased pedestrian volumes it is likely that some degree of additional walking has been induced. The final entry of Table 16-1 includes research evidence that the 46 percent average increase in child pedestrians observed in response to sidewalk improvements at five California 2002–2003 Safe Routes to School program sites did indeed reflect additional walking (Boarnet et al., 2005a and b). Additional evidence of additional walking is provided by the cross-sectional and comparative analyses summa- rized in Table 16-2 below, as part of the “Sidewalk Coverage and Traffic Conditions” discussion. Sidewalk Indirectness Effects A discussion is provided in the “Underlying Traveler Response Factors” section, under “Trip Factors”—“Walk Trip Distance, Time, and Route Characteristics,” on the relative importance of travel time in pedestrian and bicycle utilitarian travel choices. The consensus is that time (or distance) is particularly important for the potential or actual pedestrian, with distance minimization as the dominant factor in route choice for utilitarian pedestrian trips (Weinstein et al., 2007). This sensitiv- ity to time/distance manifests itself both within segments of trips and with respect to the overall trip from origin to destination. Pedestrians notice sidewalk indirectness and seek to avoid it if they can. This phenomenon is addressed in design literature, but generally without quantitative support. Five quantified examples of pedestrian response to trip segment indirectness are presented in the “Special Mini-Studies in Montgomery County, Maryland” case study, under “More—Sidewalk Indirectness.” In the one example where count data were obtained, 80 percent of pedestrians were found to walk in the street behind parked cars in preference to incurring a 27 percent deviation involved in walking around via the sidewalk. Four supplementary examples look at what side- walk deviations are avoided (or were avoided before the pedestrian traces were paved) by cutting across grass or parking. The deviations, measured as percentage of existing (or original) sidewalk distance, range from 17 down to 12 percent for the segment involved in the deviation. The median deviation encountered, unacceptable to many or most pedestrians, was 15 percent. It is of interest to note that in the instance of the 17 percent deviation, caused by a zigzag sidewalk, a substantial 16-37

16-38 Study (Date) Process (Limitations) Key Findings 1. Cao, Handy, and Mokhtarian (2006) (see this section for more information) Models were utilized to reexamine a 6-neighborhood Austin, TX, data set to explore built environment and residential self-selection effects on walking for its own sake (strolling) and utilitarian walking to the store. The 2 neighborhoods with the worst commercial pedestrian access had the only 100% complete residential sidewalk systems. (Some evidence of survey bias; perceived measures of environment dominated analysis.) The proportion of strollers was modestly higher in older traditional neighborhoods, and no significant difference was found in mean strolling rates. Frequencies of walking to stores, even after accounting for self-selection, were positively related to pedestrian connections to stores, store quality, and store closeness, and negatively related to residential and retail area traffic. Self-selection was on the basis of walk access to stores but affected all walking. 2. Schneider (2011) (see “Underlying ... Factors” — “Envi- ronmental Factors” — “Ambiance” for more information) Interviewed San Francisco Bay Area pharmacy shoppers. Socioeconomic and trip data obtained for 959 tours were augmented/used in 3 mode choice models. (Sidewalk coverage of 91% may have given insufficient variability for variable calibration.) Sidewalk coverage not statistically significant in the one-shopping-district to/from mode choice model, full-tour model, or the within-shopping-districts model, but in the latter, number of driveway/alley crossings per mile was a significant negative for walking. 3. Moudon et al. (2007) (see “Ped…cycle Friendly Neigh- borhoods” for more information) Cross-sectional analysis of walking activity, socio-demographics, attitudes, and objectively measured environmental variables covering 608 adults in King County, WA. (Only major-road sidewalks were documented/taken into account.) Major road sidewalk length was found significantly related to walking in one of two modeling approaches (about 9% more walking per sidewalk-mile within 0.62 miles). Neighborhood measures such as store proximity were generally more important. 4. Lee and Moudon (2006a) (see “Ped…cycle Friendly Neigh- borhoods” for more information) Similar analysis to above study, also making use of public health and physical activity survey data and GIS-based physical features augmentation, but covering city of Seattle respondents only (438) and all city sidewalks. (Self-reported minutes and frequency of walking.) Sidewalk extent positively but not sig- nificantly related to minutes of walking (odds ratios from 1.05 to 1.12), with no consistent relationship to utilitarian walking frequency, but statistically sig- nificant relationship to frequent recrea- tional walking (odds ratio 1.12). Some proximity measures more important. a 5. Giles-Corti and Donovan – 2002 [Prev. Med.] as summarized per SR 282 Cross-sectional survey and analysis of Australian adults using measures of activity accessibility, neighbor- hood perceptions, and transporta- tion features. Examined utilitarian walking (UW), recreational walking (RW), and walking as recommended for health (WH). (Used self- reported walking and perceived sidewalk availability measures.) UW 65% higher, RW 41% higher, and WH 65% higher with perceived presence of sidewalks. UW 3 times more with perceived access to shops, less with beach access, and more in presence of lots of traffic. RW more with beach access and favorable perception of neighborhood. WH more with high access to public open space and favorably perceived neighborhood. 6. Reed et al. – 2006 as summarized in Saelens and Handy (2008) Analysis of Sumter County, SC, survey of amount walked per week and perceived presence of sidewalks in neighborhood. (Relationships not significant in race-stratified models.) Persons walking 1 to 149 minutes/week were found more likely to report presence of sidewalks than persons not walking at all. No relationship found for walking more than 150 min./week. Table 16-2 Summary of Research Findings on the Relationships of Sidewalk Prevalence and Street Traffic Characteristics with Walking Activity

16-39 Study (Date) Process (Limitations) Key Findings 8. Van Lenthe, Brug, and Mackenbush – 2005, as summarized in Saelens and Handy (2008) Related walking by adults in 78 Netherlands neighborhoods to various environmental conditions perceived by professional observers. Higher likelihood of walking was associated, for adults under 50 years of age, with less traffic noise, and for older adults, with greater proximity to food shops. 9. Krizek et al. (2007) (see “Related Info and Impacts” — “Public Health Issues…” for a study description) NMT Pilot Program Evaluation Study asked about neighborhood sidewalks and determined walk mode shares. (Any relationship between the two is circumstantial evidence: the study authors themselves did not infer causality.) Some 97% of Minneapolis respondents agreed there were sidewalks on most streets in their neighborhood, versus 59% to 63% in the 4 other communities surveyed. The 2006 Minneapolis walk mode share was 17.6% compared to 6.6%-11.8% for the other 4 areas. 10. U.S. EPA – 2003 as summarized per SR 282 Utilized 2 surveys of the Gainesville area of Florida to examine various density and pedestrian environment variables along with NMT travel times to school. Choice of walking to school positively influenced by sidewalk availability and shortness of time to school from home. Choice of cycling significantly influ- enced only by bicycle travel time. 11. Ewing et al. – 2004 as summarized by Davison and Lawson (2006) Utilized objectively measured cross- sectional data to model the effect on walk/bike school access of sidewalk characteristics, bike lanes or paved shoulders, accessibility, and density. Student walk-to-school shares showed a significant positive relationship with main road sidewalk availability and a negative relationship with estimated walk/bike travel time to school. 12. Fulton et al. – 2005 as summarized in Saelens and Handy (2008) Utilized survey of U.S. parents to relate area type, perceived sidewalk availability, and perceived play safety to usual mode of travel to school. (Used perceived and self- reported measures.) Active transportation to school was more likely in non-rural areas, in areas perceived to have sidewalks, and when the child felt safe playing in the neighborhood. The safety variable was not significant in the full model. 13. Timperio et al. – 2004 as summarized by Davison and Lawson (2006) Conducted cross-sectional analysis of various area conditions. (Used parental perceptions of conditions and parental reporting of walking and cycling.) Lesser walking/cycling was, among Australian 5-6 year olds, associated with parental perceptions of heavy traffic and poor public transportation. 14. Carver et al. – 2005 as summarized by Davison and Lawson (2006) Cross-sectional analysis of parent and child perceptions of various facilities and environmental conditions. (Used self-reported physical activity measures as well as perceived environment measures) Australian adolescents (male, female, or both) were found to walk/bike more where traffic was less problematic, roads were perceived to be safe, and there were fewer unattended dogs and more good places to be active. Note: Where substantial additional information on individual studies is provided in text and tables or figures, this is noted — and the location within the chapter is given — in the first column. a See text Footnote 21 in the “Shared Use, Off-Road Paths and Trails” subsection for a brief explanation of odds ratios. Sources: As indicated in the first column. The notation “SR 282” is shorthand for Committee on Physical Activity, Health, Transportation, and Land Use (2005) together with Handy (2004). 7. Kitamura, Mokhtarian, and Laidet (1994) and SR 282 Cross-sectional analysis of travel behavior, built environment char- acteristics, and attitudes surveyed in 5 diverse San Francisco Bay Area neighborhoods using incrementally expanded regression models. (Aggregate facility measures.) Number or share of NMT trips positively related to North S.F. location, rapid transit and bus access, sidewalks in neighborhood, high density, and closeness of nearest park (thought to have actually served as a disaggregate land use mix indicator). Table 16-2 (Continued)

16-40 9 An attempt at city-wide before-and-after analysis of pedestrian system improvements has been underway as part of the five-city Nonmotorized Transportation Pilot Program Evaluation Study, with findings publi- cation scheduled for after a 2010 follow-up survey (Krizek et al., 2007). proportion of short-cutting pedestrians and cyclists appear to be satisfied with a 6 percent devia- tion relative to a possible but grassier straight-line routing. Trip origin to destination indirectness was examined as one of many explanatory variables in a set of walking activity research models developed for the Seattle area (King County). Route directness from home to the nearest grocery store and from home to the closest school both proved to be sig- nificant variables in two out of three final models, with directness being associated with more walk- ing overall per week (Moudon et al., 2007). More background on this analysis is given in Table 16-2 and in the “Pedestrian/Bicycle Friendly Neighborhoods” subsection under “Diversity.” Sidewalk Coverage and Traffic Conditions Assessing importance of broad-area sidewalk system coverage and improvements is generally done by means of empirical investigation of static situations, including paired-community descriptive comparisons of walking conditions and activity, and through survey-based cross-sectional model- ing.9 Such research often covers other environmental factors that potentially affect walking, thus sometimes addressing effects of traffic conditions. Traffic conditions are typically expressed in a gen- eral way, but some of the studies support inferences about traffic effects on streets where there are no sidewalks and pedestrians must share the street space with motorized vehicles. Table 16-2 sum- marizes a wide selection of broad-area studies, starting with eight cross-sectional analyses relating adult walking activity to built environment factors, followed by one descriptive analysis, and clos- ing with five studies relating child and adolescent walking to the physical environment. The 1st Table 16-2 entry pertains to Austin, Texas. The research done on walking in Austin neigh- borhoods stands out for its examination of “strolling” (walking for its own sake) separate from util- itarian walking (to meet a travel need). It is also notable for analytical refinements accomplished in waves of research extending over more than a decade. Most of the Austin analyses draw from a 4-page 1995 mail-out/mail-back survey that achieved a 23 percent response rate. This provided 1,368 completed questionnaires covering walking behavior, neighborhood perceptions, and atti- tudes. The built environment was also quantified through site visits, geographic information sys- tems (GIS), and network analysis. Six neighborhoods were studied. Two are traditional, laid out on more-or-less of a grid just beyond the downtown, with stores focused on sidewalks. Another two are early modern, immediately post- World-War-II, located somewhat farther out and more reliant on auto-oriented strip commercial for shopping. The last two are late modern 20th Century, 10 to 15 miles from downtown, with retail layouts requiring walking though parking lots for access. The late modern neighborhoods, how- ever, have the only complete residential sidewalk systems. Residential sidewalk systems are only partial in the four older neighborhoods. One each of the traditional and late modern neighborhoods have a large park with extensive walking trails. Residential street widths are 26 feet in the tradi- tional neighborhoods, 26 to 30 feet in the early modern developments, and 36 to 40 feet in the late modern locales. Despite efforts to match populations in the research, some differences were found. Residents of the early modern neighborhoods reported, for example, being somewhat older (Cao, Handy, and Mokhtarian, 2006, Handy, Clifton, and Fisher, 1998). The 1st entry in Table 16-2 per- tains to a reexamination of the Austin data. Table 16-3 summarizes the measured walking activity.

Almost every walk activity parameter presented in Table 16-3 progresses steadily downward from traditional to late modern neighborhood types. However, statistical significance (taking into account all six neighborhoods individually) is exhibited only by the walk trips to store statistics, which decline dramatically, and by the percentages who strolled. As can be seen, walking for its own sake varied least among neighborhood types. Reasons for engaging in strolling were varied, but exercise/health, pleasure, and dog-walking predominated (Cao, Handy, and Mokhtarian, 2006, Handy, Clifton, and Fisher, 1998). Findings of the six-neighborhood Austin studies indicated that persons highly rating “stores within walking distance” as important in their choice of residence location (so-called “self-selection”) were walking for its own sake more frequently. Ambient environment impacts on walking were all statistically modeled taking into account both self-selection and demographics. In decreasing order of importance, perceptions of traffic and personal safety combined with light traffic volumes, shade from trees, and opportunity to see people were all significantly and positively related to strolling. Utilitarian walking to stores was significantly related to residence closeness to the nearest store, perceived quality of commercial area pedestrian facility connections, perceived advantage of walking (including parking hassle avoidance), usefulness and quality of stores, commercial area walking comfort, and amenable residential area traffic conditions, with a negative relationship to measured commercial street traffic volumes (Cao, Handy, and Mokhtarian, 2006). Although the available quantitative data did not support separate examination of traffic and safety effects for streets with versus without sidewalks, focus group results from 1995 suggest that low traffic volumes and speeds are required for lack of sidewalks to be not perceived as a hindrance to walking. Conversely, heavy traffic was viewed as being detrimental even with sidewalks available (Handy, Clifton, and Fisher, 1998). Attitudinal intercept surveys in four neighborhoods found con- tinuous sidewalks or trails and tree shade to be important to persons exercising or strolling but not to persons walking for utilitarian purposes (Shriver, 1997). The conflict between this finding and the frequency data for strollers in Table 16-3, which shows the pairs of neighborhoods with incom- plete residential sidewalk systems (traditional and early modern) to have somewhat higher strolling frequencies (Cao, Handy, and Mokhtarian, 2006) suggests that other factors are at work as well. Though not a stated conclusion of the various Austin researchers, one possible interpreta- tion of the results is that narrow and pleasant low-volume streets, when associated with lower traf- fic speeds, can tend to compensate—at least for able-bodied adults—for partial lack of residential area sidewalks. Behavior of child pedestrians and perceptions of their guardians were not studied. 16-41 Table 16-3 Walking Trips for Strolling and Shopping in Six Austin Neighborhoods Neighborhood Type (One Pair Each): Traditional Early Modern Late Modern Strolling Trips Percent strolling at least once in 30 days 83% 77% 78% Average trips/30 days for those who strolled 12.7 12.1 11.2 Average trips/30 days for all respondents 10.5 9.2 8.4 Walk Trips to Store Percent walking at least once in 30 days 62% 43.5% 21.5% Average trips/30 days for those who walked 6.3 4.3 3.9 Average trips/30 days for all respondents 4.2 1.9 0.8 Source: Cao, Handy, and Mokhtarian (2006), with elaboration by the Handbook authors.

A San Francisco Bay Area study of trip tours involving a pharmacy shopping stop among the tour’s activity stops (2nd entry in Table 16-2) provides a finding similar to that of the Austin studies in that sidewalk coverage did not achieve statistical significance or contribute to any of the three mixed-logit mode choice models developed. Unlike the Austin research, however, tree canopy cov- erage was found significant and positive for shopping trip walk mode choice (Schneider, 2011). What is not clear, however, is whether the sidewalk insignificance outcome reflects nothing more than too little variability in sidewalk coverage (a very high 91.3 percent mean coverage within the sample) for sidewalk variable calibration or whether sidewalk presence truly was unimportant in the choice to walk or not walk when shopping. A highly significant variable in the model for trips made internal to shopping districts was, how- ever, the number of major driveways and alleys per mile that had to be crossed to walk along the main commercial roadway. Major driveways were defined as active commercial driveways or res- idential driveways serving more than 10 dwelling units. Survey respondents who would encounter more such crossings to walk between stores were less likely to do so and more likely to drive. The calibrated model parameters suggested that having 10 fewer major driveway and alley crossings per mile was worth walking an extra minute, and it also appeared that beyond 30 cross- ings per mile walk mode share dropped sharply (Schneider, 2011). These results are consonant with the Austin determination that perceived quality of commercial area pedestrian facility con- nections is important in the choice of walking for shopping trips. Driveway/alley crossings per mile may be in part a surrogate for auto orientation including presence of front-of-store parking facilities, but the policy implications are basically the same in any case. The other six adult-focused cross-sectional walking studies in Table 16-2, the 3rd through 8th table entries, generally found presence of neighborhood sidewalks or major road sidewalks to be posi- tively related to walking activity, although typically not the strongest indicator. Three of the study summaries identified some form of store proximity to be important, as in Austin. Two found pos- itive significance in some friendly neighborhood measure and two found presence of open space to be a positive. The Netherlands study joined Austin in finding traffic impacts to be a negative, while the Australian study was unique in finding a positive relationship between walking and traf- fic, perhaps as a reflection of density or high activity. The previously introduced King County research (the 3rd table entry) tested several objective traf- fic and road-size measures without finding any significant relationships, although attitudes and perceptions did contribute to explanation of walking activity. Major road sidewalk length was sig- nificant in one of two modeling approaches (Moudon et al., 2007). The closely related Seattle-only research (the 4th table entry) was, unlike the county-level analysis, able to use as a model variable total length of all sidewalks within a 1 km. buffer. It found that sidewalk extent was neither a con- sistent nor a significant variable for explaining choice to walk for transportation (utilitarian trips), but was significant as an explanatory variable for frequent recreational walking. Similarly, per- ceived architectural variety was associated with frequent recreational walking but not utilitarian walking. Traffic volume was not significant for either type of walking trip, and neither was pres- ence of parks or trails (Lee and Moudon, 2006a). The county-level research found that trails attracted walk trips but did not appear to induce more (Moudon et al., 2007). The one descriptive analysis included in Table 16-2, the 9th entry, supports the importance of side- walk system completeness. The first-phase NMT Pilot Program statistics offer only circumstantial evidence, but the five-city comparison is striking, with double the walk mode share in the one city (Minneapolis) where all but a few residents agree that most streets have sidewalks. Also parti- cularly telling are the Seattle neighborhood comparisons, covered below in the business districts discussion and tabulated in Table 16-4, which find three times the walking to and from the neigh- 16-42

borhood commercial district in those cases where blocks are small and the sidewalk system is largely complete. The last five studies summarized in Table 16-2, the 10th through 14th entries, are child-oriented active transportation studies. All three of the U.S. studies found walking to be positively related to sidewalk availability, while the two Australian studies found it to be negatively related to heavy or problematic traffic. Perceptions of safety were mentioned as a positive in one U.S. (“personal safety”) and one Australian (“traffic safety”) study summary. Good transit service was positively related to walking in one of the child-focused studies, as it was in one adult-focused study (see Table 16-2 for citations). Sidewalk coverage is an NMT feature that has probably been studied as much for its health impacts as for its effects on travel demand. Active living research relating to sidewalk availability is covered within the “Public Health Issues and Relationships” subsection of the “Related Information and Impacts” section (see both “Health Benefits for Adults of Enhanced NMT Systems and Policies” and “Health Benefits for Children of Enhanced NMT Systems and Policies”). 16-43 Table 16-4 Summary of Descriptive Studies of Sidewalk Extent/ Enhancements and Traffic Calming Affecting Business Districts Study (Date) Process (Limitations) Key Findings 1. Hess et al. (1998), Moudon et al. (1997), and SR 282 (see case study “Pedestrian Activity... Seattle”) Descriptive, comparative analysis of 12 Seattle area shopping districts and their surrounding neighborhoods, 6 urban and 6 suburban, controlled for density and mix, using 16-hour shopping area pedestrian cordon counts. (No statistical testing.) The urban examples, with small blocks and averaging 38 miles of sidewalks, averaged 38 pedestrians/hour cross- cordon flows per 1,000 residents. Suburban examples, with large blocks and averaging 8 miles of discontinuous, incomplete sidewalks, had 12 pedestrians/hour/1,000 residents. 2. Harkey and Zegeer (2004) Main Street in the mountain town of Hendersonville, NC, was 4 lanes plus parking on a 100-foot right-of- way. In late 1970s a 2-lane traffic- calmed design was installed with mid-block lateral shifts defined by bulb outs with crosswalks, framing 1/2-block angle plus parallel park- ing sections. (No before counts.) Recalled as being virtually lifeless in the mid-1970s, with 17 closed stores, the pedestrian volume 25 years later on Main Street averaged 1,750/day. Designated a National Trust “Main Street City,” 100 retail businesses were in place downtown with a waiting list for occupancy, despite after-condition regional shopping mall competition. 3. PBIC and APBP (2009) As part of a post-1999 revitalization, East Main Street in downtown El Cajon, CA, was converted from 4 to 2 lanes, with angle parking and sidewalks widened for shared use activities. Pedestrian connections were bettered. (NMT impacts must be inferred from economic impacts.) Starting with a downtown that was partially vacant in the 1980s, and aided by a denser mixed-use land use plan, circa 2008 property values have risen by 181% relative to 1996 (versus 75% citywide) and leasing rates have increased 56%. Shopping and dining customers are up 91% relative to 2002. Note: Where substantial additional information on individual studies is provided in text and tables or figures, this is noted — and the location within the chapter is given — in the first column. Sources: As indicated in the first column.

Residential and Mixed-Use Traffic Calming Traffic calming in the United States has tended to focus narrowly on crash prevention, whereas in Europe this objective has been joined for some time by other objectives including enhancement of walking and bicycling (Ewing, 2008). Traffic safety benefits in residential neighborhoods are well established for vehicle crashes overall, although not specifically for pedestrians and bicyclists. A comprehensive study of Vancouver, BC, Canada, and international experience found 85 traffic calming projects throughout the developed world to have reduced crashes by 8 to 100 percent (8 to 95 percent where at least five crashes were recorded in the before period). In four Vancouver case study neighborhoods, the crash reduction was 18 to 60 percent, the average was 40 percent, and the annual claims cost reduction was 38 percent. Effects on pedestrian and bicycle crashes were reported only for the Vancouver neighborhoods, and it appears that the projects where such crashes decreased were counterbalanced by projects where they increased (Zein et. al., 1997). Compilations of effectiveness for various physical traffic calming measures, such as street narrow- ing and small traffic circles, indicate average speed reductions ranging from none (diagonal divert- ers) to 23 percent (speed humps), and volume reductions from 20 percent (choker) to 44 percent (full street closure) (Traffic Calming.org, 2011). Among studies covered in the previous “Sidewalk Coverage and Traffic Conditions” discussion, roughly 1/4 identified some favorable effect on degree of walking of lesser, slower, and/or less problematic street traffic. This is suggestive that traffic calming is likely to have some positive impact on extent of walking. Also suggestive is the experience with bicycle boulevards (essentially traffic-calmed streets with bicycle preference) that shows these streets to be attractive for bicyclists (see both “Popularity, Preferences, and Route Choice” and “Bicycle Lane Variations, Bicycle Boulevards, and Other Signed Bicycle Routes” in the “Bicycle Lanes and Routes” subsection). In addition, one would expect the public perception that traffic calmed streets are safer would sup- port walking and bicycling activity. Two studies of pedestrian and bicycle street traffic volumes, before and after traffic calming, both support a supposition of favorable impact on walking and bicycling.10 Analysis of street use before and after the 1990 traffic calming of Milvia Street in Berkeley, California, found afternoon peak hour vehicular traffic decreases from about 520 to 420 autos, approximately a 20 percent decrease. Pedestrian traffic increased from about 55 to 95 and bicycle traffic increased from 65 to 110, increases of roughly 70 percent. In Vinderup, Denmark, 7-1/2 hour daylight counts showed pedes- trians increasing from about 850 to 1,150 and bicyclists increasing from 1,050 to 1,950 in response to a circa 1984 traffic calming project. The combined increase was thus over 60 percent, with a greater effect on bicycling (Ewing, 2008). Sidewalks and Traffic Calming for Business Districts The Austin neighborhoods research described above and entered in Table 16-2 highlights the importance of commercial area sidewalks, and good sidewalk connections to stores, in attracting more persons to the walk mode for shopping trips (Cao, Handy, and Mokhtarian, 2006). The San Francisco shopping-tour research, also entered in Table 16-2 and described above, found minimum driveway/alley crossings of commercial district sidewalks to be strongly associated with more 16-44 10 These were the only pedestrian and bicycle impact studies encountered for traffic calming projects. The numerical values reported are approximate, having been scaled by the Handbook authors from charts in the source.

walking between stores (Schneider, 2011). This importance of customer-friendly commercial-area sidewalks is underscored by the already-mentioned comparative analysis in Seattle, which focused on volumes and characteristics of pedestrians walking between residences and the local commer- cial centers of 12 neighborhoods (Moudon et al., 1997). The 1st Table 16-4 entry pertains to this analysis, with further detailing in the case study “Pedestrian Activity Effects of Neighborhood Site Design—Seattle.” The Seattle area study sites were described in terms of a 1/2-mile pedestrian travel catchment area around each of the 12 neighborhoods’ commercial centers. The six sites classified as “urban” had in their catchment areas almost 5 times as many miles of sidewalks as the six sites classified as “suburban.” Their commercial parking was on-street or in small lots as contrasted to large expanses of parking. At these “urban” sites about 3 times as many pedestrians per 1,000 residents—38 per hour over a 16-hour period—were found to be walking between residences and the commercial centers. Evaluation of pedestrian makeup relative to neighborhood resident characteristics suggested that those who did walk in the sites classified as “suburban” were—including persons under 18 years of age—probably disproportionately among the transportation disadvantaged. It was thus more often those persons with limited mobility options who were left to navigate inadequate pedestrian infrastructure (Hess et al., 1998, Moudon et al., 1997). The two other studies entered in Table 16-4 offer no or limited pedestrian flow information, but present an anecdotal picture bolstered by economic resurgence information of substantial sidewalk- oriented business activity increases. Both of these smaller cities, one in rural North Carolina and one in the San Diego metropolitan area, put their Main Street on a “road diet” while at the same time engaging in economic redevelopment activities. Traffic calming and sidewalk connectivity enhance- ments were part of overall programs that successfully engendered increased overall activity and business retail viability in old downtowns (Harkey and Zegeer, 2004, PBIC and APBP, 2009). In the central business districts (CBDs) of large metropolitan regions, sidewalk systems are usually almost complete. Sidewalk improvements are typically along the line of “tweaking” the system, such as through selective sidewalk widenings, removal of sidewalk obstructions, introduction of ADA provisions, or enhancement of street furniture amenities. There are also “beyond-sidewalk” improvements for CBDs, such as pedestrian malls and skywalks. These types of actions and their effects on walking are covered in the “Pedestrian Zones, Malls, and Skywalks” subsection to follow, and a major example is provided by the case study, “50 Years of Downtown NMT Facility Provisions—Minneapolis.” Sidewalk Use by Bicyclists Sidewalks are built primarily for pedestrian use, although some sidewalk facilities have been designed specifically to accommodate bikeways. There are issues of safety with sidewalk use for cycling, more for the cyclists themselves than for the pedestrians. Cyclist use of sidewalks is not insignificant. One U.S. national survey reported that about 14 percent of respondents who had cycled in the previous 30 days used mostly sidewalks for their trip (NHTSA and BTS, 2002). A sep- arate U.S. national survey, taking all trip purposes into account, arrived at a figure of about 11 per- cent for bicyclists traveling mostly on sidewalks (Bureau of Transportation Statistics, 2002). The significant usage level of sidewalks for cycling could be attributable to lack of acceptable alter- natives or misperceptions of risk. Indeed, many non-cyclists and beginning cyclists think riding on sidewalks is safer than riding on the street (Zehnpfenning et al., 1993). This perception may actually be “close enough” for children: Sidewalk safety problems appear to apply primarily to 16-45

adult cyclists, likely because they bicycle faster and thus surprise motorists at points of conflict (Wachtel and Lewiston, 1994, Turner et al., 2006). Even for adult cyclists, agreement on sidewalk- riding safety is not universal (Lusk et al., 2011). A brief examination of these safety issues is provided in the “Related Information and Impacts” section (see “Safety Information and Comparisons”—“Facility Type Safety Comparisons”—“Cycling Crashes on Sidewalks versus Other Facilities”). Obviously, sidewalk vehicle-conflict safety concerns do not apply to long stretches of sidewalks or side paths free of driveways, alley crossings, and intersections. Reductions in bicycle use of sidewalks have been achieved by offering parallel on-street bicycle provisions. The only relevant study data encountered apply to instances where the parallel provi- sions have been bicycle lanes. The findings presented here are extracted from studies more fully covered in the “Bicycle Lanes and Routes” subsection under “Bicycle Lane Implementation” (see Table 16-11) and in the case study, “Anderson Road Bicycle Lanes—Davis, California.” Fell Street, in San Francisco, must have been a challenging environment for bicyclists before imple- mentation of bicycle lanes. PM peak period 2-hour counts found 37 out of 71 cyclists (52 percent) to be using the sidewalks in the “before” condition. After bike lane installation, sidewalk use by cyclists dropped to 7 out of 94 cyclists (7 percent). Results were less striking in Fort Lauderdale, Florida, where narrow 3-foot bicycle lanes were installed along a beachfront state highway. Saturday afternoon counts totaling 1 hour found 29 out of 68 cyclists (43 percent) to be using the sidewalks in the “before” condition, along with 344 pedestrians. With bike lanes, off-season side- walk use by cyclists was still 23 out of 51 cyclists (45 percent), along with 206 pedestrians (Chaney, 2005). As tentatively hypothesized in the “Bicycle Lanes and Routes” subsection, bicycle lanes may be less effective where there are large proportions of less skilled cyclists. This may be particularly so when the bike lanes are narrow. Bicycle lane provision along Anderson Road in Davis, California, was accompanied by before and after counts and surveys that identified age and sex. This process involved observer estimation in the case of the counts. Bicyclists on sidewalks were not quantified directly, either before or after, and the absolute numbers of children counted bicycling on Anderson Road actually declined slightly. In the “after” bike-lanes-implementation count, out of 1,577 cyclists on Anderson Road during 3 peak period hours, seven were estimated to be of age 11 and under, and 41 were judged to be between 12 and 17 years old. Anderson Road cyclists picked up in the “after” survey included five in the youngest age category and six in the age 12-through-17 category. Among these, two in the 11-and-under group (both female) and three in the 12-through-17 group (all male) were children and adolescents who had switched from sidewalk bicycling to on-road cycling. No cyclists of age 18 and up were identified in the survey as having previously used the sidewalks (Lott, Tardiff, and Lott, 1979). None of these three case studies involving bicycle lane provision suggests any significant adverse effect relative to the objective of having fewer bicyclists using sidewalks. The Fell Street example achieved a major reduction in on-sidewalk bicycling that probably involved predominantly adult activity, likely high-risk, although age group identification was not provided in the source and is only a guess. The Anderson Road example appears to have had a positive effect on child bicyclist behavior, vis-à-vis sidewalk use, but incomplete information prevents a firmer conclusion. Street Crossings Pedestrian crossing improvements are intended to make the crossing of roadways easier and safer for pedestrians, and bicyclists as well. Street crossings figure prominently in most pedestrian trips. A survey of Florida residents found, based on 175 “most recent” pedestrian trips reported, that 16-46

76 percent of trips required crossing streets and 53 percent involved crossing at intersections (NuStats International, 1998). Unfortunately, about 30 percent of all pedestrian fatalities are related to improper crossing of a roadway or intersection (Institute of Transportation Studies, 2003). Traffic control devices such as pavement markings, signs, and signals may be used to facilitate and chan- nel pedestrian crossings. Alternatively, normally at high fixed cost, a pedestrian and/or bicycle underpass or overpass may be constructed to provide absolute separation from vehicular traffic. Both at-grade and grade-separated crossings are covered here. The bulk of the studies encountered on crossing improvements have focused on safety and design issues rather than on travel demand response, the core subject of this “Traveler Response” Handbook. From a travel behavior standpoint, the primary underlying traveler response factors addressed by these improvements are travel time and perceived safety. Crossing improvements may also help maintain the continuity of the pedestrian network by mitigating barriers to pedes- trian movement. Long crossing delays, indirect pedestrian routings, high vehicle speeds, or fre- quent vehicle-pedestrian conflicts can all contribute to a barrier effect. High-quality crossings can contribute to a sense of connectedness and enhance the overall value of pedestrian facilities in an area. Measures of street crossing ease have been found to be related to transportation mode choice (Replogle and Parcells, 1992). An on-street survey covering seven U.S. marked-crosswalk sites in three southern-tier states found “that as the control at a pedestrian crossing increases through the addition of signs, flashing lights, and/or signals the pedestrians’ perception of safety also increases.” On a scale of 5 (unsafe) to 1 (very safe), perceptions shifted from an average score of greater than 4 in cases of simple marked crosswalks to better scores in the range of 3 to 2 or less for cases of signalized crosswalks (Fitzpatrick, Ullman, and Trout, 2004). Table 16-5 provides a summary compilation of usable pedestrian and bicyclist travel behavior impact studies. It includes both quantitative research and less formal reporting, but the findings are consistent to the extent that—in their totality—they largely demonstrate provision of safe and attractive crossings is an essential and full-partner element of providing an overall NMT system that will attract and induce additional walking and bicycling. The table starts with studies involv- ing crosswalks, associated traffic controls, and major street crossings in general, that address gen- eral-purpose (mostly adult) pedestrian and cyclist usage. These are followed by similar at-grade crossings studies focused on the school commute of children and adolescents. The last three entries of Table 16-5 involve grade-separated crossings. While traffic calming may properly be considered a tool for making street crossings less of a barrier to pedestrians and cyclists, that strategy is cov- ered in the preceding “Sidewalks and Along-Street Walking” subsection under “Residential and Mixed-Use Traffic Calming” and also “Sidewalks and Traffic Calming for Business Districts.” 16-47

16-48 Table 16-5 Summary of Before and After Studies and Research Findings on Relationships between Street Crossing Provisions and Walking/Biking Activity Study (Date) Process (Limitations) Key Findings 1. Knoblauch, Nitzburg, and Seifert (2001) (see this section for more information) A study was conducted at 11 inter- sections in 4 U.S. cities where paint- ed crosswalks were installed or, in 1 case, upgraded. Before-and-after 8 AM - 7 PM observations covered the crossings and vicinity (All streets had the same 25 mph posted speed). Percentage use of crosswalks increased <1% in Stillwater, MN (2 sites), 4% in Sacramento, CA (3 sites), 11% (signifi- cant) in Richmond, VA (3 sites), and 12% (significant) in Buffalo, NY (3 sites). No statistically significant change was observed in total observa- tion-area crossings (up <1% overall). 2. Zegeer et al., (2005) (see this section for more, including safety information cross- reference) Study of crashes, with exposure rate assessment, at 1,000 marked cross- walks and 1,000 matched unmarked mostly nearby crosswalks in 30 U.S. cities. (Cross-sectional analysis.) Study could not examine changes in volumes, but compared to 66% total choosing marked crosswalks, 73% of children 12 and 73% of seniors age 65 chose marked crosswalks. (Vol- umes may exhibit legacy characteristics from before marking: See Footnote 11.) 3. PBIC and APBP (2009) An intersection of 2 state roads in Edwards, CO, had a typical “rural” layout with free right turns separa- ted from through traffic by “pork- chop” traffic islands containing the signal poles. It was rebuilt and upgraded in an “urban” configura- tion. (Limited NMT analysis.) Removal of the right turn islands and substitution of sharper corner radii provided shorter total walking distance in the intersection, signal control of all movements, and ADA compliance. Pedestrian use of the intersection more than doubled and other traffic functions were improved as well. 4. Harkey and Zegeer (2004) (see “Pedestrian/ Bicycle Systems and Interconnec- tions” for more) As part of downtown Ft. Pierce, FL, revitalization, the pedestrian- unfriendly intersection at the gate- way to the waterfront was rebuilt with a traffic circle, sidewalk exten- sions, and median refuge islands. (Analysis approach not reported.) The intersection improvement was part of an overall program to slow traffic, widen sidewalks, and improve beach access. Intersection traffic remained at about 14,000 vehicles/day, but for pedestrians, increased from about 50 to approximately 1,000 pedestrians/day. 5. UK Department for Transport – 2004 as summarized in Booz Allen Hamilton (2006) (see this section for more information) When one-way traffic flow was reversed in London’s Shoreditch Triangle, the number of traffic- signal-controlled intersections was increased, with placement in accord with pedestrian desire lines, and sidewalks were selectively widened. (Evaluation approach, likely “before and after,” not summarized.) The evaluation consultant, Intelligent Space, found a 56% increase in pedestrian use of assigned crossing areas, a 61% decrease in jaywalking, and a 9% increase in overall pedestrian crossings. Parties to the scheme believe crash risk has been reduced and that “with the roads easier to cross, their severance impact has been reduced.” 6. Troped et al. (2001) (see also “Shared Use, Off-Road Paths and Trails” — “Preferences... Walk/Bikesheds”) Conducted cross-sectional mail sur- vey in Arlington, MA, with multi- variate analysis including various neighborhood feature and rail-trail access variables. (GIS-identified busy crossings and perceived steep grades between home and trail not statistically significant for trail use.) Minuteman Trail use was twice as likely, taking other factors into account, if survey respondent perceived they did not have to cross a busy street for trail access. Other access factors signifi- cantly deterring use were distance from trail entry point and GIS-measured presence of a steep grade.

16-49 Study (Date) Process (Limitations) Key Findings 8. Boarnet et al. (2005a and b) (for more see “NMT Policies and Programs” — “Schoolchild- Focused Pro- grams” within this “Response by Type of NMT Strategy” section) Of 10 CA schools surveyed to ascer- tain 2002-03 SRTS impacts, 2 had received full traffic signals and 3 had received crossing improvements. Parents were asked retrospective questions about changes in walking and cycling to school. Counts were made 2 days running of child pedestrians at project sites, before and after improvement. (Survey obtained parent perceptions, not expressed in numbers. Seasonality of counts not reported.) Walk/bike increases were more likely to be reported for children passing via new traffic signals (16%) than for study control subjects (4%). Similarly, increases for other crossing improvements were 12% (vs. 6% for controls). These crossing improvement results compare to 17% (vs. 2% for controls) for sidewalk projects. After-signalization counts showed a weighted average 2-site 24% increase. Overall, crossing improvement counts were inconclusive. Traffic yielding improved, significantly so, at 3 of 5 sites. 9. Timperio et al. – 2004, Timperio et al. – 2006, both as summarized by Davison and Lawson (2006) Conducted cross-sectional analysis of parental or adolescent perceptions (2004) and objective measures (2006) of various area or school access conditions in Australia. (Parental reporting of walking and cycling in both studies.) Lesser walking/cycling among 10 to 12 year olds was associated with multiple roads to cross, lack of signals and crossings, and other factors. Lesser walking/cycling to school by both 5 to 6 and 10 to 12 year olds was associated with busy road barriers and a commute over 800 m. (1/2 mile). 10. Two additional studies of school access examined by Moudon, Stew- art, and Lin (2010) Cross-sectional evaluations of the effect on active commuting to school of need to cross a major street en route. (No methodological or background details reported.) A model variable representing a major street crossing was found, in Switzer- land, to be associated with less NMT school commuting, but no association was identified in an Oregon study. 11. Harkey and Zegeer (2004) In Phoenix a 7-lane arterial was built across a field previously crossed by elementary students. Later a bridge was installed with ramp and spiral staircase access. (Limited NMT impact information.) Before pedestrian bridge installation, 2 crossing guards proved no match for the 50 mph Greenway Parkway traffic. Over 60 students now use the bridge. One school crossing guard enforces bridge use. 12. Moore and Older (1965) (see this section for more information) Investigated some 30 or so origin- destination pairs with at-grade versus grade-separated pedestrian route options, counting use and timing trips. (Hand-fitted curve.) Pedestrians showed a very small route choice tolerance for added travel time (Fig. 16-1) in U.K. context examined, with some tolerance for undercrossing use and none for overcrossings. 13. Pedestrian and Bicycle Information Center (2010) A pedestrian and bicycle overpass was built in Clark County, WA, to connect growing residential/ commercial communities on opposite sides of a 4-lane parkway with limited crossing opportunities. (No analysis beyond counting.) February/March 2004 2-hour counts (no rain) were 8 pedestrians and 5 bikes in a Wednesday AM peak period, 29 pedestrians and 7 bikes in a Friday PM peak period, and 9 pedestrians and 10 bikes on a Saturday midday. NMT travel distance saved not reported. Note: Where substantial additional information on individual studies is provided in text and tables or figures, this is noted — and the location within the chapter is given — in the first column. Sources: As indicated in the first column. 7. Gårder, Leden, and Pulkkinen (1998) Cycle tracks on 5 streets involving over 30 intersections in Gothenburg, Sweden, were improved: primarily by replacing painted bike cross- walks with raised/red-colored speed tables. Before/after volumes obtained on 2 streets, with controls. (No details on volume analysis.) Bicycle flows on the 2 streets with volume investigations increased 75% (one side) to 79% (other side) on one street and 100% on the other, compared to 20% at control intersections. (No information on diversion.) The rais- ed/colored crossings were judged to have led to at least 30% greater safety. Table 16-5 (Continued)

Crosswalks and Traffic Controls Pedestrian Crossings. The first-listed research in Table 16-5 is a before-and-after study, covering four U.S. cities, of 11 intersections where painted crosswalks were implemented—or in one case upgraded. (As indicated earlier in Footnote 2, intersection crosswalks legally exist whether they are marked or not.) The study featured a comprehensive quasi-experimental design. Traffic speeds at all 11 locations were signed for 25 miles-per-hour. As indicated, the proportions of pedestrians crossing within the one-block stretch extending 1/2-block to each side who chose to use the cross- walk increased by less than 1 percent up to 12 percent (city averages). The percentage increases were statistically significant in Buffalo and Richmond. There was also an increase of somewhat less than 1 percent in pedestrians overall, not statistically significant, in the crosswalks and the 1/2-blocks to either side (Knoblauch, Nitzburg, and Seifert, 2001). A brief look at crosswalk safety issues is provided in the “Related Information and Impacts” sec- tion under “Safety Information and Comparisons”—“Other Traffic Safety Issues and Findings”— “Street Crossing Safety.” There it will be seen that, out of 11 studies from 1965 to 2005, only two did not find lesser safety in the presence of plain marked crosswalks as compared to unmarked crosswalks (Chu, Guttenplan, and Kourtellis, 2007). The four-city study of 11 intersections described above is one of the two studies not identifying lesser safety where crosswalks without traffic con- trols were marked. The four-city study results do mesh, however, with findings of a 2005 study that looked separately at uncontrolled crossings of two-lane streets and uncontrolled crossings of multi-lane arterials. (An uncontrolled crossing in this context is one with no stop sign or signal on the crosswalk approaches.) That study found no significant difference in vehicle-pedestrian crash rates with or without crosswalk markings where two-lane or low-traffic volume streets were involved, but sev- eral times higher crash rates with marked crosswalks than without where multi-lane roads with higher volumes were involved (Zegeer et al., 2005). The 2005 study is included as the 2nd entry in Table 16-5 because of its finding that marked cross- walks seem to be especially attractive to the young and the elderly. Whereas 66 percent overall of all pedestrians observed at all 1,000 pairs of crossings studied used the marked crosswalks in the marked/unmarked pairings, the proportion for persons age 65 and older having to cross four-or- more lanes was 81 percent. The proportion for children up through age 12 under comparable con- ditions was 76 percent.11 The four-or-more lane facilities (with more than 12,000 or so vehicles per day) involved precisely the types of roadways found to have the higher crash rates within marked as compared to unmarked crosswalks. Marked crosswalks in this study excluded any with active warning devices (Zegeer et al., 2005). The 3rd and 4th entries in Table 16-5 are case examples that do not appear to have extensive travel demand research behind them. Nevertheless, they illustrate that conventional urban signalized intersection design is much more attractive to pedestrians than typical rural design, even when the latter is signalized, and that a well designed traffic circle installation with pedestrian safety and traffic calming features is more attractive to walkers than an unfriendly intersection (PBIC and APBP, 2009, Harkey and Zegeer, 2004). 16-50 11 Caution should be used in deducing crossing choice behavior from these percentages. The original selec- tions of which crosswalks to mark may have reflected already-established crossing-volume characteristics (Zegeer, 2011).

The 5th table entry is of special interest because it covers, apparently with “before and after” eval- uation, a sector approach to crossing improvements. When one-way circulation was reversed in the Shoreditch Triangle, an East London arts and entertainment destination adjoining the finan- cial district, pedestrian needs were examined in terms of desire lines and pedestrian concentra- tions. The number of signalized crossings was increased, their locations were aligned with the pedestrian desire lines, and road space was reallocated to widened sidewalks where need was indi- cated. Among quantitative findings listed in Table 16-5, it is notable that a 9 percent increase in total pedestrian crossings was identified in the evaluation (Booz Allen Hamilton, 2006). Given the system approach, it may reasonably be presumed that this overall increase does not reflect walk- ing route shifts, but rather changes in travel mode, choice of destination, and/or amount of walk- ing activity. Bicycle or Mixed NMT Mode Crossings. The more extensive investigation encountered of effects on bicycling of crossing conditions is from Portland, Oregon. It is more fully described in the “Bicycle Lanes and Routes” subsection under “Popularity, Preferences, and Route Choice”—“GPS- and-Network-Based Revealed Preference Research,” and is not included in Table 16-5. In this research, an explanatory model was developed based on the routes a cross-section of cyclists were observed to use, compared to the minimum-path routes available to them. The study approach was a form of cross-sectional analysis, not a before-and-after study, and focused on route choice rather than mode choice or propensity to cycle. The explanatory-model results provide elasticities to quantify the negative route choice effects of various intersection conditions in terms of presence or lack of traffic controls. Effects for bicyclists making a right turn were found to be minimal, and the results summarized here pertain to cyclists not turning right. It was found that even having to pass through stop signs and signals was a mea- surable deterrent to use of a route, reflected in negative route choice elasticities of −0.24 for num- ber of stop signs encountered per kilometer and −0.28 for number of traffic signals per km. The negative effect was only slightly stronger for encountering one unsignalized crossing per km. of a street with daily vehicular volumes in the 10,000–20,000 range (elasticity of −0.33). On the other hand, the negative effect was 3 to 4 times as substantial for one unsignalized crossing per km. of streets with vehicular volumes over 20,000, producing an “elastic” value for elasticity of −1.08 (Broach, Gliebe, and Dill, 2009a).12, 13 More immediately graspable statistics have also been produced through further application of the Portland route choice model. It is estimated, for example, that the typical cyclist will go 1.5 percent 16-51 12 An elasticity for route choice response to traffic control conditions of −0.3, for example, indicates an 0.3 per- cent decrease (increase) in route choice probability in response to each 1 percent increase (decrease) in the crossing condition examined, calculated in infinitesimally small increments. The negative sign indicates that the effect operates in the opposite direction from the cause. An elastic value is 1.0 or greater (negative or pos- itive), and indicates a demand response that is more than proportionate to the change in the impetus. Elasticities reported in this chapter are thought to be point elasticities or closely comparable values, although none were explicitly defined in the source documents. (For additional background, including application procedures, see “Concept of Elasticity” in Chapter 1, “Introduction,” and Appendix A, “Elasticity Discussion and Formulae.”) 13 All numerical values presented here that are based on the Portland bicyclist route choice modeling derive from the initial research model of 2009, which encompasses all utilitarian trip purposes in a single model. For information about subsequent modeling, see Footnote 16 in the cross-referenced “GPS-and-Network- Based Revealed Preference Research” discussion.

out of their way to avoid one more stop sign per mile, and 2.5 percent to avoid one more traffic sig- nal per mile. (Cyclists apparently do not like turns in their route, either: it is estimated that they will go 6.5 percent out of the way to avoid one more turn per mile.) With respect to avoidance of unsignalized traffic conflicts, in the context of a 3.5 mile trip, it is estimated that cyclists will deviate by 16.5 percent to avoid each unsignalized major arterial crossing. Comparable values for other conflict/delay situations include 2.5 percent per minor unsignalized arterial crossing, 11.5 percent for each unsignalized left turn from a major arterial, and 4.5 percent for each unsignalized left turn from a minor arterial (Broach, Gliebe, and Dill, 2009b). The 6th entry in Table 16-5 touches on barriers to trail use by neighborhood pedestrians and cyclists. Although it fails to find significant impact of actual GIS-determined need to cross a busy arterial, to access a shared use trail, it estimates that the perception of such need can cut trail use in half (Troped et al., 2001). The 7th entry provides evidence of major bicyclist volume increases on urban cycle tracks (bike lanes with physical separation from traffic) in response to carrying the cycle tracks through intersections in the form of raised speed tables (raised crossings intended to alert and slow vehicular traffic) (Gårder, Leden, and Pulkkinen, 1998). School Access Street Crossings. The 8th through 10th entries in Table 16-5 pertain to situations where students must cross streets on the way to school. The 8th entry provides an extraction of street crossing improvement response information from the early California Safe Routes to School (SRTS) program. The study itself is further described within this “Response by Type of NMT Strategy” sec- tion under “NMT Policies and Programs”—“Schoolchild-Focused Programs.” Assessments based on child route-to-school choices obtained in surveys of parents indicated that the effectiveness of intersection improvements was only moderately less than the impact of paved sidewalk projects such as sidewalk gap closures. Before-and-after 2-day child pedestrian counts showed a 24 percent increase in schoolchild usage of the two intersections that had been newly signalized. Counts at the three intersections receiving crossing improvements without traffic signal installation were incon- clusive, especially where only a marked crosswalk was provided, in contrast to the in-pavement crosswalk lights deployed at the other 2 locations (Boarnet et al., 2005a and b). The 9th and 10th Table 16-5 entries cover four research efforts in Australia (two studies), Switzerland, and Oregon that used cross-sectional analysis to examine the effect on walking and bicycling to school of necessity to cross multiple, busy, or major roads. A negative impact was iden- tified in three out of the four studies. In addition, the first of the two Australian studies isolated a negative impact for lack of traffic signals and crossings (Davison and Lawson, 2006, Moudon, Stewart, and Lin, 2010). Pedestrian and Bicycle Grade Separations Constructing pedestrian and/or bicycle grade separations—overpasses (or bridges) and under- passes (or “subways”)—entails major capital investment to achieve traffic safety through total seg- regation of motor vehicle and crossing NMT traffic. Pedestrian and bicycle grade separations are used where roadway volumes, conditions, NMT volumes, user group characteristics, or facility type cannot reasonably accommodate at-grade pedestrian crossings. However, if not carefully placed and designed, there may be drawbacks in addition to the investment cost. Walkers and cyclists have a basic resistance to changes in elevation and often avoid using special grade-separated facilities to cross roadways. In addition, such facilities may isolate or obscure pedestrian activity and thereby generate personal safety concerns (AASHTO, 2001, Zegeer, 1998). The grade and safety concerns may not apply in greenway and other applications where topography is favorable and visually open construction is possible. Boulder, Colorado, offers many examples. 16-52

As with other pedestrian and bicycle facilities, travel time is an important determinant of use. Those facilities where land uses or topography permit direct connections without large up or down grade changes may be the most successful (Zehnpfenning et al., 1993, Moore and Older, 1965). Where pedestrian bridges are integrated with second-storey land development and connected with one another, they become skywalk systems as covered in the upcoming “Pedestrian Zones, Malls, and Skywalks” subsection. Bridges over major barriers are examined further on in this chapter under “Pedestrian/Bicycle Systems and Interconnections”—“River Bridges and Other Linkages.” Many pedestrian bridges and “subways” were built in the middle decades of the 20th Century to provide safe school crossings of major arterials, and this type of application remains relevant in special applications. The 11th entry in Table 16-5 provides an example from Phoenix. Note that a crossing guard has to be employed to enforce bridge use by students (Harkey and Zegeer, 2004). Omaha, Nebraska, is an example of a city with a history of using pedestrian overpasses as a strat- egy to provide safe routes to school for children. In 2000, Omaha had 23 pedestrian overpasses, many built in the 1970s. The city traffic engineer found that children over age 11 or so tended not to use the bridges, but instead to cross at-grade, engendering proposals to meet future crossing needs with traffic signals or crossing guards (Urban Transportation Monitor, 2000). This is not always feasible, of course, with higher-type highway facilities. The inclination of pedestrians to choose the shortest path is made note of, in connection with inter- preting observed phenomena, at a number of points in this chapter. Perhaps nowhere is this ten- dency so vividly illustrated than in the case of pedestrian grade separations. Grade separations are costly investments, yet if adolescent and adult pedestrians can save time and effort by avoiding them, many (or most) will do so. This phenomenon began to attract attention early on, and in the 1960s, an extensive study was made in Great Britain of the travel route choice outcomes of pedestrian decisions to use or avoid pedestrian grade separations (12th entry in Table 16-5). Figure 16-1 illustrates the striking results. With percentages of pedestrians choosing to use a bridge or “subway” crossing plotted against a value “R,” which is the ratio of time via the grade-separated route to time via an at-grade route (and which may be interpreted as a convenience measure), a highly sensitive response to travel time is shown. In the study, virtually no one used an overcrossing requiring 25 to 50 percent more crossing time (R = 1.25 to 1.5) than the at-grade route. Undercrossings were shown to be slightly more attractive, perhaps because they typically involve lesser grade changes to access. With equal travel time via either the grade separation or an at-grade route, the study results suggest (at least under 1960s urban English conditions) that an underpass will be chosen by 95 percent of pedes- trians and an overpass will be chosen by 20 to 70 percent (Moore and Older, 1965, Zegeer, 1998). The 13th and final entry in Table 16-5 gives an example of a suburban pedestrian/bicycle over- crossing with peak-period and Saturday midday 2-hour NMT volumes in the 13 to 35 users range (Pedestrian and Bicycle Information Center, 2010). Such volumes would not meet quantitative grade-separation justification warrants such as those in the 1984 Federal Highways Administration Report No. FHWA/RD-84-082, “Warrants for Pedestrian Over and Underpasses,” but they could well pass muster under more qualitative benefit analyses that include such NMT system connec- tivity considerations as maintenance of neighborhood continuity and support of existing and future land uses (Zegeer, 1998). Guidebooks such as NCHRP Report 240: A Manual to Determine Benefits of Separating Pedestrians and Vehicles, offer procedures for structuring grade separation benefit analyses involving multiple con- siderations, many not readily quantifiable (Roddin, 1981). The struggle to address unknowns such 16-53

as potential pedestrian and bicyclist use of grade separations bridging barriers of long standing can be clearly seen in innovative approaches such as New Jersey’s efforts to prioritize pedestrian and bicycle crossings using such tools as pedestrian potential indices, bicycle demand models, and GIS systems (Swords et al., 2004). Pedestrian Zones, Malls, and Skywalks Pedestrian zones, malls, and skywalks all serve to more extensively separate walkers from motorists, and to provide more walking space, thus facilitating pedestrian travel. Historically, however, the impetus for establishing the pedestrian treatment has often been less about trans- portation than about efforts to secure the economic health of a business district, normally a central business district (CBD). Installations of pedestrian zones, malls, and skywalks are often intended as strategies for stabilizing or enhancing the viability of CBD retail and office space (Robertson, 1992 and 1994). Economic perspectives, although introduced here, are quantified primarily under “Economic and Equity Impacts,” in the “Related Information and Impacts” section. Although the same general themes run through all three project types, in this subsection experiences with skywalks are looked at separately from pedestrian zones and malls. Skywalks, and their underground concourse coun- terparts, are unique in providing total separation from street traffic. They also have had a signifi- cantly lower failure rate in terms of overall success. Pedestrian Zones and Malls The distinction between pedestrian zones and pedestrian malls is not clear-cut. Pedestrian zones are areas in which vehicle traffic is restricted and pedestrian travel is encouraged, typically com- 16-54 Figure 16-1 Street crossing route choice in response to pedestrian grade separation Source: Moore and Older (1965).

posing a small-area network of pedestrian streets in an urban commercial core. In most countries, enhancing central area commerce has been the main impetus, although Sweden is reported to have placed priority on enhancing pedestrian and traffic flow and safety. Many cities outside North America now have such areas in their core, with some in Europe dating back to post-World-War-II reconstruction. There has been steady growth, until today there are over 1,000 cities with such treat- ments in Germany alone. There are only very few strictly comparable examples in the United States. More common in the United States are pedestrian malls, created by closing and beautifying a single street, albeit often for several blocks or sometimes involving sections of two intersecting streets. The vast majority— at one time found in some 200 cities across the United States—were implemented in the 1960s and 1970s in a wave of interest in city center revitalization. The pedestrian mall was envisioned as an enticing alternative to the suburban shopping center (Robertson, 1994 and 1995). Physical and Economic Context. To understand the travel impacts of downtown pedestrian malls in the United States, it is necessary to appreciate both what physical forms they can take and how they have been affected by secular (long-term) trends. “Traditional Pedestrian Streets,” including the vast majority constructed in the 1960s/70s, are designed for pedestrian use only, and a num- ber have even given pedestrians the right-of-way at cross streets. “Shared Malls” are predomi- nantly pedestrian but accommodate a narrow traffic-calmed passage for vehicles, typically a single lane with or without parking. “Transit Malls” are likewise pedestrian oriented but with exclusive transit vehicle lanes and amenities for waiting passengers. These three basic facility types generally apply to both foreign and domestic pedestrian zones and malls, except the European pedestrian streets are typically minimalist in landscaping (if any) and street furniture. They are more dedicated to accommodating substantial pedestrian volumes on narrow street cross-sections (Robertson, 1994). In addition, recent systems approaches to pedestrianization have used combinations of facility types to match specific needs. New York City (Broadway) and Oxford, England, examples are covered here under a “Combination Projects” classification (New York City Department of Transportation, 2010, Booz Allen Hamilton, 2006). With regard to secular trends, many traditional pedestrian streets (malls) were superimposed in the 1960s/70s on U.S. downtowns in decline. Many were unable to stem the tide toward suburban shopping in the years to follow. In a dying downtown with low pedestrian volumes, even a thoughtfully designed and promoted pedestrian mall can feel empty and actually discourage fur- ther use. Since 1980, few new traditional pedestrian streets have been implemented and many existing malls have either been totally “re-streeted” or altered into shared-mall constructs. Norfolk, Virginia, and New London, Connecticut, for example, have converted pedestrian malls back to streets. For some downtowns, this has recreated a sense of street life by concentrating pedestrians on smaller walking spaces and reintroducing the bustle of motor vehicle traffic (Levinson, 1986, Project for Public Spaces, 1993, Robertson, 1993 and 1995). In other cases, pedes- trian streets have been and are highly successful. Given the circumstances, the limited amount of pedestrian activity response data there is on pedestrian zones and malls must be viewed through the lens of short-term impact, recognizing that short-term gains may have become permanent in a stable or thriving economic environment, or may have been negated in an environment of over- all area decline. Pedestrian Zones. Perhaps the only U.S. location where an area-wide vehicle traffic restriction has been introduced is the 12-block Downtown Crossing in Boston, Massachusetts. Within and around the zone are about 125,000 employees and numerous retail establishments. Boston’s 16-55

pedestrian zone is actually characterized by the narrowness of streets so common in European applications. At critical points in the core retail district, original sidewalk widths averaging 9 to 10 feet had effective widths of only 5–6 feet because of obstructions. Pedestrian levels of service on four contiguous blocks of Washington Street in the zone were E, C, D, and E, on a scale of A to E (collisions probable), with worse conditions at intersections. The Downtown Crossing project was created in 1978 by closing 2/3 of the street segments in the zone to general traffic while improving the transit service and parking management. Some of the street segments were made pedestrian-only while others continued to allow transit service and taxis. Several local bus routes were extended into the zone. Street furniture, brick pavement, new lighting, and information kiosks were introduced in 1979. The changes eliminated sidewalk con- gestion on the affected streets along with conflicts at the affected intersections along Washington Street. When surveyed, both businesses and pedestrians responded positively about the project (Weisbrod and Loudon, 1982, Replogle, 1995). Pedestrian activity and store purchases increased overall following closing of the streets, despite increasing competition from other areas. Much of the increase was attributed to midday activity by nearby office workers. The volume of 10:00 AM to 4:00 PM weekday visitors to the area was up 11 percent in 1980 compared to 1978, from 74,200 to 82,400, based on pedestrian counts. Weeknight visitors from 6:00 PM to 8:00 PM were up 8 percent, from 11,300 to 12,200, and 10:00 AM to 4:00 PM Saturday visitors were up almost 10 percent, from 57,800 to 63,400. Pedestrian volumes between noon and 2:00 PM, as a percentage of total weekday pedestrian volumes, went from 45.8 to 48.4 percent. Sidewalk volume changes varied throughout the pedestrian zone. The northern blocks, closest to the government and financial office districts, saw pedestrian traffic increase by more than 15 per- cent. Southern blocks experienced pedestrian traffic decreases that were similar percentage wise, but smaller in the absolute. The largest volume increase occurred on a northerly block of Washington Street that had sidewalk widening rather than total pedestrianization, leading the researchers to conclude that proximity to activity generators can be more of an influence than the form of auto restriction. Weekday pedestrian volumes on the block in question were 38,000 in 1980 during a 6-hour period, including 8,000 pedestrians/hour volumes during the mid- day peak. Over the 2-year 1978 to 1980 period, the weekday mode shares for worker and shopper trips into and in the pedestrian zone shifted from 48 to 54 percent walk, 37 to 39 percent transit, and 11 to 6 percent auto. As shown in Table 16-6, weeknight and Saturday walk shares also increased, with auto shares decreasing, but transit usage shifts varied in direction. The increased weekday transit usage was a result of the extension of the bus routes, but given the no-free-transfers fare structure of the time, it came at the expense of subway transfer revenue. The net increase in fare box revenue overall covered only 5 percent of the cost of the extended service in the first year. The project evaluation also examined effects on Downtown Crossing worker’s and shopper’s mode choice for the trip from home to downtown Boston, as contrasted to the trip into the pedestrian zone, which often started from elsewhere in the downtown. These home-based walk and transit shares either held essentially constant or increased, as illustrated in the shaded portion of Table 16-6, and the corresponding auto travel to downtown was down for all time periods (Weisbrod and Loudon, 1982). 16-56

Limited information is also available on effects of overseas pedestrian zone implementations, but must be inferred from available retail sales statistics. Sales increased by 30 percent on Copenhagen’s Stroget, actually three contiguous streets in the main shopping district, after it was closed to motor vehicles in 1962. Technical studies of 1968 conditions showed the facility to be filled to near capacity with people walking, sitting, standing, and lingering. London Street in East Anglia, England, saw sales increases of 5 to 20 percent (Robertson, 1994). An alternative that has been tried in New York City, and also is used in Japan, is to close streets to vehicle traffic temporarily during certain hours of the day. Midday closure of Fulton Street in lower Manhattan increased pedestrian activity by 11 percent, with nearby workers flooding the street from 11:00 AM to 2:00 PM. An average of 4,132 pedestrians per hour was observed before the clo- sure. After the closure, usage grew to an average of 4,594 pedestrians per hour (University of North Carolina, 1994, Replogle, 1993). Traditional Pedestrian Streets. The first U.S. pedestrian mall opened in downtown Kalamazoo, Michigan, in 1959. By 1978 the documented number of such malls was approaching 100. Non- transportation factors found to be associated with success—generally defined in terms of usage, popularity, and perceived effect on sales—have included development of a sound organizational structure for mall management and preexisting sound economic health of the downtown. A key transportation factor linked to success is presence of a major nearby pedestrian traffic generator such as a college campus, government center, or medical complex (Robertson, 1994). A survey of 36 downtown pedestrian malls taken in 1989 by the City of Eugene, Oregon, found seven malls that “were doing well or great.” Some 25 had either been or were to be removed, or were reported to be doing poorly (Rathbone, 2006). In a number of cases, particularly those with- out supplemental pedestrian traffic generators, success or failure has had more to do with retail economics than transportation issues. A not uncommon example of a mall deemed to be hobbling along when examined in 1988 was the nicely maintained three-block Mall Germain in St. Cloud, Minnesota. Retail included one blank- walled department store and a small-business retail mix observed to be outdated vis-à-vis student and office worker populations nearby. It lacked a 1-block extension that would bring it to the river and a conference center, and was “empty of pedestrians most of the time.” 16-57 Table 16-6 Mode Shifts Accompanying Implementation of the Downtown Crossing Pedestrian Zone in Boston Time Period Mode Shares to Pedestrian Zone Mode Shares to Downtown Boston Year Walk Transit Auto Other Walk Transit Auto Other Week- 1978 48% 37% 11% 4% 10% 62% 23% 5% days 1980 54% 39% 6% 2% 9% 75% 13% 3% Week- 1978 48% 38% 13% 1% 11% 71% 17% 1% nights 1980 60% 36% 3% 0% 12% 80% 7% 1% Satur- 1978 21% 54% 20% 4% 13% 59% 23% 5% days 1980 32% 49% 14% 4% 19% 59% 18% 5% Note: Includes all interviewed visitors to the pedestrian zone study area irrespective of trip purpose. Source: Weisbrod and Loudon (1982).

Another pedestrian street examined in 1988 and found not to be doing well was Westminster Mall in Providence, Rhode Island. Highly active between 11:00 AM and 3:00 PM thanks to workers from the nearby financial district, the mall was relatively deserted otherwise, and devoid of use after 5:00 PM. Negative factors included retail that was declining in the face of intense competition from 12 suburban shopping centers, exacerbated by low levels of mall maintenance and an undoubtedly related perception of crime (Robertson, 1994). Both the Mall Germain in St. Cloud and the Westminster Mall in Providence are among those changed back to a conventional street and sidewalk cross-section. The Westminster Mall decline is an example of the strong role of retail economics in pedestrian mall success or failure. This mall was quite successful for a number of years. The final straw was a business decision by the key retailers to build a conventional modern enclosed mall nearby and relocate. In places that are thriving, and have high pedestrian volumes, traditional pedestrian streets have done well. Examples include Seattle (Occidental Street), San Francisco (Maiden Lane), Las Vegas (Freemont Street), and Santa Monica, California (Third Street Promenade). Other malls perceived to be faring well include those in the college towns of Charlottesville, Virginia; Boulder, Colorado; and Burlington, Vermont. A standout example in Madison, Wisconsin, is actually 3/4 transit mall and 1/4 conventional pedestrian street. Known as the State Street Mall (see “Transit Malls,” below), it links the University of Wisconsin and the state government complex (Harkey and Zegeer, 2004, Robertson, 1994). Shared Malls. A broad-scale overseas application of the shared mall approach is the Japanese “community street” concept. More than 140 were introduced in the 1980s in Japan after a success- ful demonstration project in Osaka. There, a 10-meter wide street was converted into a 3-meter (9.8-foot) wide zigzag space for vehicles. Motor vehicle traffic dropped by 40 percent and pedes- trian and bicycle traffic increased by 5 and 54 percent, respectively (Replogle, 1993). Such applications blur the distinction between shared malls and traffic calming. In the United States, the Santa Cruz, California, Pacific Garden Mall is an example of a shared mall cited as suc- cessful in the previously noted 1988 review. Pedestrian information is lacking, but of the six malls examined, this mall was second highest in number of mall businesses (106 establishments) and in percentage selling retail goods (67 percent), and lowest in ground-floor vacancies at less than 1 per- cent (Robertson, 1994).14 The Portland Mall, covered below under “Transit Malls,” actually has a significant shared mall component. Close cousins to shared malls are downtown streets that have been put on a “road diet,” decreasing the number of traffic lanes, introducing traffic calming, and allowing angle parking, mid-block crosswalks, and/or widened sidewalks. Two successful exam- ples, Hendersonville, North Carolina, and El Cajon, California, were included in Table 16-4 of the earlier “Sidewalks and Along-Street Walking” subsection. Transit Malls. A shared-use approach that creates more activity without necessarily allowing pri- vate vehicles is creation of a “transit mall.” Such malls dedicate a portion of the street right-of-way to use by public transit vehicles, potentially enhancing transit operations and maintaining or adding transit patrons in the pedestrian mix (Levinson, 1986, Robertson, 1993). Different reviews of transit malls have arrived at disparate conclusions about their success. One examination of six U.S. pedestrian malls in 1988 concluded that, as viewed from the perspective of 16-58 14 These 1988 data were collected 1 year before the Loma Prieta earthquake, which in 1989 disrupted the Pacific Garden Mall by destroying more than 1/3 of the buildings along it (Robertson, 1994).

economics and mall vitality, “the most successful malls . . . [were] the three transit malls and the shared mall” (Robertson, 1994). Certain transit malls have arguably not had success. Commercial activity on Howard Street in Baltimore suffered from the construction activity of building street- level Light Rail Transit (LRT) and never rebounded (Calvert, 2001). The State Street transit mall in Chicago was disliked for its concentration of large buses and the thinned crowds spread across too much walking space. It was restored to a conventional streetscape with 22-foot sidewalks, a better match for the pedestrian volumes (Engelen, 2004, Kamin, 2009). A factor in its removal was a reduced bus transportation role once Orange Line rail rapid transit service opened in Chicago’s southwest corridor. An elaborate glass-enclosed transit mall in Canada’s capital city of Ottawa, the Rideau Mall, created a confining space, blocked views of storefronts, and sheltered “undesirables.” Financial difficulties faced by merchants and property owners precipitated a decision to revert back to a traditional street. The negative view is summed up in the contention that, “in nearly every city where they have been built, transit malls are being rethought or have been altered from their original concept” (Project for Public Spaces, 1993). Despite individual failures, U.S. transit mall development has produced solid and enduring suc- cess stories. Lacking a broader quantitative success and failure tabulation, it is instructive to look at the five transit malls and mall proposals selected in the 1970s for an Urban Mass Transportation Administration (UMTA) Service and Methods Demonstration Program (SMD) review (Koffman and Edminster, 1977) and actually implemented. Of the five implemented transit malls, four remain in full use. The Minneapolis, Minnesota, transit mall was given a 1990–1991 upgrade, fol- lowing the original design outlines, to restore it after a quarter century of hard use. The Madison, Wisconsin, and Denver, Colorado, installations apparently remain essentially as built. The Portland, Oregon, bus mall was reconstructed in 2007–09, after 3 decades, primarily to add LRT. Only in Philadelphia has the transit mall examined in the UMTA/SMD review been dismantled. Opinions diverge on whether Philadelphia’s Chestnut Street Mall hastened retail decline or whether retail decline occurred as a result of competition once the nearby Market Street East commercial area was developed. The Minneapolis, Madison, Denver, and Portland transit malls are examined in the following para- graphs. It should also be noted that elements of Boston’s Downtown Crossing pedestrian zone, dis- cussed above, operate—in effect—as transit malls. The Nicollet Mall in Minneapolis was apparently the first U.S. transit mall when built in 1967. It features a 24-foot serpentine travelway for buses, heated sidewalks ranging from 20 to 36 feet wide, and numerous amenities. Originally eight blocks in length, it was extended to 12 blocks in 1982. Circa 1988 ridership on six bus routes serving the mall was 30,000 riders (Robertson, 1994, Project for Public Spaces, 1993). Average 11- to 12-hour pedestrian counts per side for the six blocks cen- tral to retail activity were 12,400 to 12,800 in 1958 well before introduction of the transit mall, 13,600 in 1973 after transit mall development (Koffman and Edminster, 1977), 7,400 in 1976 after Skyway interconnection of major retail centers (Edminster and Koffman, 1979), and 7,200 in 2002 in the con- text of an 8-mile Skyway system. Recent pedestrian count data show that Skyway usage now does tend to dominate, but that the mall holds its own (Bruce, 2002a and 2002c), with 38 to 46 percent of the immediately parallel pedestrian flow on a September day in 2002. Additional detail is provided in the case study “50 Years of Downtown NMT Facility Provisions— Minneapolis,” which concludes that the Nicollet Mall has played a supporting role in the city’s suc- cess in stabilizing and enhancing its downtown area and its NMT attractiveness. As covered in the case study, the block-wide corridor centered on the Nicollet Mall is estimated (as of 2002) to be attracting an 11-hour pedestrian flow averaging 15,600 to 18,700 per side (Skyway traffic included), contrasted to the 12,400 average per side in 1958—an increase of some 25 to 50 percent. (In the case 16-59

study, under “More . . . ,” see Table 16-131 for a full presentation of Nicollet corridor pedestrian flows over time.) The State Street Mall in Madison, Wisconsin, opened in stages between 1977 and 1982, draws much of its layout from the Minneapolis example. Despite being introduced into a downtown suffering from 1960s Vietnam-era turmoil, it is an economic success (see “Related Information and Impacts”—“Economic and Equity Impacts”—“Land Value and Commerce Impacts”—“Downtown Pedestrianization Effects” for quantitative measures). The six-block transit mall is anchored at one end by the state Capitol, with nearly 25,000 government workers, and at the other end by a two- block section of traditional pedestrian street and the University of Wisconsin, which represents 13,000 employees and 44,000 students (circa 1988). The mall is served by 700 buses running on 16 routes. It features a retail mix representative of college towns and has an active night life. Bicycle traffic is a major component of the pedestrian-bicycle-bus mix (Robertson, 1994). A refurbishing plan in 2002 featured a “cleaner look” but retained the original basic layout (Harkey and Zegeer, 2004). Quantitative NMT volume data are not available. The 16th Street Mall in Denver opened in 1982 over a 13-block, 1-mile distance between two concurrently planned urban bus terminals. Along the mall runs a very frequent fare-free low/ no-emissions-vehicle shuttle-bus service, distributing passengers from and to the bus terminals and more recently constructed LRT. Interesting design features and ability to hop on a free bus mitigate the spread-out character of the mall. Bus shuttle operating parameters and results circa 1997 are described in Chapter 10, “Bus Routing and Coverage,” under “Response by Type of Service and Strategy”—“Circulator/Distributor Routes”—“Transit Terminal and Parking Distributors.” Chapter 17, “Transit Oriented Development,” describes changes in downtown Denver’s development regulations and land use mix in that chapter’s “Response by TOD Dimension and Strategy” section, under “Response to TOD by Regional Context”—“City Center TODs.” As reported in Chapters 10 and 17, average weekday free bus shuttle ridership was 47,000 in 1997, and 60,000 in 2004, the latter after extension to serve residents of an adjacent rail-yard redevelopment. Pedestrian usage of the 16th Street Mall was estimated in the late 1980s at 90,000 walkers daily (Robertson, 1994). The Portland Mall, opened in Portland, Oregon, in December 1977, is actually a combined pedestrian/ transit/shared mall through the primary office district and into the downtown retail area. It first served successfully as a bus transit mall, for three decades, followed by reconstruction and addi- tion of LRT in 2007–2009. The twin mall occupies a one-way street pair, 5th and 6th Avenues. As constructed and operated initially, it was 11 blocks in length, with two exclusive bus lanes on each street, plus a lane for general traffic access, 26-foot wide sidewalks along the right sides where buses load, and mostly 18-foot sidewalks on the left. Originally the general traffic lane was inter- rupted every 4th block by a block of 30-foot left-side sidewalk, but the general traffic lane has been made continuous in the 2007–2009 reconstruction. Traffic signal timing is currently set for 12 mph, appropriate for exclusive-lanes transit service with heavy passenger loading and unloading at multiple stops, and allowing bicycles and autos to move together on the general traffic access lane. Originally bicycles were prohibited. Portland bicycle mode shares have increased in recent years (see Figure 16-7 under “NMT Policies and Programs”). One feature of the mall rehabilitation is addition of more bicycle parking, including covered “bike oases” (Edminster and Koffman, 1979, Dueker, Pendleton, and Luder, 1982, TriMet, 2009). Simulation-aided pedestrian estimates circa 1980 indicated that the Portland mall had focused pedestrian activity on the mall area and nearby sections of cross-streets, as compared to a more even distribution of pedestrians (without the mall) on streets in the downtown. Of all downtown 16-60

bus runs, 88 percent had been concentrated on the two mall streets. An average of 13 passengers boarded or alighted at each bus at each stop along the mall in the PM rush hour. Total mall pedes- trian volumes were estimated to be 75 percent bus patrons. (This is a much higher percentage than in Minneapolis, where as detailed in the downtown Minneapolis case study, 16 percent of sur- veyed Nicollet mall pedestrians were headed to or from a bus stop.) Together, these statistics led to an estimate that 800 persons per hour were passing along the average block of the transit mall during the peak hour. The midday estimate was 600 persons per hour. The only pedestrian vol- ume data for the before condition are 1975 counts indicative of a 565 persons per hour average flow on 5th and 6th Avenues through the retail district, mid-morning and mid-afternoon. Portland Mall employees, bus riders, and pedestrians were separately surveyed about their sat- isfaction, using a 5-point scale ranging from “1” for strongly disagree to “5” for strongly agree with various statements. “The Transit Mall is attractive” engendered a 4.2 to 4.6 mean response (between agree and strongly agree) across the three surveys. The means for “The Mall is a good place to shop” ranged from 3.8 to 4.1; “The Mall is a good place for entertainment,” from 2.8 to 3.3 (basically neutral); “The Mall is a good place to relax,” from 2.5 to 3.2; “The Mall is safe,” 2.9 to 3.7; and “The Mall is a good place to walk,” 3.8 to 4.5. In each of these instances, employees were the least affirmative, bus riders were more so, and pedestrians tended to be the most posi- tive by a small margin. Both the intent and results were unique for the survey statement “The Mall sidewalks are crowded.” Here the means were 3.5 for both employees and bus riders (between neutral and agree), while the mean for pedestrians was 2.8 (fairly neutral but tilted toward dis- agreement). The researchers felt these mid-range responses were close to ideal, reflecting enough crowding for comfortable social interaction, but not too crowded. This interpretation meshed with the generally positive responses to the perceived safety and “good place to walk” questions, which were offered despite actual crime statistics suggestive of more off-street crime on the mall- frontage blocks than further away (on-street crime distributions could not be assessed) (Dueker, Pendleton, and Luder, 1982). Combination Malls. Two major pedestrian mall systems employing combinations of mall facility types offer roughly comparative data. Each reports overall project area pedestrian volume increases on the order of 8 percent. Table 16-7 provides a summary. 16-61

“Green Light for Midtown,” the name given to New York City’s pilot project for Broadway in Midtown Manhattan, may seem an odd name for a project involving extensive areas of pedestrian mall (labeled “plazas”). The reference is to the greater traffic signal green time afforded to the Manhattan streets and avenues with removal of Broadway’s diagonal-to-the-grid traffic flows.15 The selection of physical layouts for the different components of the pilot project, from Columbus Circle (59th Street) south to 23rd Street (with progressive extension further south), had much to do with traffic lane layouts and conflict-elimination intersection geometrics designed specifically to improve Midtown vehicular traffic and pedestrian crossing conditions. From 47th Street to 42nd Street, inclusive of Times Square, and again from 35th Street to 33rd Street inclusive of Herald Square, all roadway space exclusive to Broadway has been converted to pedes- trian plazas with tables, chairs, and awnings. Street space on the alignments of 7th Avenue (at Times Square) and 6th Avenue (at Herald Square) remain dedicated to vehicular traffic flow. From 59th Street south to 47th Street, and 42nd Street to 35th Street, plus 33rd Street to 23rd Street (and on south in project extensions), a partial-mall cross-section has been employed. Most partial-mall blocks from Columbus Circle to Herald Square have two southbound traffic lanes, two lanes of parking, a southbound buffered bike lane, and a minimum of one traffic lane’s worth of added pedestrian space with seating. South of Herald Square, Broadway is narrower, there is only one southbound traffic lane in many blocks, and the narrow reserved space has no mall furniture in the pilot project arrangement. 16-62 Table 16-7 Summary of System-Scale, Combination Pedestrian- Mall-Type Application Effects in New York City and Oxford, England Study (Date) Process (Limitations) Key Findings 1. New York City Department of Transportation (2010), Grynbaum (2010), Philip Habib & Associates (2011) Pilot project partial-mall-with-bike- lane and full-pedestrian-mall combi- nations for all of Broadway within Manhattan’s Midtown, imple- mented in May, 2009. (No means for separating pedestrian attractiveness effects from strong secular trends.) Times Sq. pedestrian volumes along Broadway and 7th Ave. up 11% overall, Herald Sq. pedestrian volumes up 6%, but 80% fewer walking in road at Times Sq., injuries down 35% for ped- estrians, 63% for vehicles, net bus and car traffic effects neutral to positive. 2. UK Department for Transport – 2004 as summarized in Booz Allen Hamilton (2006) Oxford, England, June 1999 central- area closure of Cornmarket St. to all traffic, daytime closure of High St. to all but cyclists, buses, and taxis, Broad St. closure to through traffic, bicycle network improvements, 300 more bicycle parking spaces. (Analysis approach not reported.) Central area pedestrian flows up 8.5% (6,000/day) 1998-2000, 11% bicycle mode share maintained, local bus and park-and-ride use up 50% 1991-2000 (2,000/day) , parking at 3 central facilities down 14% (700 cars/day) relative to 3 previous years, total attraction of people to central area up. Note: See this section for more information. Sources: As indicated in the first column. 15 Broadway angles across 10th to 4th Avenues and roughly 77th to 17th Streets, creating numerous awkward intersections and problematic traffic conflicts as it follows the pre-street-grid trace of the original road from Albany (Grynbaum, 2010, New York City Department of Transportation, 2010).

The term “shared mall” has not been applied here to the partial-mall blocks given the lack of overt traffic calming, although the remaining traffic lane(s) do generally have parking on both sides. With a bike lane in all partial-mall blocks, the design clearly draws from “complete streets” principles. Bike lanes are provided in all blocks except those with a full-mall cross-section. In some blocks, left turn lanes are included (Grynbaum, 2010, New York City Department of Transportation, 2010). It was announced in early 2010 that the pedestrian plazas would be made permanent even though not all traffic congestion relief goals had been met (Urban Transportation Monitor, 2010). The project substantially increased sidewalk and other pedestrian space. It needs to be understood that the term “square” is a misnomer as it applies to Times and Herald Squares. There is essentially no outdoor public space not in the street-right-of-way except for narrow triangles at each square. Times Square reputedly has the highest concentration of pedestrians in the world. With the new pedestrian plazas in place, sidewalk flow is vastly improved, in part because “stopping” activities such as reading a map, taking a picture, or looking at billboard displays now tend to take place in the plazas instead of on sidewalks as before. Pedestrian flows have increased in part from growth in pedestrian visits and in part because higher pedestrian capacity has allowed choice of more direct routes by those who formerly deviated out of the way to avoid the pedestrian congestion. Pedestrian volumes at Times Square increased by 17 percent on the most historically crowded sidewalk sections and by up to 112 percent on popu- lar crosswalks, which are now afforded more crossing time. As noted previously, overall weekday volumes are reported (apparently on the basis of peak hours averages) to have increased by 11 per- cent in the Times Square area and by 6 percent in the vicinity of Herald Square (New York City Department of Transportation, 2010). These impacts have occurred in a context of pedestrian traffic that has been increasing over the course of a decade, likely in response to area revitalization. In 2010, cumulative growth in pedes- trians on summer Wednesdays for an aggregation of 14 Times Square area traffic counts reached 50 percent for the 11 years since 1999. Saturday summer counts grew 89 percent over the same period, including a sharp increase between 2009 and 2010. The busiest section of 7th Avenue in the Times Square area, between 43rd and 42nd Streets, handled 109,793 pedestrians between 8:30 AM and midnight during Wednesday, August 11, 2010, counts. During the same hours, the highest count on Broadway, between 46th and 45th, was 52,897 on the sidewalk plus 43,419 in the plaza (Philip Habib & Associates, 2011). Herald Square pedestrian capacity improvements have allowed peak hour pedestrian flow increases in the range of some 30 to 60 percent (New York City Department of Transportation, 2010). Satisfaction with the “Times Square experience,” before and after pilot project implementation, grew from 80 to 91 percent for Tri-state residents, 78 to 89 percent for New York City residents, and 43 to 74 percent for Times Square area employees. The percentage of Broadway pedestrians agreeing that “I would avoid walking on this part of Broadway if I could” dropped from 28 to 16 percent, “It is too crowded here” dropped from 62 to 45 percent, and “I feel safe crossing the street here” increased from 80 to 90 percent. Pedestrian signal compliance shifts ranged from slight improvement at several Times Square locations to a 36 to 82 percent compliance increase at 7th Avenue and 47th Street. Herald Square compliance changes, although generally starting from lower levels, were comparable. Pedestrian and traffic injury reductions of 35 and 63 percent, respectively, attributed to simplified intersections and shortened crosswalks, are noted in Table 16-7 along with a summary bus service and traffic flow assessment (New York City Department of Transportation, 2010). 16-63

The central area pedestrianization project in Oxford, England, employed separate approaches on three different streets, as delineated in Table 16-7. The street closed to all traffic, Cornmarket Street, is a major shopping street. A central area pedestrian flow increase of 8.5 percent was measured between the 1998 “before” year and the 2000 “after” year, reversing a declining trend. The 11 per- cent bicycle mode share, which held steady, includes a journey-to-work bicycle share of 17 per- cent. These are among the highest bicycle shares in the United Kingdom. The project involved a bus priority route, around the central area, with general traffic pushed outward. Nevertheless, sur- rounding streets did not experience traffic volume changes. Opinion surveys, starting in 1993, “show overwhelming public support for the Strategy” (Booz Allen Hamilton, 2006). Pedestrian Skywalks Pedestrian skywalks offer direct connections among buildings, parking facilities, and transit ter- minals, including peripheral facilities. Most U.S. skywalk networks are above grade (hence the name) and, between street over-crossings, link through buildings or above alleys at the second story. At least 16 cities in the United States and Canada have downtown skywalk networks that interconnect 12 or more city blocks. Cities with skywalk systems include Calgary, Cincinnati, Des Moines, Duluth, Minneapolis, St. Paul, and Sioux City. Several systems have underground seg- ments as well. A notable underground equivalent to skywalks is the Houston Downtown Tunnel System, while the largest underground system may be in Montreal, connecting some 300 retail and business establishments plus the area’s subway system. Toronto’s “PATH” underground network is nearly as large. There are also systems of underground concourses in additional North American cities, large and small, including Rochester, Minnesota, and the “Oklahoma City Underground.” Skywalks designed for general public use form a network of walkways that allow pedestrians to travel from one location to another—typically within a downtown—without having to deal with motor vehi- cle conflicts or weather. In addition to providing safe traffic crossings and weather protection, they usually offer a climate controlled environment with retail opportunities. Some in-building compo- nents resemble a shopping mall interior. Skywalks save travel time for many downtown trips, because of avoiding street crossings, and offer time-saving access to quick-stop retail services along their corridors. The first skywalks were built in the 1960s in cold weather cities, where they have continued to be expanded. Later, air conditioned skywalks were developed in warm weather cities such as Charlotte, Dallas, and Fort Worth (Corbett, Xie, and Levinson, 2008, Robertson, 1993, 1994, and 1995, Wikipedia, 2009, Bandara et al., 1994, Podolski and Heglund, 1976, Heglund, 1980). Skywalk Impacts on Walking. No studies have been encountered that explicitly examine the rela- tionship between presence or extent of skywalks and prevalence of walking, although a rough esti- mate of induced walking is provided below following Table 16-8. However, historic counts of pedestrian and transit passenger volumes and mode shares at the Minneapolis CBD cordon along with corresponding data on the extent of the Minneapolis Skyway system help to assess the role of that city’s extensive system in downtown travel choices. The cordon and Skyway data are pre- sented and examined in “50 years of Downtown NMT Facility Provisions—Minneapolis,” in the “Case Studies” section. The case study finds that NMT cordon volumes have been heavily influ- enced by economic conditions, but that overall an NMT growth of roughly 1/2 of 1 percent per year since the mid-1960s can be discerned. Circumstantial evidence suggests a correlation with the development of the Skyway system and more recent introduction of bicycle facilities, while the Nicollet transit mall appears to play a positive supporting role. An analysis covering 6 individual years from 1969 through 1974 indicated that opening of new Skyway bridges in Minneapolis was accompanied by at least a proportional increase in total 16-64

Skyway system bridge crossings. Average December/July daily Skyway crossings were 10,100 in 1969 with two bridges and 11,600 in 1974 on nine of 10 bridges, with the excluded bridge function- ing primarily as an intra-hotel facility. In between, with five bridges not forming a cohesive sys- tem, the average sagged to 8,600 daily pedestrian Skyway crossings per bridge. Downtown redevelopment was a factor, including opening in 1973 of the multi-level interior Crystal Court, which linked separated parts of the system (Podolski and Heglund, 1976). As of 2002, with the Skyway system having grown to 82 bridges, it appeared that traffic per bridge in the core area had held steady (if one adjusts for the post-9/11 economic downturn), while aver- age volumes on the outer reaches of the now-vast system were less. In September 2002 counts, the volume average for the nine core area bridges counted (out of 15 internal to the three by four block area originally connected as of 1974) stood at 10,050. The range for this area was 17,100 to 4,700 per bridge (Bruce, 2002a). Retail and office center Skyway volumes grew/rebounded by almost 1/4 between 2002 and 2007. Thus the proportional growth assumption may still hold for the core area, and then some. However, the 24 less-central Skyway bridges counted (out of 67 lying outside of the core area) averaged 3,700 in 2002. These lower counts suggest that overall system bridge vol- ume averages will drop as a skywalk system is extended beyond a certain point to serve additional businesses, garages, transit terminals, public buildings, and the like. The range for these “outer” 16-65 Table 16-8 Noon Hour Pedestrian Usage by Month in 1974 of Six Twin Cities Skyways in Comparison to Competing Crosswalks W eekday Noon Hour Volumes Percent Using Skyway Month 6 Skyways Crosswalks Totals Highest Lowest Average January 16,400 5,400 21,800 90% 62% 76% February 18,600 6,600 25,200 86% 48% 72% March 19,400 6,400 25,800 85% 50% 71% April 15,000 9,000 24,000 78% 62% 66% May 10,800 11,400 22,200 75% 36% 56% June 10,400 13,000 23,400 76% 25% 46% July 10,200 11,800 22,000 79% 26% 47% August 10,000 12,000 22,000 66% 30% 47% September 11,600 11,800 23,400 76% 36% 52% October 13,000 11,000 24,000 76% 32% 60% N ovember 15,800 8,400 24,200 95% 24% 68% December 17,200 5,600 22,800 82% 51% 67% Notes: N oon hour pedestrian counts made on three Mi nneapolis and three St. Paul Skyways in 1975, when the extent of each system was 4 blocks north-south and three to four blocks east-west. The first two columns of volumes are weekday noon hour subtotals for the six Skyways and for the competing crosswalks. Both are scaled from hand-graphed Figure 5 in the source document. Discrepancies in totals, and vis-à-vis the average Skyway usage percentages (from Table 3 of the source document), have not been fully resolved. Newly computed totals are substituted for the graphed totals, reducing monthly aggregate noise in the surviving record of this unique data set to an equivalent of roughly plus or minus 800 to 2,800 pedestrians per volume observation total (equivalent to 4 to 12 percent of the individual monthly observation totals). Two competing crosswalks were counted for each Skyway crossing to give total inter-block pedestrian flows. For example, in the case of the east-west Skyway crossing of Mi nnesota St. between 5 th and 6 th Sts. in St. Paul, the east-west Mi nnesota St. crosswalk at the north end of the block ( 6th St.) and the corresponding crosswalk at the south end of the block (5 th St.) were counted (Podolski and Heglund, 1976). Source : Adapted from Heglund (1980), with substi tute totals by the Handbook authors.

downtown Minneapolis bridges was 14,400 pedestrians (actually within the core area if one assumes it has shifted one block south over time) to 400 pedestrians (Bruce, 2002a, 2009, and 2002b; Case Study, “50 years of Downtown NMT Facility Provisions—Minneapolis”). Analyses covering specific aspects of walking choices as affected by skywalks, mostly based on Twin Cities Skyway data, also exist. Spring of 1975 all-day counts found roughly 1/4 of weekday Skyway crossings to occur during the noon hour (23 percent in Minneapolis and 28 percent in St. Paul). Nearly 1/2 occurred during the 3 hours from 11:00 AM to 1:00 PM. All-day counts on a selected Twin Cities Skyway bridge and its competing crosswalks showed the percent of pedestrians using the Skyway to be relatively constant throughout the day, ranging for the Skyway in question between 40 and 65 percent choosing the Skyway. The midday Skyway choice averaged close to 50 percent, while the lowest percentages occurred at the beginning and end of the business day, when many pedestrian trips start or end at street-level bus stops. These statistics pertain to relatively small systems—four blocks each way in extent—as compared to today’s Twin Cities Skyway systems with their exten- sions to serve peripheral parking and (in the case of Minneapolis) transit terminals. The 1974 counting program in Minneapolis and St. Paul included monthly noon-hour counts dur- ing a 12-month period that saw no system expansion. Table 16-8 shows the results in terms of vol- umes on six representative Skyways and their competing crosswalks, and reported ranges and averages of the percentage of pedestrians choosing Skyway use in preference to street-level cross- walks. A notable finding is that total volumes throughout the year of Skyways plus competing crosswalks varied relatively little, no more than 10 percent from the average. The lowest and high- est totals both occurred in freezing-temperature months (January and March), suggesting little relationship between total volumes and season. The downtown pedestrians simply shift more to the Skyway systems in cold-weather months (Heglund, 1980). The data in Table 16-8 lends itself to a parametric exploration of how much downtown pedestrian travel may be induced by the presence of skywalks in a cold-climate city like Minneapolis or St. Paul. The 1975 counts found 71 percent of pedestrians traversing Skyway-served blocks to be using the Twin Cities Skyways in the 6 months of November through April, compared to 48 per- cent in summer months. Thus roughly 1/3 of the November-April users were cold-weather users only. If 1/4 of the observed bridge crossings during those months are assumed to represent walk- ing that would not occur without the weather protection of the Skyway systems, then the induced pedestrian blocks of travel represent 9 percent of the total annual observed walking and 15 per- cent of the Skyway traffic. Other parametric trials deemed reasonable by the Handbook authors give an induced-walking range of 6 to 12 percent of total annual observed walking and 9 to 20 per- cent of Skyway traffic. Whatever induced walking there actually is would represent some combi- nation of new walk trips and walk trips that are longer than they otherwise would be. Evidence exists that skywalk systems do encourage longer walk trips, though all that can be stated with certainty is that Skyway trips have been observed to be longer than sidewalk trips in one late 1970s study in St. Paul. There the median CBD walk journey via sidewalks was found to be approximately 2-2/3 blocks, while the median for trips making use of the Skyway system was some 3-1/3 blocks. The proportion of sidewalk trips in any given trip distance increment dropped off fairly steadily from 1-1/2 blocks on, while the proportion of Skyway trips across distance increments held relatively steady up to 4-1/2 blocks in length (Barton-Aschman, 1978). Indeed, the sharp drop-off in Skyway trips after 4-1/2 blocks may possibly have been a product of the limited extent of the St. Paul Skyway system at the time. Responses to a five-city skywalker preference survey, conducted in 1985 and summarized in Table 16-9, articulate for a broader range of cities both the year-round appeal of skywalks to pedes- 16-66

trians and the variations which do occur in response to outside temperature. Preference for sky- walk over sidewalk was reported by 72 to 100 percent of survey respondents except in Duluth, the northernmost city surveyed, where a warm day was cause for preferring the outdoors (Robertson, 1993 and 1994). It is not clear why the Minneapolis and St. Paul Skyway preference percentages obtained in this survey substantially exceed the Skyway usage percentages measured on the basis of actual counts in 1975. The differences may relate to growth of those cities’ systems over the inter- vening decade, to biases inherent in preference surveys, to the fact that only summertime pedes- trians on skywalks and not users of sidewalks were interviewed in the preference survey, or some combination of these factors. Not all pedestrians prefer to use skywalks. Street-level entrances may be far inside buildings and thus inconvenient for short travel segments. Also, and not just in tunnel systems, there may be a disorient- ing lack of visual cues as to the user’s location. Landmarks are not as visible as they might be at street level and the twists and turns of interior corridors can lead to wrong turns. In addition, some skywalk segments may close or become deserted at night and on weekends (Robertson, 1993). Despite their popularity in most applications, there is one known instance where skywalks have been dismantled— the Rosslyn district of Arlington County, Virginia (Fisher, 2005). The Rosslyn system had been built piecemeal by developers as a building approval requirement. It was characterized by relatively nar- row walkways open to the weather, and never quite achieved full “system” status. 16-67 Table 16-9 Percentages of Skywalk Users Preferring Skywalk Over Sidewalk by Outdoor Temperature Temperature (Fahrenheit) Cincinnati, Ohio Des Moines, Ohio Duluth, Minnesota Minneapolis, Minnesota St. Paul, Minnesota 5 Cities Overall Cold day (20 degrees) 100.0 97.1 99.0 96.0 99.0 98.2 Average day (50 degrees) 90.1 84.5 69.0 71.7 90.9 81.3 Warm day (80 degrees) 84.2 83.5 31.0 71.7 86.9 71.5 Notes: Survey conducted in summer of 1985, of skywalk users only, with 502 samples total (99 – 102 respondents per city). See discussion in text above of possible survey biases. Source: Robertson (1993). Urban Planning Considerations. Skywalks are not universally liked among city planners and observers of the cityscape, although 97 percent of skywalkers themselves interviewed in the 1985 five- city skywalk survey agreed that they “thought skywalks added to the visual attractiveness of the down- town.” The concern is not just architectural effect on sightlines and building facades, or the potential to segregate people by social class, discussed further in the “Related Information and Impacts” section under “Economic and Equity Impacts”—“Equity Issues”—“Equity of Access.” The most fundamental issue is whether skywalks draw pedestrians (or too many pedestrians) off of the sidewalks and away from the ground level, leaving lightly populated streetscapes not attractive to retailers and dominated by the automobile (Robertson, 1988 and 1995, Bandara et al., 1994, Peale, 1999). The case of St. Paul garners the most attention in this regard. Disparities in ground floor versus Skyway- level rents and retail activity in downtown St. Paul are covered under “Land Value and Commerce Impacts”—“Downtown Skywalk Impacts” in the aforementioned “Economic and Equity Impacts” subsection. A factor not always considered in using St. Paul as a case example is that many of their

Skyways were installed in conjunction with urban redevelopment, inclusive of the new retail core, and that it was this urban redevelopment that moved much of the retail up to the second level (Heglund, 2004). With reference to the street level of key urban redevelopment components, St. Paul has even been criticized as “the blank-wall capitol (sic) of the United States” (Roberts, 2001). It appears that a substan- tial portion of the activity transfer that has caused 3/4 of the downtown retail to be on the second level in St. Paul is not directly attributable to any inherent characteristic of skywalk outcomes but rather to deliberate 1960s/70s city-planning and redevelopment-project-design decisions. Bicycle Lanes and Routes Bicycle lanes provide designated travel ways on roads for preferential or exclusive use by bicycles. They are created through the use of pavement markings and traffic signs and, in the case of cycle tracks, physical delineations (AASHTO, 1999, NACTO, 2011). Many researchers have concluded that bicycle lanes are advantageous, compared to streets with no bicycle space delineation, in that they make bicyclists and motorists more predictable and comfortable with each other’s presence (RTC and APBP, 1998). However, there have been some bicycle advocates who have not supported bicy- cle lane development, particularly where bike lane use is mandatory when present (MacLachlan and Badgett, 1995). There are a number of subcategories of bicycle lanes, as described in the “Overview and Summary” under “Types of Pedestrian and Bicycle Improvements/Programs,” but travel demand response of trip makers to bicycle lanes tends not to be differentiated at that level of detail. Shared-roadway bicycle provisions other than bicycle lanes and tracks do not incorporate lane-line or separator designation of road space for bicyclists but do generally feature signage and other con- siderations. Included are wide curb lanes, bicycle boulevards, and other signed bike routes. This subsection first provides a review of findings concerning bicyclist preferences and travel behav- ior with regard to on-street bicycle facilities, set in a context of comparisons to undifferentiated streets and also multi-use, off-road paths. This is followed by examination of actual changes in volumes and travel choices of bicyclists in direct response to bicycle lane implementation, both individually and as systems of bicycle lanes. Such information, being limited, is supplemented with research on the effects of bicycle facility system extent on overall cycling levels. The comparative and systems stud- ies overlap substantially with research on impacts of off-road bicycle paths, a subject covered further in the subsection to follow on “Shared Use, Off-Road Paths and Trails.” Finally, information specific to cycle tracks is offered along with findings about shared-roadway bicycle route applications includ- ing wide curb lanes, bicycle boulevards, and ordinary signed bike routes. Popularity, Preferences, and Route Choice Research on preferences and route choices offers a sound basis for concluding that most adult bicy- clists prefer bicycle lanes relative to use of undifferentiated streets if vehicular traffic volumes are moderate to high. The picture is less clear when it comes to understanding preferences for bicycle lane use as compared to the alternatives of off-road paths or bicycle routes with or without special provisions. Seemingly conflicting findings are common. Revealed preference research from Portland, Oregon, utilizing global positioning system (GPS) and computer network analysis tech- nologies, is providing added evidence of preference for off-road facilities and even bicycle boule- vards over bicycle lanes where there is a reasonably direct option for any given bicycle trip. Contradictory findings for lanes versus paths arise in part from different reactions and needs of differing bicycling populations. Bicyclists run the gamut from highly experienced bicycle com- 16-68

muters on the one hand to inexperienced recreational cyclists on the other. There is also the com- plexity introduced by need to get between specific points when bicycling for utilitarian purposes. Among facilities that are physically and operationally attractive, whichever type provides the most direct routing in any given circumstance is the most likely to be used for utilitarian travel. Bicycle lanes are often more direct because they can and do make use of the street system, while path facil- ities often follow natural features or former railway roadbeds and canals. Such path alignments may or may not offer linkages useful for commuting or other travel aligned to specific destinations. Bicycle boulevards are only beginning to be addressed in quantitative preference or route choice analysis, and little has been encountered that covers other special provisions or signing of bicycle routes. Similarly, information on the bicycling preferences of children and their guardians is very thin. Opinion Surveys and Observational Studies. Among opinion surveys is a case where respon- dents expressing interest in cycling were asked to allocate 100 points among different facility improvements to indicate their effectiveness in encouraging bicycle commuting. The importance assigned to safe bike lanes was (on average) more than three times higher than any other type of improvement among the choices offered (MacLachlan and Badgett, 1995). A 1991 Bicycling Magazine poll found 49 percent of active bicycle riders and 20 percent of all adults felt that “safe bike lanes” would encourage them to ride a bicycle to work (Goldsmith, 1992). Such surveys are influenced by question structure and wording, and only indicate what respondents might do, not what they actually will do (Dill and Carr, 2003). The popularity of bike lanes encountered in preference surveys could possibly be attributed as much to respondents’ experience as motorists as to their experience as cyclists. Some motorists like it that bicycle lanes take cyclists “out of the way” of the motor vehicle and vice versa. A Florida study confirms this function of bike lanes. The study relied on more than 1,500 observations of passing-vehicle interactions between cyclists and motorists, between intersections only, on both bike lanes and wide curb lanes. The lateral separation between motorists and bicyclists, lateral position of bicyclists, and motor vehicle encroachments when passing bicyclists were all examined. There were not huge variations between the two facility types, but bike lanes resulted in the smallest movements on the part of motorists and the least spatial separation between bicyclists and motorists. Bike lanes seemed to give motorists greater confidence about the likely movements of cyclists, encouraging them to accept smaller separations, while cyclists seemed to be less timid about road position. On average, the separation of motorists from bicyclists was 5.9 feet for bicycle lanes versus 6.4 feet for wide curb lanes. Motorists moved to the left an average of 1.0 feet for facilities with bike lanes versus 2.4 feet for passing bicyclists on facilities with wide curb lanes. Cyclists did not feel the need to ride as close to the edge of the road on facilities with bike lanes (riding 2.6 feet from the edge) as on wide curb lane facilities (1.4 feet from the edge). Only 8.9 percent of motorists passing cyclists shifted into the left lane on facilities with bike lanes as compared to 22.3 percent with wide curb lanes. The study did not report on crash rate differentials (Harkey, Stewart, and Rodgman, 1996). Stated Preference Experiments. While stated preference surveys have tended to indicate cyclists have an increased comfort level on bicycle lane facilities (Hunter et al., 1999), they have not resolved discussions of which is the better solution—bicycle lanes or off-road facilities. A more recent pub- lished review of both stated and revealed preference research found “results [that] seem somewhat mixed” on the subject of bike lane versus off-road facility preference. Among five stated preference experiments examined as part of the review, two specifically identified travel time as being of utmost importance. One-third found safety to be of top priority, and posited that safety improvements were more important than travel time reductions for encouraging bicycling. Among the four studies with 16-69

reported results for facility type preferences, one found a preference for bicycling on residential streets and an aversion to cycling alongside parked cars. It also estimated that either bicycle lanes or off-road facilities added value, with the greater added value for bicycle lanes. Another identified a trip routing preference for off-road facilities, along with low-traffic residential streets. One of the studies found surface quality to be of more importance than type of facility or traffic volumes. A study that looked only at bicycle lanes versus wide shoulders, in the context of transit access, found bicycle lanes to have the greater positive influence on access mode choice—more strongly so in the case of inexperienced cyclists (Tilahun, Levinson, and Krizek, 2007). The same researchers conducted their own stated preference experiment with employees (faculty and students excluded) at the University of Minnesota. Summer and winter facility conditions at selected St. Paul locations, shown in video clips, were presented in separate summer/winter ses- sions. Paired comparisons were employed, and a travel time was associated with each option. Participants were asked to choose their preferred route in the context of commuting to work. In an iterative process, the maximum added travel time each participant would tolerate to use his or her preferred facility type, within paired comparisons, was determined. Using combined summer and winter results, the estimated marginal utility of an off-road bicycle facility relative to having a bicycle lane with no parking alongside was small, while the marginal util- ity of a bicycle lane relative to having no lane was large. The estimated marginal utility of not hav- ing parking alongside (on either a bicycle lane or on a street with no bicycle lane) was intermediate in value. That said, the estimated order of participants’ facility preference was: (1) off-road bicycle facility (most preferable); (2) bicycle lane with no parking; (3) bicycle lane with parking; (4) street without bicycle lane, no parking; and (5) street without bicycle lane, with parking (Tilahun, Levinson, and Krizek, 2005 and 2007). Findings were found to be independent of regularity of actual bicycling to work. A greater willingness to accept longer travel times to travel on preferred facility types was exhibited, however, by female and older participants (Tilahun, Levinson, and Krizek, 2005). Retired persons and children were, obviously, not included in this work-trip-based experiment. GPS- and Network-Based Revealed Preference Research. The previously alluded to GPS- and network-based research in Portland, Oregon, provides revealed preference information in the form of actual routes taken in comparison to minimum time paths through a bicycle network. GPS tech- nology was employed to track bicycle trips made during one week by a sample of 164 adults in the region, primarily within the city limits. Volunteers were obtained, not working through groups of avid cyclists, but instead using more general appeals. An extra effort, somewhat but not entirely successful, was made to include infrequent cyclists—deemed to be a surrogate for less skilled cyclists. Quota sampling was used to obtain roughly equal representation for men and women. Data collection took place from March through November, 2007. After processing the bicycle trip data, and determining trip origins and destinations, minimum- distance-path traces were determined for the same origin-destination pairs utilizing standard transportation-planning network algorithms. This, together with further network processing, allowed analysis of deviations from minimum-distance paths in terms of bicycle facility types uti- lized. Recreational and exercise bicycle trips were omitted from the minimum-path tracing and comparative analyses, as were transit access trips, but all types of bicycle-mode-only utilitarian trips—not just commute trips—were included (Dill and Gliebe, 2008). Cycle tracks were not included in the research, for lack of such facilities in Portland at the time of data gathering. The route choice implications of this research are explored in the “Underlying Traveler Response Factors” section under “Trip Factors”—“Bicycle Trip Distance, Time, and Route Characteristics”— “Bicycle Route Choice.” (See Table 16-67 and subsequent discussion. Table 16-67 compares the bicycle 16-70

mileage actually accumulated on different types of facilities compared to the mileage that would be accumulated if all cyclists making utilitarian trips followed a minimum distance path. It shows the surveyed adult bicyclists overall rode 4 percentage points more miles on bike lanes than minimum- distance routings would suggest, 6 percentage points more miles on bicycle boulevards, and 8 percent- age points more miles on off-road trails. It also shows 17 percentage points fewer miles were ridden on busy and moderate traffic streets without bike lanes than minimum-distance routings would predict.) Subject to the limitation that only the Portland, Oregon, urban area was included in the research, this indicates a hierarchy wherein bicycle lanes are preferred over all categories of undifferentiated streets except maybe quiet streets, bicycle boulevards are preferred over bike lanes, and off-road trails are preferred over bicycle boulevards. (A hierarchy derived such as this one assumes ideal comparabil- ity, in other words, equivalent connectivity and directness between origin and destination for each alternative facility type.) Although the strength of these relationships varied among subgroups of adult bicyclists, only infrequent cyclists showed a negative response to any type of bicycle facility. Bike lanes held a slight negative attraction for the infrequent cyclists, though not nearly as negative as riding on non-quiet streets without bike lanes (Dill and Gliebe, 2008). The minimum-distance and actual route data for bicyclists from this study were subsequently used to develop an explanatory model to mathematically describe cyclist preferences (Broach, Gliebe, and Dill, 2009a). Using this model, the analysts predicted how far out of his or her way an average adult cyclist would go in order to make use of various types of bicycle facilities for the full trip, or in the case of major bridges, a full bridge crossing. It was estimated, for example, that the average cyclist will be willing to bicycle 31 percent farther to avoid a moderate traffic street without a bike lane and be able to use a bike lane instead. This result and other estimates using the explanatory model are shown in Table 16-10 (Broach, Gliebe, and Dill, 2009b). The same basic hierarchy of pref- erences can be seen here as discussed above and illustrated later (see Table 16-67).16 16-71 Table 16-10 Estimated Percent Out-of-the-way a Cyclist Would Go to Avoid a Street or Bridge Without a Bicycle Lane To Avoid A… (Without Bike Lane) And Use, for the Entire Trip or Bridge Crossing,… A Bicycle Lane A Bicycle Boulevard An Off-road Trail Quiet Street 0% 14% 26% Moderate Traffic Street 31% 45% 57% Highway Bridge 19.5% n/a 34% Source: Broach, Gliebe, and Dill (2009b). 16 All numerical values based on the Portland bicyclist route choice modeling that are presented here in Chapter 16 derive from the initial research model of 2009, which encompasses all utilitarian trip purposes in a single model. Subsequently, the model has been refined for use in regional forecasting, including stratification into work and non-work trip purpose components. The purpose-stratified models continue to exhibit the same on- street hierarchy with off-road trails ranking highest in preference, bicycle boulevards next, followed by quiet streets along with bicycle lanes on streets of any volume, and lastly by busy streets without lanes (Annual Average Daily Traffic {AADT} of 10,000 vehicles or greater), which rank progressively lower with higher traf- fic volumes. The on-street hierarchy holds for both commute and non-commute travel purposes, with the higher sensitivity to facility type found with non-commute trips, as might be expected. Bridge bicycle facility types have been subdivided into bike lane and separate bike facility categories, both exhibiting substantial pref- erence, especially the one bridge with a separate bike facility (Broach, Gliebe, and Dill, 2011).

Bicycle Lane and Route User Makeup. The user makeup of bicycle lanes, as compared to other types of facilities, may possibly be tilted toward use by adults commuting to work. A weekday 7 to 9 AM survey of bicycle lane users in the Seattle CBD found 97 percent of survey respondents were making a utilitarian trip. Of these, 92 percent considered themselves to be regular commuters. By comparison, an equivalent peak period survey on the Burke-Gilman off-road trail north of the University of Washington (UW) found only 56 percent to be making a utilitarian trip, with some 86 percent of the utilitarian tripmakers considering themselves to be bicyclists who commuted reg- ularly (Niemeier, Rutherford, and Ishimaru, 1995b). Of course, the CBD location of the bicycle lane survey would tend to give more emphasis to commute travel even with the comparable timing. Bureau of Transportation Statistics (BTS) 2002 data indicate that only 5 percent of persons who bicycled in the previous month did so as part of a work or school commute. (In the “Overview and Summary” section, see “Analytical Considerations”—“National and Regional Non-Motorized Transportation (NMT) Data”—“Most-Recent Trip Versus Trip-Day Travel Data” for a discussion of the limitations of using “most recent trip” information like this.) Of those who were commut- ing, 11.0 percent reported primarily using bike lanes. The comparable bike lane usage figure for recreational bicycle trips was 5.6 percent (Dill and Carr, 2003). Whether this differential is the result of preference or of facility orientation is not known. A comparison of parallel-facility user characteristics is provided in the case study “Special Mini- Studies in Montgomery County, Maryland” under “More . . .”—“Off-Street Versus On-Street NMT User Mix.” It focuses on weekend and off-peak use of an on-parkway bike route (not a bike lane) versus a proximate off-road trail, both in the same parkland. Within these limitations, markedly different usage patterns are shown for the two types of facilities. Although total facility NMT vol- umes were similar, the on-parkway bike route attracted many more bicyclists in cycling gear, fewer females, and virtually no children, to say nothing of the fact that all walkers and joggers (save one) used the trail. Numerical comparisons are given in the case study and equity implications are explored in the “Related Information and Impacts” section under “Economic and Equity Impact”—“Equity Issues”—“Equity of Access.” The case study, taken in conjunction with other evidence, is certainly suggestive of the proposition that many investigations of facility preference have been too aggregate. Examination of the preferences and needs of distinct user groups is begin- ning to show promise for identifying different bicycling patterns and reducing occurrence of con- flicting findings, thereby providing a basis for better facility planning. Bicycle Lane Implementation Before-and-After Counts and Surveys. Most before-and-after evaluation studies report increased bicycle volumes on streets where bike lanes have been introduced. In those that also examine off- facility data, however, it becomes apparent that a portion of the demand attracted to bicycle lanes is simply shifted from presumably less desirable routes. Count-based studies tend to leave as an open question the extent to which introducing bike lanes will result in higher total bicycling demand. Two of the locations with comprehensive before-and-after evaluations, Davis, California, in the United States, and Toronto, Ontario, in Canada, are featured in the “Case Studies” section of this chapter. In each of these examples, the introduction of bike lanes on a single street or multiple streets resulted in increased cycling along those streets, but a substantial portion of the increase was found attributable to shifts in route choice rather than changes in the prevalence of bicycling (Lott, Tardiff, and Lott, 1979, Macbeth, 1999). Table 16-11 summarizes before-and-after bicycle survey or count data for implementation of bicy- cle lanes. Although some of the individual counts cover as little as 1 hour before implementation, 16-72

16-73 Table 16-11 Summary of Before and After Studies of Individual and System Bicycle Lane Provision Examples Study (Date) Process (Limitations) Key Findings 1. Lott, Tardiff, and Lott (1979) (see also case study “Anderson Road Bicycle Lanes — Davis, California”) Bicycle lanes were introduced on Anderson Road in Davis, CA, after a basic bicycle lane grid was already established. Interviews, including 108 “after” with retrospective ques- tions, covered 5 blocks each side of Anderson. Peak period counts were made before and after. (Differing findings from interviews vs. counts.) Of 57 interviewed cyclists using other streets in the before condition, 44% had changed route to Anderson Road. The 1-hour AM and 2-hour PM counts showed overall bicycling growth on Anderson of 7%, compared to growth on 2 parallel roads with bike lanes of 9% to 12%. Overall cycling growth was ascribed to season and school calendar. 2. Barnes, Thomp- son, and Krizek (2006) (see this subsec- tion including Table 16-12) In Minneapolis-St. Paul 3 major bike lane facilities and 4 major off-road trails were opened 1990-2000. Com- mute trip bike mode share changes were computed for TAZs within 1 mile (1.5 miles for facility termini). (Work purpose trips only.) Average bike mode shares inside the commutersheds were 4 to 5 times the shares in the rest of the Central Cities to start with. The trail and lane commutershed bike shares increased by averages of 1.38 percentage points each, up 64% in the case of bike lanes. a 3. Cleaveland and Douma (2009) (see this subsec- tion including Table 16-13) Commute trip bike mode share changes were computed for Census block groups within 1.55 miles of bike facilities opened 1990-2000 in 6 additional U.S. cities. (Work purpose trips only, on-street facility types not reported for all cities.) In the city of Chicago, the implemented on-street facilities were bike lanes. There was also promotion and a major bike rack installation program. Bike lane commutershed bike shares increased by an average of 0.32 percentage points, up 91%. a 4. Macbeth (1999) (see case study “Bicycle Lanes in the Downtown Area — Toronto”) Between 1993 and 1997 bicycle lanes were installed on 6 streets in the central area of Toronto. Before and after bicycle and vehicle counts were performed. (Diversion of cyclists to bike lanes was not fully explored.) Bicycle volume increases averaged 23% on the 6 streets where bicycle lanes were installed, while citywide cycling remained static or possibly declined. Declines were most noticeable on streets without bicycle lanes. 5. Fertig (1996) In the city of Santa Barbara, CA, bicycle counts were made at 62 loca- tions in 1973 and 1996. None had bicycle lanes in 1976 and 23 had bike lanes in 1996. The 12-hour 1973 counts were adjusted to 2-hour PM counts to allow comparison with 1996. (No adjustments for secular trends other than population.) Total cyclists counted at 62 street locations increased by 48% on average over the 23-year period without popu- lation growth adjustment or 19% with adjustment. Adjusted growth at loca- tions with bike lanes in 1996 was 46% to 47% (with or without conversion to one-way street traffic) vs. a 1% decline where bike lanes were not installed. 6. Chaney (2005) A 2-mile stretch of Valencia Street in San Francisco was restriped from 4 through lanes to 2 through lanes, 1 turn lane, and 2 bike lanes. (One 1 hour PM peak count before/after.) Valencia Street bicycle usage increased from 88 to 215 (up 144%) in the 1 hour PM peak. No investigation of possible bicyclist diversion to Valencia. There was parking on both sides before/after. 7. Chaney (2005) Part of 10 blocks of Polk Street in S.F. was restriped from 3 through lanes (2 SB, 1 NB) to 2 vehicle lanes, 2 bike lanes, and 2-sides parking, while the other (narrower) part became 2 wide lanes (2-hour AM, PM count locations not reported.) Polk Street bicycle usage (average of several before/after counts) rose from 37 to 52 (up 41%) in the 2-hour AM peak, while 2-hour PM peak usage increased from 43 to 55 (up 28%), apparently in the wide-lanes section. No diversion investigation. (continued on next page)

16-74 Table 16-11 (Continued) Study (Date) Process (Limitations) Key Findings 9. Chaney (2005) Before/after provision of bike lanes on Oriental Blvd. in Brooklyn, 11-hr. weekday and 8-hr. weekend counts were made in May, July, and Sept. (No diversion investigation.) The average weekday bicycle, skater, and scooter totals were 68 before and 103 after 1 year, up 52%. Correspond- ing weekend average totals were 61 before and 65 after, up 7%. 10. Chaney (2005) Non-standard 3-foot bike lanes were implemented along a 2-3 mile beachfront state highway in Fort Lauderdale, FL. Before and after counts, 4 15-min. Saturday afternoon counts each, were obtained in February and May. (Fewer tourists in May.) “Before” 1-hour 2-way totals were 344 pedestrians, 39 bikes in street, 29 bikes on sidewalk, 68 total bikes. “After” totals were 206 pedestrians, 28 bikes in lane/street, 23 bikes on sidewalk, 51 bicycles total. Ratio of bicyclists to pedestrians increased from 1:5 to 1:4. a 11. Davies (2007) (see also “Related Info…” – “Time to Establish… Use”) One hour AM peak bicycle counts made on St. Kilda Rd. in Melbourne, Australia, starting 1 year before bicycle lane implementation. (No details about the facility or context.) Bicyclist count grew from 42 the year before opening to 76 the year after, up 81%. Reached 160 after 5 years (almost 4 times the before count) and 511 after 10 years, with 1/5 “before” injury rate. 12. Boarnet et al. (2005b) (see also “NMT Policies and Pro- grams” – “School- child-Focused…”) Of 10 CA schools surveyed to ascer- tain 2002-03 SRTS impacts, one had received a bike lane in addition to improved sidewalks. Before/after 2-day counts were made of child cyclists. (Volumes deemed too small for inferences to be drawn.) Child bicyclist volumes before and after the installation of on-street bicycle lanes were 4 and 14 cyclists, respec- tively, during regular school access/- egress hours. The authors concluded “that there was little observed impact on bicycling.” Notes: Where substantial additional information on individual studies is provided in text and tables or figures, this is noted — and the location within the chapter is given — in the first column. a Percentage increase(s) or ratios and some totals calculated by the Handbook authors. Sources : As indicated in the first column. 8. Chaney (2005) A 3-block 1-way section of Fell Street in S.F. was restriped from 3 through lanes and 1 tow-away lane to 3 vehicle lanes, 1 bike lane, and parking on both sides (2-hour PM peak counts, only 1 “after” count.) Fell Street bicycle usage in the 2-hour PM peak rose from 25 on south side traffic lanes, 9 on north side lanes, and 37 on sidewalks (71 total) to 82 in south side bike lane, 5 in traffic lanes, and 7 on sidewalks (94 total, up 32%). a and 1 hour after, the 12 quantitative studies present a nearly-consistent pattern of apparent bicy- cle usage or count growth from the “before” to the “after” condition. The first three table entries cover studies that either made use of project-specific survey results or utilized decennial U.S. Census journey-to-work data. Excluding the Davis, California, case where results are somewhat ambiguous (1st entry), the 2 investigations that examined travel mode shifts in response to bicycle lanes are those in Minneapolis-St. Paul and Chicago (2nd and 3rd table entries). These studies found average increases of 64 percent and 91 percent, respectively, in work commute travel bicycle share with the introduction of bicycle lanes. The response in Chicago was almost certainly amplified by publicity and bicycle parking enhancements (Barnes, Thompson, and Krizek, 2006, Cleaveland and Douma, 2009). Aside from the influence of concurrent actions in Chicago, the mode shift findings may be viewed as particularly robust, because they are not inflated by effects of route choice shifts.

The 4th through 9th Table 16-11 entries, if one combines the three San Francisco studies into one data point, offer straightforward combined count-based examples from four North American cities (see Table 16-11 for sources). (Deliberately held aside in this array of cities are the Fort Lauderdale, Florida, example with counts in different seasons relative to tourist activity, the California Safe Routes to School (SRTS) results for children, and the overseas example from Melbourne, Australia.) The average bicycle count increase per city among these four, on streets with bike lanes added, ranges from 23 percent in downtown Toronto to 70 percent in San Francisco. The simple average across the four cities is a 48 percent increase in bicyclists on affected streets. The Melbourne St. Kilda Road bike lane results (11th table entry), comparing year-before and year- after counts, show a somewhat higher increase at 81 percent. What is particularly notable in the case of Melbourne is the continued strong growth for a number of years after implementation (Davies, 2007). In addition to the 5- and 10-year results in Table 16-11, a 14-year record of St. Kilda Road bicycle volumes and injury crashes is provided in the “Time to Establish Facility Use” dis- cussion within the “Related Information and Impacts” section (see Table 16-114). A key weakness in most of the count-based studies, already alluded to, is the lack of information on what travel changes actually make up the increases in cycling on streets where bicycle lanes have been installed. The added bicycles represent an unknown combination of diversions from parallel routes (route shifts), trips previously made by other means (mode shifts), and even possi- bly some trips diverted to new destinations (destination choice shifts) and trips not previously made (induced travel), all manifestations that may occur when a travel route is improved. The count-based studies also lack information on the purposes of the bicycle travel before and after. Three of the studies do provide some information on route diversion/shifting. The Anderson Road research in Davis (1st entry in Table 16-11) found 57 cyclists, among those interviewed between Anderson Road and the parallel previously existing bike routes, who reported use of routes other than Anderson Road in the before condition. Among these cyclists, 44 percent had shifted to Anderson Road after implementation of the new bicycle lanes (Lott, Tardiff, and Lott, 1979). Note that this is different than making a statement about the proportion of Anderson Road bike lane users who had diverted from other routes, a value that could not be meaningfully computed with the Davis survey findings obtained. The downtown Toronto bicycle lane study (4th table entry) did not quantify route diversion effects, but found the 23 percent average bicycle count increase—on streets where bike lanes had been installed—in a context of citywide lack of change, or possibly decline, in bicycle usage. This out- come strongly implies shifting of pre-existing bicycle trips from streets without bike lanes to streets where they were installed. Anecdotal evidence of pronounced cycling declines on unmodified streets adds support for the implication (Macbeth, 1999). The before-and-after bicycle count analysis in Santa Barbara (5th entry in Table 16-11) provides similar but more explicit evidence of route shifting. With a 46 percent adjusted average growth in bicycling on streets with bike lanes installed during the 23-year analysis period, paired with a 1 percent decline on streets with no such lanes (Fertig, 1996), bicyclist route shifting is clearly demonstrated. Nevertheless, it is obvious that an overall increase in bicycling per capita also took place. What cannot be determined, in the absence of full screenline counts or equivalent, is exactly what the overall growth was or what proportions of bicycle count growth are attributable to route shifts versus other responses such as mode shifts. Longitudinal Commute Mode Share Research. The 2nd entry in Table 16-11 encapsulates the first of two before-and-after studies found to have information directly bearing on whether or not 16-75

mode shifts to bicycle riding are brought about by bicycle lane introduction. In this research, 1990 and 2000 bicycle mode shares were obtained from U.S. Census journey-to-work data for the traf- fic analysis zones (TAZs) within the commutershed of three bicycle lanes and four off-road trails opened during the decade within the city limits of Minneapolis and St. Paul. The analysts experi- mented with alternative buffer zone and trip definitions for delineation of the commutershed. During this experimentation, it was discovered that many of the trips most affected were longer than the 5-mile limit initially imposed. Accordingly, the commutershed definition used for the final results covered all work purpose trips over 1 mile in length, generated within 1 mile of the facility or within 1.5 miles of the ends of the facility, but with inter-facility trips allowed. Table 16-12 gives the facility mileage and bicycle share results for the three bicycle lanes and four off-road trails (Barnes, Thompson, and Krizek, 2006). 16-76 Table 16-12 Before and After Commutershed Work Trip Bicycle Mode Shares for Three Bicycle Lane and Four Off-road Trail Provision Examples in Minneapolis-St. Paul Bicycle Facility Facility Mileage 1990 Bike Share 2000 Bike Share Percentage Point Change Percent Increase Park/Portland Bike Lanes 4.0/4.2 3.49% 4.54% 1.05% 29.9% Summit Ave. Bike Lane 4.6 1.00% 2.36% 1.36% 135.0% University/4th Bike Lanes 1.6/0.8 6.10% 7.82% 1.72% 28.2% Cedar Lake Trail a 7.8 2.50% 3.55% 1.05% 41.9% Kenilworth Trail a 1.8 1.73% 3.04% 1.31% 76.0% West River Parkway 8.0 5.48% 7.18% 1.70% 30.9% U of MN Transitway a 1.9 6.37% 7.83% 1.46% 23.0% Center Cities - All Work Trips n/a 1.15% 1.39% 0.23% 20.2% Notes: All facilities listed were implemented during the 1990-2000 period. Trips under 1 mile in length excluded, except in “Center Cities - All Work Trips” row. a Frequency of intermediate access points limited by topography or built environment. Source: Barnes, Thompson, and Krizek (2006), with elaboration by the Handbook authors. One circumstance that immediately stands out is that the 1990 “before” mode shares in the cor- ridors slated for bicycle lanes and for off-road trails are substantially higher than for the Minneapolis-St. Paul Center Cities as a whole. This remains true even if commutershed trips of less than 1 mile are included for better comparability, although the differential is reduced. This find- ing will be referred back to in the discussion of causality in the “Bicycle Lane System Coverage” discussion to follow. The Minneapolis-St. Paul research not only provides bike lane results, but also allows a compari- son of the effect on commute trip bicycle mode shares of the three on-street bicycle lanes relative to the four shared use, off-road trails. As can be seen from Table 16-12, the increases in commuter- shed bicycle mode share ranged from 1.05 to 1.72 percentage points for the three bike lanes and from 1.05 to 1.70 percentage points for the four trails. The simple average bicycle commute mode share gain was 1.38 percentage points for both the three bike lanes and the four trails. Because the starting shares in the bicycle lane corridors tended to be lower, these gains translate to a 64 per-

cent average increase for the bicycle lane commutersheds and a 43 percent increase for the off-road trail commutersheds. The absolute mode share gains in commuter bicycling were, however, essen- tially identical. The results may also be compared to bicycle mode share growth in areas outside the facility com- mutersheds. In St. Paul, excluding trips under 1 mile in length, the 1990–2000 secular growth amounted to only 0.22 percentage points of commute trip bicycle mode share, a 50 percent increase over the low 0.453 percent 1990 share. In Minneapolis, the comparable statistics are 0.23 percent- age points of commute trip bicycle mode share gain, a 24 percent increase over the 0.942 percent 1990 share (Barnes, Thompson, and Krizek, 2006). Subsequent research applied the same general study approach in six additional U.S. cities and regions. Commutersheds were, however, defined as extending 2.5 kilometers (1.55 miles) from the various bicycle facilities studied and there may have been other analytical differences. Table 16-13 tabulates the findings for all types of facilities studied, by city/region. On-street facilities were ana- lyzed in four of the cities, but only in Chicago were they explicitly identified as being bicycle lanes. 16-77 Table 16-13 Before and After Commutershed Work Trip Bicycle Mode Shares for Various Bikeway Types in Six Additional U.S. Cities and Regions City/Region and Bicycle Facility Type Statistical Significance 1990 Bike Share 2000 Bike Share Percentage Point Change Percent Change Austin – signed routes Yes 0.87% 1.19% +0.32% +36.8% Austin – off-road trails Yes 2.64% 3.52% +0.88% +23.9% City of Austin, TX, overall Yes 0.76% 0.95% +0.19% +25.0% Chicago – bike lanes Yes 0.35% 0.67% +0.32% +91.4% City of Chicago, IL, overall Yes 0.28% 0.50% +0.22% +78.6% Colorado Springs – off-road paths No 0.72% 0.76% +0.04% +5.6% City of Colo. Springs, CO, overall No 0.49% 0.55% +0.06% +12.2% Madison – on-street bikeway No 1.30% 1.62% +0.32% +24.6% Madison – off-road paths No 5.83% 5.70% -0.13% -2.2% City of Madison, WI, overall No 3.40% 3.28% -0.12% -3.5% Salt Lake City – on-street bikeways No 1.54% 1.53% -0.01% -0.6% Salt Lake City – off-road paths No 1.67% 1.27% -0.40% -24.0% City of Salt Lake City, UT, overall No 1.52% 1.49% -0.03% -2.0% Orlando area – off-road trails No 0.77% 0.61% -0.16% -20.8% Orange County, FL, overall Yes 0.66% 0.46% -0.20% -30.3% Notes: All bikeways studied were implemented during the 1990-2000 period, but not all bikeway segments implemented during the period were deemed relevant for inclusion in the study. Source: Cleaveland and Douma (2009), with elaboration by the Handbook authors. The Chicago bicycle lanes were implemented during a period of increased bicycling advocacy and awareness campaigns. Implementation was also more-or-less concurrent with a major bicycle rack installation program (see 6th entry, Table 16-36, in the “Point-of-Destination Facilities” subsection under “Bicycle Parking and Changing Facilities”). The bike rack program and promotional campaign

effects are impossible to disentangle from the bike lanes mode share effects. It may be noted, how- ever, that the percentage point bicycle mode share gains were 45 percent greater within the defined bicycle lane commutersheds than for the city of Chicago as a whole (Cleaveland and Douma, 2009). While the bicycle work commute share percentage point gains were more modest along Chicago bicycle lanes than in Minneapolis, averaging 0.32 percentage points, the percentage increase was higher on average and the absolute numbers of cyclists involved were presumably appreciable given that radial routes to Chicago’s CBD were involved. The off-road path findings in Table 16-13 are discussed in the “Shared Use, Off-Road Paths and Trails” subsection under “Shared Use Path Implementation”—“Other Path Information,” as are possible area-specific causes of the smaller to negligible work commute mode share impacts found for most types of facility introductions in the smaller and more spread-out urban areas. The on-street facility findings, other than the Chicago bike lanes already discussed, are examined fur- ther under “Bicycle Lane Variations, Bicycle Boulevards, and Other Signed Bicycle Routes.” The researchers in the six-region study take pains to emphasize that the Census-based results say nothing about effects on trips for errand-running or recreation/exercise, or even about student commuting to major universities as are found in Austin and Madison (Cleaveland and Douma, 2009). Neither this nor the Minneapolis-St. Paul research provides information on route shifting or induced bicycle travel. (Route shifting would not directly affect the reported mode share changes.) Substantial route shifting is almost certain to have occurred in combination with such major mode shifts as were identified in most of the cities where on-street facility bicycle mode shift effects were examined. Thus, the average volume of commuter cyclists on the treated streets presumably increased by a larger percentage—likely a substantially larger percentage—than the 25 to 135 per- cent mode share gains identified in all but Salt Lake City. On the other hand, induced travel and destination shifts are probably negligible in the context of work purpose travel, the focus of the research. Additional Information. Other relevant information can be gleaned from several of the studies. In San Francisco, the three examples reported on in Table 16-11 (averaging a 70 percent growth in bicycling along the affected streets) took place in a broader context involving 10.5 miles of new bike lanes. In this larger context, before-and-after analyses showed increases in bicycle counts ranging from 23 percent to 148 percent with an average of 50 percent (Morris, 2001). On the 2-mile Valencia Street corridor (6th entry in Table 16-11), the one-year evaluation that showed a jump in cycling volume from 88 to 215 bikes per hour also identified a slight reduction in vehicular Average Daily Traffic (ADT), from 22,200 to 19,700, concurrent with the 25 percent reduction in number of vehicular traffic lanes. Reported injury crashes in the corridor among road users, including pedes- trians, bicyclists, and auto occupants, decreased by 15 percent (San Francisco Bicycle Coalition, 2001, BikeSummer ‘99, 1999). Availability of both weekday and weekend before-and-after data from Oriental Blvd. in Brooklyn (9th entry in Table 16-11) indicates that the relative weekday impact of bicycle lane implementation was some 7 to 8 times the weekend increase in bicycling, skating, and scooter use. The Saturday-only data from the Fort Lauderdale, Florida, beachfront (10th entry) shows the proportion of NMT vol- umes made up by cyclists using the street (not the sidewalk) as increasing from 9.5 to 10.9 percent with the implementation of bike lanes (Chaney, 2005). This is a rather modest increase, essentially the same order of magnitude as the 7 percent Brooklyn weekend cycling increase. The California SRTS studies (12th entry in Table 16-11) found no statistically significant evidence of an effect on bicycling to school with bicycle lane installation (Boarnet et al., 2005b). In Davis, 16-78

California, after installation of bike lanes on Anderson Road (1st entry in Table 16-11), only seven bicyclists were estimated to be of age 11 and under, while 41 were judged to be between 12 and 17 years of age, out of 1,577 on Anderson during 3 peak-period hours. That is 0.4 percent and 2.6 percent, respectively, a total of 3 percent children and adolescents (Lott, Tardiff, and Lott, 1979). These findings from Brooklyn, Fort Lauderdale, and California could be the beginning of a still very tentative thesis that bicycle lanes offer relatively little attraction for increased cycling at times or by groups likely to be characterized by presence of youngsters and high proportions of bicy- clists with modest skill levels.17 The Fell Street counts in San Francisco (8th entry in Table 16-11) and the counts in Fort Lauderdale (10th entry) provide information on efficacy of bicycle lanes in attracting cyclists off of parallel sidewalks. These findings are examined in the “Sidewalk Use by Bicyclists” discussion at the close of the “Sidewalks and Along-Street Walking” subsection, along with similar information by bicy- clist age category from the Davis research. Not covered in Table 16-11 are Portland or Corvallis, Oregon, both of which have very extensive bicycle lane systems relative to their size. Portland’s dramatic results are addressed further on, in the “Pedestrian/Bicycle Systems and Interconnections” and “NMT Policies and Programs” sub- sections. Corvallis, a community of roughly 50,000 population, early on had the highest bicycle commute mode share in the state of Oregon, at 8 percent. Some credit the fact that over 90 percent of the collector and arterial streets have striped bike lanes (RTC and APBP, 1998). Better established is the finding that crashes in the community involving bicycles dropped from 40 in the year before the lanes (October 1980 through September 1981) to 16 in the year following lane installation (Environmental Working Group et al., 1997). Also not listed in Table 16-11 are results for multiple bike lane installations in Hull, England. Six monitored locations exhibited cycling increases from before to after bike lane introduction rang- ing from no change to 138 percent growth. The average increase was approximately 36 percent. This average lies within the range of North American city averages. Roughly paralleling the Corvallis experience, a 45 percent reduction in bicycle casualties was observed. This reduction was accompanied by an 11 percent decline in pedestrian casualties (Booz Allen Hamilton, 2006). Bicycle Lane System Coverage Table 16-14 summarizes research on the overall effect of bicycle lane coverage on prevalence of bicy- cle riding. The first four study entries are progressively more advanced works done on a national level by examining facility extent and bicycle commute shares in 18 to 90 U.S. cities. Limitations to be kept in mind with respect to these four studies are that all use city-level aggregate data, focus only on adults, and address neither non-work-purpose utilitarian travel nor recreational/exercise activity. Also, the first two use a combined bikeway coverage measure including both on-road bicycle lanes and shared use, off-road paths and trails. The 3rd study began by exploring a com- bined measure but found the strongest relationship for bicycle lanes alone. The 4th study explic- itly demonstrates strong roles for bicycle lanes and for off-road paths. 16-79 17 At least two areas in addition to the California example have installed bicycle lanes as an element of safe- routes-to-school infrastructure (Petal, Mississippi, and Auburn, Washington), but with usage results unre- ported (National Center for Safe Routes to School, 2010).

16-80 Table 16-14 Summary of Research Findings on the Relationships of Bicycle Lane and Other Facility Prevalence with Cycling Activity Study (Date) Process (Limitations) Key Findings 1. Goldsmith (1992) Tabulated and averaged bicycle commute mode share for U.S. cities grouped by ratio of bikeway to arterial street miles. (No statistical tests, work purpose trips only.) Bikeway/arterial ratio of less than 0.035:1 (8 cities) associated with 0.63% bicycle share versus 6.80% for 10 cities with a ratio over 0.035:1 (or 1.96% share omitting the 6 “university towns”). 2. Nelson and Allen (1997) With 16-city data from Goldsmith (1992) (Davis and Palo Alto omit - ted), plus the percentage of college students among residents, conduct - ed a cross-sectional analysis relating facility miles per 100,000 population to bike commute mode share. (City- level aggregation, work trips only.) The derived linear relationship found 0.069% more commuter cycling for each additional bikeway mile per 100,000 population. Fewer rain days/year and higher ratios of college students were also positively related with bike use. Temperature and terrain were weak/ambiguous variables. 3. Dill and Carr (2003) With Census 2000 Supplemental Survey data, plus bike lane and off- road path and other data, undertook a cross-sectional analysis for 42 large cities relating bicycle infrastructure and other measures to commute trip bicycle mode share. New York City had a “dummy variable” (negative) in the final model. (City-level aggregation, work purpose trips only.) Combined bike lane/path measures significantly related to cycle share but bike-lane miles per sq. mile itself was the strongest infrastructure variable. In the 42-city model each additional bike- lane mile per sq. mile was associated with roughly a 1 percentage point gain in commuter cycling share. Rain days and vehicle ownership were negatives; state spending on NMT was a positive. 4. Buehler and Pucher (2011) (see this section for more information) Further expanded on the Nelson and Allen (1997) and Dill and Carr (2003) approaches, using 2006-2008 3-year average American Community Survey (ACS) cycling level data plus bike lane and path supply data for 90 of the 100 largest U.S. cities, collected by others directly from each city. Three forms of regression were used along with alternative dependent variables: bike commuters per 10,000 residents and work trip bicycle mode share . (City-level aggregation, work purpose trips only.) Multiple regression coefficients on bike lanes, and on paths, highly significant. Bike lane coefficients a little different in each model set than bike path coeffi - cients, but differences never statistically significant. Inelastic demand shown, e.g.: 10% more bike lanes = 2.5% more bicycle commuters (per 10,000 population); 10% more bike paths = 2.6% more bike commuters. Days over 90º F and higher bike fatality rates were negatives; western U.S. location, overall denser/older housing/fewer cars, and higher student ratios were positives. 5. Moudon et al. (2005) (see “Ped…cycle Friendly Neigh- borhoods” for more information) Cross-sectional analysis of cycling activity, socio-demographics, attitudes, and perceived plus objectively measured environmental variables in King County, WA. (Evidence of neighborhood “self- selection” in 1/3 of cyclists.) Objectively measured presence of bike lanes was not significantly related to cycling at least once a week, though a perception of combined trail and bike lane presence did have a positive relationship with cycling, as did objectively measured closeness of trails.

All four national research efforts found a strong positive association between bicycle facility cov- erage and bicycle use for commuting to work. The quantitative relationships derived, and other factors found to have an influence, are enumerated in the “Key Findings” column of Table 16-14. The parameters examined and the detailed results obtained vary among the studies, limiting any additional specificity with which overall conclusions can be drawn. Nevertheless, the three stud- ies that developed research models (2nd through 4th table entries) all demonstrate positive but inelastic commuter bicycling demand for additional bicycle facilities (Goldsmith, 1992, Nelson and Allen, 1997, Dill and Carr, 2003, Buehler and Pucher, 2011). The 4th Table 16-14 entry, the national study covering 90 cities and utilizing 2006–2008 journey-to- work travel data, was able to estimate separate, highly significant model coefficients for both bike lanes and paths (miles per 100,000 population in all cases). Six different final research models were developed, including both ordinary least squares regressions and binary logit proportions models, and estimating bicycle commuters per 10,000 population in four cases and bicycle commute trip mode shares in two cases. Adjusted or pseudo R2 values ranged between 0.60 and 0.62, indicating a good fit with the data in each case. With a bicycle fatality rate variable (significant), 10 percent more bike lane miles (per 100,000 residents) were associated with 2.5 percent more bicycle commuters (per 10,000 population) and 10 percent more bike paths were associated with 2.6 percent more bike com- muters. The overall study results indicate an elasticity of +0.25 for the positive association of both bike lanes and off-road paths with bicycle commute levels in U.S. cities. In two of the six models for the 90-city research, the fatality variable was omitted because of con- cerns about causality and fatality-rate approximation. Without the fatality variable, path mileage was estimated to be about 1/3 more important than bike lane mileage for describing bicycle com- muters per 10,000 population. For estimating bicycle commute mode shares, the importance of lanes and paths reversed, with bike lanes exhibiting the higher coefficient. As indicated in Table 16-14, estimated differences in relative importance of bike lane extent and bike path extent were in no case statistically significant (Buehler and Pucher, 2011). The authors reporting on the nationwide city-level cross-sectional analyses all emphasize that while the positive relationship found between bicycle facility coverage and work trip usage levels is strong, it does not prove causality. Additional bicycle facilities may beget more commuting to work by bicycle, or there may have been inherently higher bicycle volumes in some areas to start with, leading to successful agitation for and construction of more bicycle facilities (Nelson and Allen, 1997, Dill and Carr, 2003, Buehler and Pucher, 2011). 16-81 Table 16-14 (Continued) Study (Date) Process (Limitations) Key Findings 6. Ewing et al. – 2004 as summarized by Davison and Lawson (2006) Utilized objectively measured cross- sectional data to model the effect on walk/bike school access of sidewalk characteristics, bike lanes or paved shoulders, accessibility, and density. Student cycle-to-school shares showed a significant negative relationship with estimated bike time to school. Failed to find any relationship between bicycle lane availability and cycling to school. Note: Where substantial additional information on individual studies is provided in text and tables or figures, this is noted — and the location within the chapter is given — in the first column. Goldsmith (1992) and Nelson and Allen (1997) combined bicycle lanes and off-road facilities in their quantitative (observed) measure of facility prevalence. Sources: As indicated in the first column.

Indeed it was observed above, with respect to the 1990–2000 decade in Minneapolis-St. Paul, that the corridors where bicycle lanes and off-road trails were built tended to have substantially higher bicycle commuting shares to start with. Growth in bicycling mode share did occur in parallel with implementation of the new facilities, but it was fractional compared to the preexisting differential in bicycle commuting between the corridors gaining facilities and other areas. The Twin Cities researchers noted this as a demonstration of “the risks inherent in trying to deduce the impact of facilities by trying to compare [. . .] different places” (Barnes, Thompson, and Krizek, 2006). It is reasonable to assume that both effects play a role in the relationships found in the cross- sectional analysis research: better bicycle facility coverage producing heightened levels of bicycle commuting, and higher volumes of bicycling supporting more facility implementation (Nelson and Allen, 1997). Nevertheless, survey data on alternative or prior modes of travel of users of new NMT facilities do show that mode shifts play a role in facility usage (see, for example, “Related Information and Impacts”—“Travel Behavior Shifts”). The King County, Washington, research entered as the 5th study in Table 16-14 provides the only bicycle lane system coverage findings in the table applicable to bicycling for all purposes, includ- ing the work commute, other utilitarian travel, and recreational/exercise activity. In this study, a perception of bike lane and/or trail presence was found to have a positive relationship to actual bicycling activity, but not objectively measured presence of bike lanes. The objective facility pres- ence measure that did have a significant relationship was measured closeness to off-road trails, not bicycle lanes (Moudon et al., 2005). This finding can be regarded as offsetting study findings such as presented in the 3rd entry of Table 16-14, namely, that the strongest relationship is between bicycle lane coverage and (commute) mode share. Indeed, the totality of evidence presented in this subsection suggests that either bicy- cle lanes or off-road paths may attract the most bicycling depending on circumstances. An alter- native interpretation may prove to be, however, that the strength of off-road paths lies more in supporting other trip purposes other than commuter cycling. Bicycling to work is the only compo- nent of cycling activity addressed by much of the bicycle lane research available, limiting conclu- sions at this point. The final entry in Table 16-14 reports results of cross-sectional modeling of school access NMT facility characteristics and mode use. This study failed to find any relationship between bicycle lane availability and cycling to school (Davison and Lawson, 2006). Although this study represents only one data point among the relatively few evaluations of bike lane use by schoolchildren, it does combine with the findings discussed with reference to Table 16-11 to lower expectations of success in employing bike lanes for school access—particularly in the case of elementary and intermedi- ate schools—or for use with any population characterized by low prevalence of cyclists with high experience levels. An additional cross-sectional child-focused study from the same review does not shed much additional light on the issue. It found Australian adolescents to walk and bike more where roads were perceived to be safe, but this logical finding was paired with others less intu- itive, such as the odd finding that boys were more likely to cycle where it was less easy (Davison and Lawson, 2006). Bicycle Lane Variations, Bicycle Boulevards, and Other Signed Bicycle Routes On-street bicycle facilities and provisions, other than conventional bicycle lanes, are addressed in the following discussion. At the high end of the cost and space-requirement spectrum are cycle tracks and buffered bike lanes. At the other end of this spectrum are wide curb lanes, bicycle boule- 16-82

vards, and streets that have been simply signed as bike routes. The available traveler response research on these options is relatively limited, either because they are somewhat new concepts— especially in a U.S. context (cycle tracks and bicycle boulevards)—or because they exist mostly “below the radar” (wide curb lanes and signed bicycle routes). Cycle Tracks. Physical separation, in contrast to only painted traffic lines and colors, is used in constructing on-street cycle tracks. This separation is accomplished with raised traffic separators, bollards, or on-street parking, or by raising the cycle track itself to introduce a grade differential. Buffered bike lanes are included in the Portland, Oregon, analysis introduced here. They employ a buffer strip between bicycles and motor vehicles that is marked with traffic paint (NACTO, 2011). A before-and-after study of cycle tracks in Copenhagen conducted 1,000 interviews and 1,500 counts, and analyzed 8,500 crashes. Usage of bike lanes and cycle tracks in Copenhagen is 95 per- cent bicycles and 5 percent mopeds. Installation of conventional bike lanes was accompanied by a 5 to 7 percent increase in cycle/moped traffic and no change in vehicular traffic volumes on affected streets. Construction of cycle tracks was, in contrast, accompanied by an 18 to 20 percent increase in cycle/moped traffic and a 9 to 10 percent decrease in vehicular traffic on the streets involved. Copenhagen cyclists were found to feel much safer on conventional bike lanes than in mixed traf- fic, and even more secure on the cycle tracks. For example, 11 percent of cyclists felt “very safe” in mixed traffic, 32 percent felt so in bicycle lanes, and 46 percent felt very safe on the cycle tracks (Jensen, Rosenkilde, and Jensen, 2007). Actual safety results could, however, be described as mediocre. Safety and cyclist-interaction conclusions from this and the other cycle track studies cov- ered here are summarized in the “Related Information and Impacts” section under “Safety Information and Comparisons”—“Facility Type Safety Comparisons”—“Cycle Track Versus Other On-Road Cycling Safety.” Montreal has been a major early adopter of cycle tracks in North America, with a longstanding network. A detailed safety study developed comparisons of bicycle usage between the six studied cycle tracks and mostly parallel “reference streets.” The reference streets had no bicycle facilities. Simultaneous 2-hour counts were used for the comparison. The cycle tracks were found to have 2-1/2 times the bicycle traffic of the reference streets. Cycle track 2-hour volumes, time of day not indicated, ranged from 109 to 1,193, averaging 668 bicycles. As covered in the “Cycle Track Versus Other On-Road Cycling Safety” discussion, the average risk of injury for bicyclists on the cycle tracks was found to be 72 percent of the risk per bicyclist cycling in the mixed traffic of the refer- ence streets (Lusk et al., 2011). Portland, Oregon’s 2009 installations of a cycle track and a pair of buffered bike lanes were ana- lyzed too soon after implementation for rigorous bicycle volume analysis. Among survey respon- dents, 70 percent felt the SW Broadway cycle track had made bicycling easier and safer as compared to the prior bike lane configuration. The proportion of bicyclists on Broadway cycling in mixed traffic, rather than on the available bicycle facility, fell from 12 to 2 percent. Surveyed bicyclist reaction to the SW Stark and State Streets buffered bike lanes was similar, with 9 in 10 indicating preference for the buffered lanes as compared to standard lanes. This one-way couplet had not had bicycle lanes previously. Stark/State bicycle counts were up at least 75 percent in the “after” condition (Monsere, McNeil, and Dill, 2011). A Burrard Bridge trial reallocation of roadway and sidewalk space in Vancouver, British Columbia, Canada, adds additional insight given that the resulting bicycle provisions, while unusual, fit the def- inition of cycle tracks. The physical arrangement is described in the “Pedestrian/Bicycle Systems and 16-83

Interconnections” subsection under “River Bridges and Other Linkages”—“Other River Bridges.” Two different analysis approaches indicated that the change from mixed-use sidewalks to segregated cycle-track equivalents increased bicycling by 26 percent (count-based results) to perhaps a doubling of bicycle-use incidence by bridge neighbors (survey results). Importantly, the count-based increase was composed of a 31 percent increase for bicycle crossings by women versus a 23 percent increase for men, suggesting that increased bicycling comfort levels had attracted more cycling by females in particular (City of Vancouver, 2009a). Wide Curb Lanes. Wide curb lanes can be considered a variation on marked bicycle lanes and have been supported by some as an alternative. They do not have lane-line or barrier separation of bicycles from vehicles, but do feature added lateral road width compared to a normal traffic lane. They do not require quite as much street width as adding a standard bike lane. The extra width enables more comfortable passing of bicyclists by motorists. Bike routes on roads with wide curb lanes are sometimes designated using signs and/or chevron pavement markings. Earlier, in Table 16-11, the 6th entry provided one example wherein a 28 to 41 percent bicycle count growth was observed with a San Francisco project (Polk St.) that involved bike lanes for part of the dis- tance and wide curb lanes for the remainder (Chaney, 2005). It has been noted in opinion survey findings that people express a preference for marked bike lanes over wide curb lanes. A comparative analysis of bicycle lanes versus wide curb lanes concluded that either facility was acceptable, but recommended that where adequate road space was avail- able regular marked bicycle lanes be used, given their apparent popularity (Hunter et al., 1999). The appropriateness of this conclusion is further supported by the bicyclist and motorist position- ing studies presented earlier, at the start of the “Popularity, Preferences, and Route Choice” discussion. Bicycle Boulevards. An approach introduced in some cities to providing on-road bikeways is “bicycle boulevards.” Bicycle boulevards are a shared-roadway strategy applied on low-volume, low-speed streets enhanced for cycling with preferential traffic calming, intersection crossing assists, pathfinder signing, and other treatments to provide a “bicycle arterial” that is mostly stop- free (Alta Planning + Design, 2009a, Ciccarelli, 2010). Vehicles and bicycles are, for the most part, not physically separated. Streets used may be local streets with a history of low vehicular volumes and speeds, streets deliberately traffic calmed, or both. For example, Berkeley’s grid system of seven bicycle boulevards evolved from a 1969 traffic calming plan and system. It was converted in 1999 by providing traffic diverter pass-through linkages, substituting alternative traffic calming devices for boulevard-facing stop signs, and adding other bicycling enhancements. Additional techniques commonly used to provide bicycle boulevard connectivity for through-traveling cyclists include linking isolated street segments with short bicycle paths or bridges and providing traffic signals or special geometric design aids for crossing busy streets (Pedestrian and Bicycle Information Center, 2010). Bicycle boulevards may hold a greater attraction for the average cyclist than conventional bicycle lanes. This has been quantified for adults in the case of Portland, Oregon (Dill and Gliebe, 2008, Broach, Gliebe, and Dill, 2009a and b, Broach, Gliebe, and Dill, 2011), as covered under “Popularity, Preferences, and Route Choice”—“GPS- and Network-Based Revealed Preference Research.” It is interesting to note that Emeryville, California’s, Horton-Overland Bicycle Boulevard was a bicycle facility solution adopted after consideration of needs of “design cyclists” (a takeoff on highway “design vehicles”) of varying skills and preferences (Pedestrian and Bicycle Information Center, 2010). The Portland studies have led to observation that “there is something more to a bike boulevard than low traffic volumes, improved street crossings, and ‘flipped’ stop signs. The something more 16-84

may be explained by attributes [. . .] such as parking or traffic speeds, or perhaps something more subtle like perceived safety in numbers or simplified navigation.” The Portland researchers have taken care to note the need for further research and especially for replication of the GPS route choice studies in other regions (Broach, Gliebe, and Dill, 2011). Table 16-15, below, encapsulates the few available reportings offering bicycle boulevard usage infor- mation. The 1st table entry is believed to be the earliest U.S. bicycle boulevard, on Bryant Street, in Palo Alto, California. As indicated, initial-phase before-and-after observations found an 85 to 97 per- cent increase in bicycling on the street. The increase was greater than citywide upward trends, but with evidence of bicyclist diversion from parallel streets. Bryant Street serves several schools. Making a later comparison between a 1997 8-hour count of 385 bicycles and the 1982 12-hour volumes ranging 16-85 Table 16-15 Summary of Studies of Individual Bicycle Boulevard Provision Examples Study (Date) Process (Limitations) Key Findings 1. Ciccarelli (2010) (see this section for more information) Bryant Street in Palo Alto, CA, was converted to a bicycle boulevard in 2 segments. The 1.9-mile southern section (1981) included a new traffic signal, with right-turn-only vehicle diverters on Bryant, and a pair of NMT-only bridges over a creek. The 1.2 mile northern section (1992) penetrates the CBD and has no street closure element. (Before/after counts for southern section only.) May 1981 and April 1982 12-hour counts found 85% and 97% bicycling increases at 2 locations on Bryant St. Volumes ranged from 475 to 725/day. Bike volumes on 2 nearby multilane streets declined by 35% and 54%. Vehicle traffic volumes near the 2 street closures went from 953 to 457 and from 481 to 170, with diversions to adjacent streets. A May 1997 8-hour intersection count found 385 Bryant St. bikes. 2. Chaney (2005), City of Vancouver (2009c) (see this section for more information) Three Vancouver, BC, bicycle routes with many bicycle boulevard char- acteristics, 5.5 to 14 km. long and implemented in the 1990s, have had multiple-location before and after counts published in the form of average 24-hr. weekday volumes estimated from 1- or 2-hour counts. (No diversion investigations.) Average 24-hour weekday “after” bicycling ranged from 39 to 1,086 on individual segments. Average “after” for Adanac Bikeway was 743 (4 loca- tions, up 272% in 5 years), average for Off-Broadway was 351 (5 locations, up 76% in 2 years), “after” average for Midtown/Ridgeway Bikeway was 114 (7 locations, up approximately 333%). a 3. Alta Planning + Design (2009a) The Lincoln-Harrison bicycle boule- vard in Portland, OR, 3 miles long, started with traffic calming in the 1980s and 1990s, with wayfinding signage and pavement markings in 2005.(No information on study methods.) “After” bicycle “extrapolated total count” of 1,900 in 2008, up 755% since 1996 (presumably including secular growth associated with development of Portland’s overall bicycle facility network). 4. VanZerr (2010) (see this section for more information) Portland, OR, dwellings facing the SE Salmon St. bicycle boulevard be- tween SE 12th and SE 35th Aves. re- ceived invitations to an on-line com- puter survey: 78 households (31%) responded. (No bike volume data, potential response-rate biases.) Of residents choosing to respond, 6% typically cycled 6-7 days/week; 29%, 4-5 days; 18%, 2-3 days; 18%, 1 or fewer days; 28%, never. Typical destinations were social/recreational (82%), shop- ping/errands (61%), work (59%). Bicy- cling rates exceeded U.S./local norms. Notes: Where substantial additional information on individual studies is provided in text and tables or figures, this is noted — and the location within the chapter is given — in the first column. a Averages and percentage increases calculated by the Handbook authors. Only those count locations with data clearly for both “before” and “after” conditions are included. Sources: As indicated in the first column.

from 475 to 725, the Bryant Street PBIC case study author surmised that commuter and other util- itarian bicycling had declined in the two decades since the peak of the 1970s gas crises, and that more parents were chauffeuring their children to school by auto (Ciccarelli, 2010). The 1982 count data in Table 16-15 may be used to estimate that the first phase traffic mix on the Bryant Street bicy- cle boulevard was, at least near street closure locations, roughly two bicycles for every single motorized vehicle. The 2nd table entry presents before-and-after count results for three “local-street Bikeways” in Vancouver, British Columbia, Canada. These are signed bicycle routes on mostly residential, mostly narrow streets, enhanced with sufficient traffic-calming and bicycle-preference engineer- ing features to be properly considered as bicycle boulevards. A few short elements of off-road paths and “paper street” path segments were apparently included at the time of implementation, but it is believed that there were no sections of bicycle lanes as of the dates of the “after” studies (Chaney, 2005, Alta Planning + Design, 2009a, City of Vancouver, 2009c, Navin and Anderson, 2009).18 Before-and-after bicycle counts, taken at multiple points on each Vancouver “Bikeway,” show weighted-average 2- to 5-year cycling increases per Bikeway of 76 percent, 272 percent, and approximately 333 percent. These increases include not only effects of each new Bikeway but also a general upward trend in bicycling within Vancouver proper (Chaney, 2005). Bicycle trips within and to Vancouver increased by 180 percent between 1994 and 2004 (City of Vancouver, 2009b), thus the upward secular trend would have been on the order of 18 percent per year. It is of interest to note that these fairly modest but carefully selected, augmented, and well-integrated Bikeway routes seem to be full players in an apparently very successful citywide bicycle facility grid. Portland, Oregon, is an example of a city within the United States that has implemented bicycle boulevards on a number of streets. The 3rd and 4th entries within Table 16-15 pertain to the Lincoln-Harrison and SE Salmon Street bicycle boulevards. The Lincoln-Harrison facility attracts an estimated 1,900 bicycle trips daily (Alta Planning + Design, 2009a). No count information is pro- vided for SE Salmon Street, but it is 1/2-mile south of and parallel to the Lincoln-Harrison bicycle boulevard, and thus part of the same Portland bicycle facility network. It is included in the tabu- lation for the related information obtained in a survey of dwellings facing Salmon Street. For exam- ple, asked if they enjoyed living on a bicycle boulevard, two out of three survey respondents liked it “A lot,” while only one in nine responded “Not at all.” The rest of respondents were indifferent or liked it “A little.” The bicycling activity reported by responding Salmon Street residents (see Table 16-15) seems to far exceed national averages. Some 59 percent of respondents fronting the Salmon Street bicycle boulevard reported cycling at least 1 day a week (VanZerr, 2010). Comparisons must be made with caution, given that survey response within households of Salmon Street survey respondents was a personal choice among household members rather than random or pre-defined selection. (The 31 percent survey response rate pertains to contacted dwellings, not individuals.) Still, compari- son is of interest considering that the 2001 NHTS found only 4.5 to 12.7 percent (among covered Metropolitan Statistical Areas) of surveyed individuals to have cycled sometime during a week (Krizek et al., 2007). 16-86 18 The term “paper street” refers to linear segments of land dedicated/acquired for use as street right-of-way but never built upon for the purpose. Their use for paths connecting between built streets does not violate their utility as impediments to undesired through traffic.

The Salmon Street study author identifies potential survey biases, but if there were no response bias at all in the survey, the identified bicycling rate would be nearly 7 times national averages. The Salmon Street analysis itself reports a work commute bicycling rate comparison that suggests the Salmon Street rate may be on the order of 10 times the citywide bicycle commute mode share (VanZerr, 2010), remarkable even if definitional differences and bias issues exaggerate the differ- ential. (More information on national bicycling rates is found in the “Related Information and Impacts” section under “Extent of Walking and Bicycling”—“Extent of Bicycling.”) Housing choice “self-selection” may be a factor for bicycle boulevards. Among responding Salmon Street residents, 18 percent indicated that bicycle boulevard status was a positive factor in housing choice. No one selected the “negative factor” questionnaire option. Persons residing on the street prior to bicycle boulevard designation constituted 33 percent of respondents, 29 per- cent didn’t know it was a bicycle boulevard when they moved in, and 20 percent knew but didn’t factor it into their housing choice decision. An analysis was made of the combined effect of self-selection—positive factoring of the bicycle boulevard into the housing location decision— and a perception, also reported in the survey, that living on a bicycle boulevard makes bicycling more likely. Four combinations were identified and group average days per week of bicycling were calculated (VanZerr, 2010): • Persons who self-selected and also perceive the bicycle boulevard presence makes them more likely to bicycle (15 percent of respondents) bicycled 3.59 days/week on average. • Persons who did not self-select (for whatever reason) but do perceive bicycle boulevard presence as making them more likely to bicycle (32 percent of respondents) bicycled 2.44 days/week on average. • Persons who self-selected but do not perceive the bicycle boulevard presence makes them more likely to bicycle (4 percent of respondents) bicycled 2.39 days/week on average. • Persons who neither self-selected nor perceive bicycle boulevard presence makes them more likely to bicycle (50 percent of respondents) bicycled 1.92 days/week on average. The limited quantifications of bicycle volume increases with bicycle boulevard introduction com- pare favorably with increases reported for bicycle lanes. The tripling of on-street bicycle volumes on average for the three early facilities in Vancouver, and the huge increase reported over time for the Lincoln-Harrison bicycle boulevard in Portland (Table 16-15), are much more than the four-city growth average for bicycle lanes of approximately 50 percent derived earlier from Table 16-11. On the other hand, absolute peak-hour “after” volume counts reported for the Vancouver facilities (not published for the more heavily used Lincoln-Harrison facility in Portland) show a moderate facility average of about 50 cyclists per peak hour (Chaney, 2005). This bicycle usage is about 1/3 less than the average for the four bike lane cases for which peak period or one-hour volumes are pro- vided in Table 16-11 (San Francisco and Fort Lauderdale).19 The point here is neither to offer a precise bicycle volume growth estimate for either bicycle lanes or boulevards, from this small sample, nor to establish bicycle lane or boulevard volume averages. 16-87 19 Peak period bike lane volumes are given for Anderson Road in Davis within case study Table 16-134, but this is a statistical outlier, where both the “before” bicycle volumes (with no special bicycle treatment) and the “after” volumes were an order of magnitude higher relative to the more typical bicycle volumes reported here.

More volume data should become available from the “National Bicycle and Pedestrian Documentation Project” or similar endeavors as they mature (see the “Additional Resources” section). The bottom line is that, on the basis of aggregate growth and volume measures, both bicycle lanes and bicycle boulevards show roughly equivalent promise in terms of volumetric traveler response to on-road facilities. Area characteristics may, of course, dictate bicycle lane versus bicycle boulevard strat- egy selection as a result of physical restraints and opportunities. These same area characteristics may also influence usage. Bike lanes are typically placed on arterial or collector streets and in down- towns, while bicycle boulevards are most suited to low vehicular volume, mostly residential, continu- ous or interconnected local streets. It should also be borne in mind that additional research may show one or the other treatment to be more effective for individual disaggregate categories of existing or potential cyclists. In this con- nection, evidence that Portland’s bicycle boulevards are especially attractive to female cyclists is presented in the “Underlying Traveler Response Factors” section (see “Trip Factors”—“Bicycle Trip Distance, Time, and Route Characteristics”—“Bicycle Route Choice”). Signed Bike Routes. One study encountered addresses commute mode shifts to bicycling in commutersheds of newly signed bike routes. Austin, Texas, implemented roughly 20 bike route seg- ments during an analysis period of 1990–2000. City staff worked with the local bicycling community to identify routes already favored by cyclists as being bicyclist-friendly. The typical such route is a residential street running parallel to major arterials. The analysis was part of the six-city commute mode shifts study summarized in the 3rd entry of Table 16-11. It produced the results entered in Table 16-13. The signed bike route commutersheds in Austin, which together encompassed roughly one-half or more of the city’s geographic area, exhibited an 0.32 percentage point gain in bicycle com- mute mode share relative to 0.19 percentage points for the city as a whole. The corresponding bicy- cle share growth rates were 37 percent around the bike routes and 25 percent citywide. Commute mode changes were also reported for two cities implementing “on street bikeways,” −0.6 percent in Salt Lake City and +24.6 percent in Madison, Wisconsin (Cleaveland and Douma, 2009). The other bit of information on signed bike route attractiveness comes from the follow-up Portland bicyclist route choice model development process. They were tested as a facility type and it was concluded that unimproved signed bike routes were “insignificant factors” once other variables were accounted for (Broach, Gliebe, and Dill, 2011). As is the case with bicycle lane count-based studies, the bicycle boulevard and route studies pro- vide only partial or no information on diversion of bicyclists to the facility relative to bicycle trips resulting from mode shifts, shifts in destination, or induced cycling. (Some diversion information is available for Palo Alto—see the 1st Table 16-15 entry.) Similarly, no surveys of bicycle boulevard or route user makeup have been encountered beyond the survey responses by adjacent residents along the Salmon Street bicycle boulevard and the one set of in-parkland bike route observations provided in the Montgomery County, Maryland, case study. Primary cycling destinations of the Salmon Street residents have been included in Table 16-15. 16-88

Shared Use, Off-Road Paths and Trails The off-road, shared use facilities addressed in this subsection are the counterpart to the combination of sidewalks, on-road bicycle lanes, and shared-roadway bicycle-preference treatments covered in previous subsections. Shared use paths accommodate pedestrians (inclusive of manual and motor- ized wheelchairs), cyclists, and other non-motorized wheeled users, including in-line skaters when pavement and design conditions allow (AASHTO, 1999). Despite the popular “bike path” appellation, such facilities are very rarely if ever restricted to bicycles alone in the United States. As noted earlier in Footnote 3 of the “Overview and Summary,” although “path” is the preferred technical term for urban applications, local-area usage of the term “trail” has been adhered to where known. The subsection title, “Shared Use, Off-Road Paths and Trails,” is intended to convey that pedestrian- only walking and hiking trails are, except in special cases, not covered. Neither are on-road separated facilities such as cycle tracks and buffered bike lanes, these having been included in the “Bicycle Lane Variations, Bicycle Boulevards, and Other Signed Bicycle Routes” discussion immediately preceding. Preferences, Route Choice, and Walk/Bikesheds The opportunity for cyclists to route their bicycle trips over specially designed facilities, such as bike paths, bike lanes, or bicycle boulevards, is demonstrably an encouragement to cycling. Each of these facility types designates physical space for bicycle use and addresses two key underlying traveler response factors: perceived safety and travel time. Different cyclists may prefer to use different facility types for different trip or recreational/exercise purposes. Bicycle lane versus off-street path preferences have been examined in the preceding “Bicycle Lanes and Routes” subsection. The comparative assess- ments are primarily contained in the “Popularity, Preferences, and Route Choice” discussion, but there is related information under “Bicycle Lane System Coverage” as well. Comparative Preferences Recapitulation. Much of the overall body of research on bicyclist prefer- ences for off-road path use as compared to bicycle lane use has produced seemingly inconsistent results. Also, a majority of the research has focused only on the bicycle-to-work commute, not other uses. A tentative conclusion that either facility type may be equally useful and attractive for bicyclists overall seemed to fit study results produced prior to findings emerging from GPS-and-network-based research in Portland, Oregon (Tilahun, Levinson, and Krizek, 2007). The Portland findings, at this point not yet replicated elsewhere, indicate a bicyclist preference hier- archy that applies to both work purpose and other utilitarian trips so long as alternative facility types afford adequately direct routings. In this hierarchy, off-road paths are the most preferred facility type, followed by bicycle boulevards, in turn followed by conventional bicycle lanes or quiet resi- dential streets, and with all of these being more preferred than bicycling on moderately or very busy streets with no special treatment (Dill and Gliebe, 2008, Broach, Gliebe, and Dill, 2009a and b, Broach, Gliebe, and Dill, 2011). Limited experience with cycle tracks and other separated on-street facilities is promising but does not yet allow estimation of their place within this hierarchy. U.S. off-road paths and trails serve a broader clientele than bicycle lanes and other on-road bicycle facilities, which obviously cannot serve NMT users such as pedestrians or joggers the way multi-use paths do. The choice for users on foot is not between paths, bike lanes, and streets, but between paths, sidewalks, and sides or shoulders of streets and roads. This is a choice afforded very little research, although it has been shown that path proximity will lead to more path use (but not nec- essarily more walking). (For more information, see discussion in connection with Table 16-19, 1st and 4th entries.) 16-89

A relevant circumstance affecting all facility types is that selection of a facility type for implemen- tation will often be dictated by individual geographic, physical, and/or traffic conditions. Those, in turn, will be determined by the potential trip origins and destinations proposed to be served. There is reason to be concerned that in GPS- and network-based studies, important differences in preferences among disparate user groups may have failed to surface in the aggregate-data and often commuter-focused analyses that predominated. Because these issues are covered elsewhere, treat- ment of them in this subsection is limited to additional information focused exclusively or substan- tively on path use. Reference should be made back to the “Bicycle Lanes and Routes” subsection for a more comprehensive overall discussion from the perspective of bicycling. In addition, the “Underlying Traveler Response Factors” section further explores differences among distinct user groups. See the “Trip Factors” subsection (discussion of Table 16-67) as well as “User Factors.” The 2002 national survey on pedestrian and bicyclist attitudes and behaviors found that bicycle and walking paths and trails were “most used” in undertaking the respondents’ most recent trip in the case of 6 percent of all walk trips and 13 percent of all bicycle trips (NHTSA and BTS, 2002). (For the full tabulation see Table 16-96 under “Related Information and Impacts”—“Facility Usage and User Characteristics”—“Frequency of Facility Usage by Facility Type.”) In contrast, a survey of Florida residents making at least 1 bike trip in the past 7 days found that 27 percent of their trips were mostly on bike paths and another 18 percent were partially so (NuStats International, 1998). As previously noted, with regard to this type of information, the facility choices reported could be the result of either facility-type preference or facility orientation/availability with respect to travel needs, or both. Route Deviation to Use Paths. Portland, Oregon, bicycle facilities region-wide in 2007 included some 550 miles of bike lanes (almost 78 percent of all bicycle facility mileage), 30 miles of bicycle boulevards (4 percent), and 130 miles of separate bike paths (18 percent). In this context, 52 per- cent of all utilitarian bike travel took place along bicycle facilities. The bicycle-miles breakdown among bike-facility types was 54 percent on bike lanes, 20 percent on bicycle boulevards, and 26 per- cent on shared use trails, evidencing a disproportionate attraction to bicycle boulevards and off-road trails. Exercise and other loop trips were less oriented to bicycle facility use in general and to use of bike lanes and bicycle boulevards. They were slightly more oriented toward shared use trails. Route choice plots showed substantial exercise and recreational travel to be taking place on undifferentiated roads in more rural parts of the area (Dill and Gliebe, 2008). Various surveys have been reported to indicate willingness to incur extra travel to bicycle on off-road shared use paths (Guttenplan and Patten, 1995). Estimates based on the Portland GPS/network-based studies (see Table 16-10 and associated discussion in the “Bicycle Lanes and Routes” subsection) indi- cate that the average cyclist will travel substantially out of the way to use an off-road trail in preference to other options. The estimated willingness to deviate is 26 percent out of the way if the other option is a quiet street, and 57 percent out of the way if the other option is a moderate-traffic street without a bike lane. These Portland estimates are derived on the basis of bicycle trips for utilitarian travel pur- poses such as commuting and running errands (Broach, Gliebe, and Dill, 2009b). In comparison to the Portland utilitarian-travel-based estimate of 26 to 57 percent, research in Minneapolis on bicycle travel for all purposes—weekends as well as weekdays—estimated that bicy- clists are traveling an average of 67 percent longer to include an off-street trail facility in their route. Findings were derived from routing information provided by cyclists intercepted in a 13-station sur- vey. Trips employing other than bicycle access to reach the trails were excluded (Krizek, El-Geneidy, and Thompson, 2007), as was the case in the Portland studies. Although the study locations and route determination methodologies of the two studies differed, the most obvious potential explanation for the greater amount of route deviation observed in Minneapolis is the inclusion of substantial recre- 16-90

ational travel in the data set. At the end of this discussion, estimates of the price elasticity of demand for recreational use of off-road trail facilities based on distance of travel to the facility are presented. Path Walk/Bikesheds. Some studies have translated willingness to travel extra distance to use off- road paths into quantification of path use as a function of distance from the path. The quantifications use various metrics, hindering comparison, but a path commutershed and walk/bikeshed phenom- enon clearly exists. (The term commutershed is used here in the context of work purpose trips and the term walk/bikeshed is applied in the context of trips for all purposes.) An examination of journey-to-work bicyclist origins obtained in a 1993 weekday peak-periods intercept survey on Seattle’s Burke-Gilman off-road trail plus a second intercept survey on a bike lane in the north fringe of the CBD found 24 percent of all commuters to have originated within 0.4 km. (1/4 mile) of the trail, 37 percent within 0.8 km. (1/2 mile), and 53 percent within 1.2 km. (3/4 mile).20 An examination of commute times for both trail users and non-trail users found travel time means of 29 minutes in the first 1/4-mile band, 35 minutes in the second 1/4-mile band, and 31 minutes in the third (farthest from the trail) 1/4-mile band. This led the researchers to conclude that commuters in the second band were traveling out of their way to use the trail, but that most commuters from beyond 1/2 mile of the trail were unwilling to incur the longer travel distance. On this basis they suggested a trail commutershed boundary for bicyclists of 1/2 mile, as measured from the trail (Shafizadeh and Niemeier, 1997). A random sample of adults in Arlington, Massachusetts, was surveyed to explore patterns of Minuteman Trail use for any recreational or transportation physical activity including both walk- ing and bicycling. The trail is a shared use rail-trail facility traversing Arlington and two other towns in the northwest Boston suburbs. An “Arlington Physical Activity and Bikeway Survey” obtained self-reported information on trail use, all types of recent physical activity, and health and socio-demographic status, along with perceived neighborhood environment, distance to trail, need to cross busy streets for access, and need to traverse steep grades for access. Survey respondent addresses were geocoded, allowing independent geographic information system (GIS) calculation of access distances, major street crossings, and grades. A Minuteman Trail user was defined as any respondent making any use of the trail in the preceding 4 weeks. Predictive models incorporating the significant variables provided estimates of odds ratios for utilizing the trail of 0.58 (based on GIS trail access measures) and 0.65 (based on perceived access measures) for every 1/4-mile increase in access distance (Troped et al., 2001), suggesting that residents were 35 to 42 percent less likely to make use of the trail for each added 1/4-mile.21 This estimate would indicate that persons living in the bands between 1/2 and 3/4 miles from the trail, and not having an extra busy street or steep grade to traverse, would have roughly 1/3 to 2/5 the likelihood of using the trail as someone living within the first 1/4 mile of the trail. The researchers did not suggest walk/bikeshed boundaries, but 1/2 to 3/4 mile each side of the trail would seem appropriate. 16-91 20 Given land mass, water body, highway, bridge, and bike facility geography at the time, these two surveys likely intercepted a large portion of all commuter bicyclists from north Seattle. 21 An “odds ratio” quantifies the relation between two odds in order to illustrate the amount by which the prob- ability of a certain outcome differs, if at all, between two groups. Generally, an odds ratio is calculated as the odds of the outcome (trail use, in this case) for the affected group (1/4 mile further away) divided by the odds of the outcome for the group not so affected (not 1/4 mile further away). An odds ratio of 1.0 implies equal likelihood, an odds ratio of more than 1.0 implies greater likelihood, and an odds ratio of less than 1.0 (in this case, 0.58 to 0.65) implies lesser likelihood (in this case, of trail use among those 1/4 mile further away).

The previously-introduced intercept-survey-based analysis of users of Hennepin County off- street trails within Minneapolis likewise covered all travel purposes over both weekdays and weekends, but was restricted to users making their entire trip via bicycle. Over half the cyclists traveled less than 2.5 km. (1-1/2 miles) from their homes to use the trails. Over 3/4 were within 5.0 km. (3 miles). Decay functions were fitted to the percentage of trips coming from different dis- tances for different travel purposes. The decay functions for each travel purpose dropped off sharply at first, tending toward flattening out at one side or the other of 5.0 km. However, the functions for work/school and shopping trips dropped off more sharply than the function for recreational trips (Krizek, El-Geneidy, and Thompson, 2007). The Minneapolis findings would seem to imply a broader bikeshed than the postulated Seattle bicycle commutershed or Arlington walk/bikeshed. These studies address either bicyclist path users or a mix of walkers and bicyclists. Addressing walk- ers per se, the Minneapolis researchers note that “the pedestrian literature widely cites that people are willing to walk a quarter of a mile or so” and trace this finding back to the early 1980s (Krizek, El-Geneidy, and Thompson, 2007). Certainly the 1/4-mile walk access limit has served for many decades as a rule-of-thumb for local bus route planning (see Chapter 10, “Bus Routing and Coverage”). Of course, this determination regarding bus access may or may not be an equivalent cir- cumstance, especially since path use for walking involves not only access but also the walking on the path itself. Path Orientation and Value. It deserves repeating here that facility-type preferences, while impor- tant in the choice of whether and where to walk or cycle, are subordinate to origin and destination access needs in the case of utilitarian active transportation. Utilitarian NMT usage is greatly influ- enced by both alignment with travel needs and how direct and logical the routing is (Alta Planning + Design, 2009a). The importance of providing connectivity to places people want and need to go led one early study report to question whether separated paths could possibly “inspire bicycle commut- ing” given the propensity of bike path alignments to “follow scenic corridors and [. . .] not necessar- ily lead to major destinations” (Goldsmith, 1992). The findings reported on below suggest that while this concern applies to some shared use, off-road paths and trails, it does not pertain to others well oriented to utilitarian travel needs. Many cases lie somewhere in-between. Also, one aspect not to be overlooked is that attractiveness of off-road paths for recreation and physical exercise provides qual- ity of life and health benefits that may be of sufficient value to compensate, in terms of public benefits, where path alignment cannot support utilitarian usage as well. Studies of path economics, as noted previously, have used path access distances to develop calcula- tions of the price elasticity of demand for recreational use of off-road path facilities. For example, stud- ies of 2003–04 use of the Washington and Old Dominion (W&OD) Trail in the Virginia sector of the U.S. National Capital Region produced price elasticities derived on the basis of cost to access the trail, which itself is free to the public. A 2003 mileage cost rate of $0.131/mile was applied to round-trip access mileage and the result was related to frequency of trail use. Two different estimating formulations were used. These gave travel cost price elasticities, calculated at the means for recreational users not living directly on the trail, of −0.34 and −0.22.22 Other studies covering different paths have produced price elasticities of −0.21 to −0.43, and −0.68 (Bowker et al., 2004). The average of −0.38 suggests a sensitivity 16-92 22 A travel cost trail use elasticity of −0.3, for example, indicates an 0.3 percent decrease (increase) in trail use in response to each 1 percent increase (decrease) in cost, calculated in infinitesimally small increments. (See also Footnote 12 in the “Street Crossings” subsection.)

to cost that is close to some other important transportation price elasticities, most notably the average fare elasticity for public bus transit ridership. Shared Use Path Implementation Facilities that are altogether new when implemented, such as most shared use paths are, present an analytical challenge in that there is no route-specific “before” data with which to compare. Faced with this constraint, available studies have employed a variety of techniques to explore travel or physical activity changes in response to path provision. Table 16-16 presents a summary of shared use path studies that have employed retrospective or “what if?” questioning, time-series work commute mode share observations, or some form of adaptive before-and-after data. A few additional studies with “after” statistics that contribute useful information, even without pairing with “before” data, are included. 16-93

16-94 Table 16-16 Summary of Retrospective and Before and After Studies of Individual Shared Use Path Implementation Examples Study (Date) Process (Limitations) Key Findings 1. Indiana University (2001) (see case study “Six… Trails — Indiana Trails Study” for more) In 6 Indiana locales, 1 trail each (5 opened in late 1990’s, 1 in 1980’s), were studied in 2000 with counts, user interviews, and surveys. Trip- based user data and perspectives were obtained. (No basis for direct measurement of NMT increases.) August weekday trail volume ranged from 1,620 (Indianapolis, Monon Trail) to 170 (Greenfield, Pennsy Trail); 2,350 to 190 on weekend days. From 14% to 19% reported engaging in their activity only because of the trail, while 70% to 87% engaged more in their activity. 2. Welzenbach (1996), Greenways Incorporated (1992) (see this section for more information) Chicago region shared use paths in- clude early rails-to-trails conversions like the Illinois Prairie Path, opened in the late 1960s and serving 18 cities and villages in 3 counties with multiple branches (55 miles circa 1990). In 1995, 54 segments of 18 paths representing 196 miles were surveyed, obtaining responses from 4,589 (42%) of on- path walkers/cyclists. (Runners/skaters not interviewed.) Purpose distributions were: Work (including station access) 9%; non- work utilitarian (incl. school), 15%; recreation, 31%; other (incl. recreational site access), 45%. Auto was alternate mode for 43% of work, 37% of non- work utilitarian, and 24% of other- purpose trips. An earlier, separate analysis of Census tracts along 5 key linear paths found 1980 Census work purpose shares to be 15.6% bicycle mode, vs. 1% regionally. 3. Puget Sound Regional Council (2000), Moritz (1995 and 2005a and b) (see this section and “Ped./ Bicycle Systems and Interconnections” for more) The shared use Burke-Gilman and Sammamish River Trails were opened in the late 1970s and joined, in 1993, into a 27-mile trail serving UW and the north fringe of central Seattle. Multi-location 7 AM - 7 PM trail user counts/surveys have been taken at 5-year intervals starting in 1980. (Major exogenous factors; multiple on-path survey points give duplicate observations of long trips.) Tuesday (Saturday) 4-station trail count averages grew from 400 (1,900) in 1980 to 2,200 (3,600) in 1995 then dropping, partially rebounding to 2,000 (2,300) in 2005. The drop was in cycling (2/3 to 4/5 of the total). The pedestrian count has grown fairly steadily. The work/ school proportion has grown from 1/10 to between 1/4 and 1/2, with a proportional decline (but stabilization in the absolute) of recreation/exercise users. 4. Guttenplan and Patten ( 1995), Ewing (1997) (see this section for more information) The Pinellas Trail along Florida’s West Coast is a 47-mile rail trail, connecting Tarpon Springs and St. Petersburg. Users were surveyed at 8 locations from 6:30 AM to 6:00 PM during a November 1993 weekday, filling out the survey only once. (Survey not keyed to a count). Some 35% of users surveyed reported using the trail for utilitarian transporta- tion purposes such as commuting to work or school, or shopping. Of those using the trail to get to work or school 87% did so at least 2 days a week, with 60% using it 5 days a week. Trail use approaches 100,000 people a month. 5. Barnes, Thomp- son, and Krizek (2006) In Minneapolis-St. Paul 4 major off- road trails and 3 major bike lane facilities were opened 1990-2000. Commute trip bike mode share Bike mode shares at the outset, inside the commutersheds, averaged 4 to 5 times the shares in the rest of the Cen- tral Cities. The trail and bike lane com- (see also “Bicycle Lanes and Routes” and Table 16-12) changes were computed for TAZs within 1 mile (1.5 miles for facility termini). (Bike/work trips only.) mutershed bike shares each increased by averages of 1.38 percentage points, up 43% in the case of off-road trails. a 6. Cleaveland and Douma (2009) (see “Bicycle Lanes and Routes” and Table 16-13 plus this section for more information) Commute trip bike mode share changes were computed for Census block groups within 1.55 miles of bike facilities opened 1990-2000 in 6 U.S. cities. Off-street paths were included in Austin, Madison, Colorado Springs, Salt Lake City, Orlando. (Bike/work trips only.) Off-road trail commutershed bike shares in Austin, TX, increased by an average of 0.88 percentage points, up 24%. a Work commute mode share outcomes for the off-road paths in the other cities were not statistically significant. The research examined neither walk trips nor non-work trips.

16-95 Table 16-16 (Continued) Study (Date) Process (Limitations) Key Findings pleted in September, 2002. The 2005 county population was about 85,000. (Counts/penetration surveys only.) have used the trail. The corresponding rate from a random Comprehensive Planning Citizen Survey in March 2005 was 53%. 8. Chaney (2005) (see this section for more information) An interim section of the Hudson River Trail in Manhattan between West 12th and 55th was replaced with a straight 15-foot wide pathway. Counts were made in May, 2001, 1 month after completion. “Before” counts were in September, 2000. (No NMT type differentiation.) Weekday 6-hour NMT user volumes (7:30-9:30 AM, 12-2 PM, 4:30-6:30 PM) increased from 731 to 2,056 (up 181%) at W. 17th and from 319 to 1,248 (up 291%) at W. 34th. Weekend 6-hour volumes (10 AM-4 PM) increased from 1,986 to 4,498 (up 126%) at W. 17th and from 868 to 3,474 (up 300%) at W. 34th. 9. Pedestrian and Bicycle Information Center (2010) A 1.2-mile path was constructed to connect Delaware Valley College to Doylestown, PA, penetrating a bar- rier created by a bypass road. (No analysis of reasons for low usage.) Counts found 10 people on average on various parts of the path each day, or about 3,000/year. Part of the path is wedged between a high traffic barrier and fence [and may be unattractive]. 10. Five additional studies examined by Pucher, Dill, and Handy (2010) (see this section for more information) A review was carried out of 5 addi- tional studies of paths before and after construction or introduction of bicycles. In 1 instance, trail opening was combined with a marketing campaign. (Results summary was limited to cycling outcomes.) In 2 studies no changes were observed in levels of cycling of nearby residents, while increases in numbers of cyclists were found in 2 studies, and an increase in minutes of cycling by residents within 1.5 km. was seen in the case where marketing was applied. Notes: Where substantial additional information on individual studies is provided in text and tables or figures, this is noted — and the location within the chapter is given — in the first column. a Percentage increase(s) or ratios and some totals calculated by the Handbook authors. Sources: As indicated in the first column. 7. Pedestrian and Bicycle Information Center (2010), Roback (2004) The Interurban Trail of Ozaukee County, Wisconsin, north of Mil- waukee and south of Sheboygan, is a 30-mile north-south mostly off- road paved rail-trail essentially com- The sum of 7-day, 14-hour counts made at 2 locations in August, 2004 was 8,825 users in 1 week. A summer 2003 Community Health Survey found 25% of respondents countywide to Indiana Trails. The 1st study entry in Table 16-16 is of particular interest, not only because of ret- rospective survey questions probing changes in physical activity. It is also uniquely informative because of the manner of positioning surveyors to obtain trip-based user information, i.e., trail visit data, rather than trail-traffic-based characteristics information. In addition, it offers the perspec- tive of being a statewide study of selected individual trails in differing locales. This study and the survey technique and its implications are more fully described under “Six Urban, Suburban, and Semi-rural Trails—Indiana Trails Study” in the “Case Studies” section. Certain trail user and use characteristics identified in the Indiana Trails study are somewhat atypical compared to most trails (see Tables 16-137 and 16-138 in the case study). It is not clear whether this simply reflects regional or other locational differences or whether it arises from obtaining actual trip- based data, eliminating a bias toward interception of and reporting on longer trips more than shorter ones. The usual male dominance of trail use is found on only half the trails, with the facilities in Greenfield and Portage exhibiting almost equal balance, and a moderate female dominance (54 per- cent female) on the Monon Trail in Indianapolis. The percentage of users on foot, walking or running, is almost 60 percent or more on 4 trails, 50 percent on the Prairie Duneland Trail in Portage, and only

dominated by cycling—together with skating—on the semi-rural Cardinal Greenway rail-trail in Muncie (Indiana University, 2001). A range of 14 to 19 percent of surveyed users on the individual Indiana trails indicated that their engagement in their chosen trail use activity was because of the trail’s availability. Another 70 to 80 percent indicated that they were walking, running, cycling, or skating more because of the trail. The Indiana trails fall in the category exhibiting low usage for work commuting, with the highest work purpose share (5 percent) found on the Indianapolis Monon Trail. Interestingly, however, a companion survey question indicates that some users are “killing two birds with one stone,” getting to work while gaining exercise, while reporting their trail use purpose as health/exercise or recre- ation (Indiana University, 2001). The actual Monon Trail use for commuting may thus be higher, although the percentage of trip makers reporting entering and exiting at the same point was 91 per- cent, placing a logical upper limit on utilitarian-purpose trip making of 9 percent.23 In discussions of travel purpose of trail users, it is well to remember overall context. Commuting to work is not the dominant purpose of trip making overall in the United States. The 2001 NHTS found only 15 percent of all U.S. person trips by all motorized and non-motorized travel modes to involve com- muting to or from work. Work-related trips and school/church trips added 3 percent and 10 percent, respectively. Other non-work travel purposes accounted for 45 percent, excluding social/recreational trips, which were another 27 percent of all U.S. trip making (Bureau of Transportation Statistics, 2003a). Greater Chicago Paths. Examples of paths and trails with higher reported commuting utilization than Indiana in the year 2000 are provided by the Chicago area. Chicago area paths overall, as per the 2nd entry of Table 16-16, exhibited a 9 percent work purpose share for walking and cycling in 1995 surveys. Another 15 percent of surveyed users had various other utilitarian purposes includ- ing school access. The reported 9 percent work purpose share includes trips to and from rail sta- tions, almost exclusively commuter rail or rapid transit. The high proportion of “other” trips, 45 percent overall, includes walking and cycling to specific recreational sites and to visit friends and relatives. This study and most other studies reported on below obtained path-traffic-based data rather than user-based data, given the on-path intercept survey methods utilized. Individual Chicago area path use for work purpose travel ran as high as the 22 percent proportion found on the North Shore Trail. Recreational/exercise use of the North Shore Trail was only 19 per- cent (Welzenbach, 1996, Greenways Incorporated, 1992). The North Shore Trail, the Illinois Prairie Path, and other principal Chicago area shared use paths are on the roadbeds of former commuter- oriented electric railways or other radial railroads, giving them natural alignment with work and other utilitarian travel demand. Average one-way bicycle trip lengths identified in the Chicago surveys were 3.6 miles for work trips, 3.2 miles for non-work utilitarian trips, and 4.7 miles for other trip types, excluding recreational trips. (The modest differences among trip types may have minimized potential trip-length biases insofar as trip purpose distributions are concerned.) One-quarter of all survey respondents reported that 16-96 23 The obscuring of utilitarian travel by “primary purpose” survey question responses of “exercise” or the like may be more widespread than realized. Interviewers on the Iron Horse Regional Trail in the San Francisco East Bay area also reported a tendency to give “recreation” as a trail trip purpose when, in fact, respondents had actual utilitarian destinations but chose to use the trail as an exercise opportunity (East Bay Regional Park District, 1998). Some newer research efforts have turned to instructing interviewees that any trip with a purposeful destination should be classified according to that purpose and not the motivation for engaging in active transportation (Dill and Gliebe, 2008).

their alternative mode for their trip, assuming the path did not exist, was auto (Welzenbach, 1996). While the implications for vehicular travel mitigation are substantive, it must be recognized that the “auto” mode terminology can encompass both auto driver—implying a car removed from the road— and auto passenger. Seattle Urban/Suburban Trails. The 3rd entry into Table 16-16 offers a further perspective on dis- tributions among NMT modes of trail traffic and on allocations of trail traffic among work, other utilitarian, and recreational/exercise purposes. The Burke-Gilman and Sammamish River Trails data also open a window on trail use changes over time as a pair of facilities and a bikeway sys- tem mature. These two Seattle-area trails—now joined end to end—wrap around the north end of Lake Washington to connect central Seattle’s north fringe with northeast Seattle, including the University of Washington, and suburbs east and north of the lake in northern King County (Puget Sound Regional Council, 2000, Moritz, 1995). Table 16-17 summarizes the counts and observed NMT mode distributions from the Tuesday and Saturday counts taken every 5 years, showing the proportions of cycling, walking, skating, and other modes. The table title deliberately identifies the NMT mode distributions as “classification count” results, because the taking of on-trail observations does not result in a direct assessment of user char- acteristics, but rather in an assessment of observed volume characteristics. Although as many as eight intercept count stations were employed for individual years, the data in Table 16-17 are limited to that from the four stations (three for 1980) with consistently available NMT mode observations. 16-97 Table 16-17 Seattle Area Burke-Gilman/Sammamish Trail Four-Station Classification Count Averages Over Time, 1980–2005, Late May, 7:00 AM–7:00 PM Year Bikes Percent Peds. a Percent Skate b Percent Other b Percent Total Tuesday 1980 c 260 64% 129 32% — — 18 4% 407 1985 790 64% 429 35% — — 20 2% 1,238 1990 624 67% 300 32% — — 11 1% 936 1995 1,590 72% 452 21% 144 7% 6 0% 2,192 2000 1,057 63% 530 31% 97 6% 6 0% 1,690 2005 1,357 68% 584 29% 48 2% 13 1% 2,002 Saturday 1980 c 1,617 83% 278 14% — — 42 2% 1,937 1985 1,747 78% 384 18% — — 59 3% 2,190 1990 2,235 81% 497 18% — — 32 1% 2,764 1995 2,874 79% 496 14% 272 7% 3 0% 3,645 2000 1,464 71% 506 24% 85 4% 20 1% 2,076 2005 1,650 72% 574 25% 46 2% 21 1% 2,291 Notes: The four count stations are Gas Works and Sheridan Beach (Lake Forest Park) on Burke- Gilman, and Woodinville and Redmond on the Sammamish Trail (see also table-Note C). a The pedestrian count includes walkers, joggers, and runners. b Skaters were entered as “Other” in 1980, 1985, and 1990. c The Gas Works Park count station was not open in 1980, thus the counts and percentages for 1980 are actually 3-station averages, with only one station on the Burke-Gilman Trail. Sources: Moritz (2005a and b), with averaging and pre-2005 percentages by the Handbook authors.

The sharp growth in trail use by bicycles in the 1985–1995 period has been ascribed to the inter- connection of the Burke-Gilman and Sammamish River Trails, accomplished in two stages, in 1988 and 1993 (Moritz, 1995, Puget Sound Regional Council, 2000). This attribution of cause is backed up by counts taken in 1990 and 1994 at each end of the final “Missing Link,” as described in the “Pedestrian/Bicycle Systems and Interconnections” subsection that follows. However, it can be seen that a drop-off in cycling was observed in 2000 and found to continue into 2005. Walking, on the other hand, grew fairly steadily in absolute terms over the quarter-century of observations. Skating jumped from less than 1 percent of all trail traffic in 1990 to 7 percent in 1995 and then declined. Possible reasons for the drop-off in bicycling on the combined trails after 1995 include:24 • There may have been precipitation forecasts, cooler weather, or other not fully recognized exoge- nous factors that affected usage on the survey days. Adverse weather on the count Saturday in 2005 forced premature closure of one-half of the count stations (Moritz, 2005a and b). • The novelty of a continuous 27-mile scenic urban trail may have somewhat worn off for recre- ational users. • The trail may have become a victim of its own success, with complaints of crowding on the trail encouraging choice of alternative routes and activity venues (Puget Sound Regional Council, 2000). • Greater development of trails and other bikeways throughout the region may have led to a better distribution of use by those with a choice, particularly cyclists accessing a trail by motor vehicle for purposes of exercise or recreation. Proportions of survey respondents reporting use of a car for trail access in 1985 ranged from 59 percent on Saturdays to 54 percent on weekdays. A 22 percent decline from 1985 to 2000 in this proportion for Saturday respondents and a 50 percent decline for Tuesday respondents (Moritz, 2005b) meshes with the postulate that some earlier users of the Burke-Gilman and Sammamish River Trails may by 2000 have been taking advantage of new alternative facilities. (Information on access mode is not available for 1980 or 2005.) Whatever the reasons for trail use variations, the average May 7:00 AM to 7:00 PM trail volumes in 2005 stood at over 2,000 for both Tuesday and Saturday trail traffic, indicating higher overall use than any other survey year except 1995 (see Table 16-17). Table 16-18 summarizes trip purpose findings from the Tuesday counts/surveys. Because the 2005 data were obtained only at four intercept stations, both all-station and four-station results are shown to the extent available. Not only were the reported all-station and four-station surveys taken using different approaches, the all-station locations were more heavily weighted toward areas of denser urbanization, likely the major factor in the higher work/school commute percentages reported for the all-station surveys. Although this circumstance somewhat clouds the results, there is a fairly consistent and substantive upward trend in the proportion commuting within each over- lapping set of time-series data (all-station and four-station). Clearly weekday utilitarian use of the trails was increasing as a proportion of total use, with a corresponding percentage decline in recreational/ exercise use (Puget Sound Regional Council, 2000, Moritz, 2005a and b). In absolute terms, however, it appears that the number of recreational/exercise users at first grew substantially and then stabi- lized, with the number of Tuesday recreational/exercise users in 2005 being likely about as high as any survey year except 1995. 16-98 24 Postulates offered in the following two series of bullets are those of the Handbook authors except in the case of the individual bullets containing a citation.

The Saturday surveys from 1985 to 2000 show trends fairly consistent with the weekday survey respon- dent trip purpose patterns. The recreation/exercise proportion declined from 98 to 79 percent. Work/ school commuting increased from 2 to 12 percent, shopping increased from nil to 6 percent, and other/multiple responses increased from nil to 3 percent (Moritz, 2005b). One can only speculate as to reasons for the shift over time toward trail use for commuting: • Closure of the “Missing Link” in 1993, extension in 1993/94 of the Burke-Gilman component into Freemont and Ballard (nearer central Seattle), and interconnection with an expanding net- work of King County NMT facilities, may together have made the Burke-Gilman/Sammamish River Trails progressively more useful for commuting and other utilitarian uses. • NMT commute-mode choice-making may be a more “sticky” decision process than recreational or exercise decisions, with change not occurring quickly, but perhaps involving evolution of work- place culture—with choices to cycle or walk/run to work by avant-garde employees being gradu- ally followed by fellow workers and progressively receiving more employer support. • NMT congestion on the trail may be dampening casual recreational use more than use for commuting. • Sharp growth in student population and employment at the University of Washington, en route on the Burke-Gilman Trail, may be showing up in trail commuting increases.25 16-99 Table 16-18 Seattle Area Burke-Gilman/Sammamish Trail Average Tuesday Trip Purpose Percentages Over Time, 1980–2005, Late May 1985 1990 1995 2000 2005 Trip Purpose All-Sta. Survey All-Sta. Survey All-Sta. Survey 4-Sta. Verbal All-Sta. Survey 4-Sta. Verbal 4-Sta. Verbal Work/School Commute 10% 44% 47% 28% 48% 26% 32% Recreation/Exercise 90% 53% 48% 67% 45% 70% 58% Shopping 1% 1% 3% 4% Other/Multiple Responses 0% 2% 2% 3% Other/Shopping 3% 1% 2% Number of Respondents 968 1,905 1,611 6,060 1,245 6,103 7,663 Sample Size 150 1,905 1,611 6,060 1,245 6,103 7,663 Notes: The all-station surveys, with six to eight intercept survey stations, had all but two survey stations on the more urban Burke-Gilman Trail. Survey hours were 7:00 AM to 7:00 PM. The all-station surveys utilized a mail-back survey approach. The four-station approach instead sought a verbal trip purpose identification from passing trail users. Survey hours were 6:00 AM to 7:00 PM. Sources: Moritz (2005a and b), with calculation of four-station percentages, “no response” excluded, by the Handbook authors. 25 Between 1991 and 2005, for example, student enrollment was up 40 percent and faculty and staff totals were up almost 30 percent (see “University of Washington’s U-PASS Program—Seattle, Washington” under “Case Studies” in Chapter 19, “Employer and Institutional TDM Strategies”).

The Burke-Gilman/Sammamish River Trails surveys have identified quite high trip lengths. May 1990 reported Tuesday and Saturday on-trail trip distance medians were 5 and 14 miles, respectively (Moritz, 1995). One analysis of the May 1990 survey calculated that the prevalence of commute trips among bicyclists within the trail traffic was over five times the proportion of commute trips among intercepted walkers (Guttenplan and Patten, 1995). Non-bike commuters had a median Tuesday trip length of 2 miles, while bike commuters exhibited a median trip length of 4 miles (Moritz, 1995).26 Other Path Information. The Pinellas Trail (4th entry in Table 16-16) provides a further example of significant path use for utilitarian transportation. In 1993 over 1/3 of weekday use was found to be for work, school, or shopping access. The trail provides a direct route toward central St. Petersburg, Florida, and passes several major employment sites and 5 schools, in addition to its recreational role of connecting numerous parks and natural areas. The trail cross-section, except in constricted areas, provides a 10-foot paved way for bicycles and a separated 5-foot paved way for pedestrians. A num- ber of grade separations, only one of which was part of the former railroad infrastructure, cross busy arterials (Guttenplan and Patten, 1995). Researchers in an area of variable path quality have observed that “commuter cyclists for the most part use only higher quality paths” (Aultman-Hall, Hall, and Baetz, 1997). Apparently the Pinellas Trail provides the required quality in addition to meeting the utilitarian trip requirement for directness and linkage of residential areas with professional employ- ment, schools, and other activity sites. This requirement is also met by two Washington, DC, area trails surveyed in September 1993: the inner Rock Creek trail in the District of Columbia itself, with 67 percent transportation use, and the W&OD Trail of Northern Virginia, with 51 percent. Two outlying more rural trails outside of Baltimore, the Northern Central and Baltimore and Annapolis rail trails, were found to be primar- ily used for recreation and exercise (Guttenplan and Patten, 1995). This pattern of higher transporta- tion use in urban locations parallels that found in Indiana. More information on W&OD trail use and economics, with greater emphasis on weekend activity, is located in the “Related Information and Impacts” section under “Economic and Equity Impacts”—“Commerce Impacts of Off-Road Paths.” The 5th and 6th research entries in Table 16-16 reflect a quite different analytical approach, provid- ing before-and-after perspectives by examining U.S. Census journey-to-work trip data for the Census years prior and subsequent to facility implementation. These key studies were fully described in the preceding “Bicycle Lanes and Routes” subsection under “Bicycle Lane Implementation”— “Longitudinal Commute Mode Share Research” (see especially Table 16-12) but apply with equal importance to shared use paths. The journey-to-work mode share impact findings for paths are sum- marized here along with contextual factors that may help explain why the new paths in some urban areas were more successful in attracting work commute trips than in other areas: • The Minneapolis-St. Paul research identified increases in work-commute bicycle shares averaging 1.38 percentage points in the commutersheds of both new bicycle lanes and new shared use trails. The corresponding bicycling percentage increase for the trail corridors of 43 percent was numer- ically less than the increase for bicycle lane corridors only because trail corridor shares were higher to start with. The four trail commutersheds had a simple average before-trails bicycle commute mode share of 4.02 percent, and a 5.40 percent average after the decade of trails implementation (Barnes, Thompson, and Krizek, 2006). The studied Twin Cities trails are in attractive locations for 16-100 26 Since these data are from multiple on-trail intercept locations, as contrasted to the Indiana surveys discussed above, they represent a snapshot of use characteristics of trail traffic at points along the facility, not a report- ing of average trail user characteristics. Average trail users would be making somewhat shorter trips (inter- cepted less), and thus likely would exhibit more use of the walk mode.

commuting, are oriented toward downtown and university employment areas, and are integrated into a well-establish network (Cleaveland and Douma, 2009). • In Austin, TX, the 1990–2000 bicycle work-commute mode-share gain along new trails was 0.88 percentage points, representing a 24 percent increase, reaching 3.52 percent. The studied trail locations are close-in to the downtown, and other favorable characteristics noted for the Twin Cities are present in Austin as well (Cleaveland and Douma, 2009). • Results for Colorado Springs were not statistically significant. The 1990–2000 gain in bike work commute mode-share was 6 percent overall along new trails. The north trail did well, with a tripling of bike mode share, but the south trail dragged the average down. Colorado Springs lacked an overall bike facility network (Cleaveland and Douma, 2009). • In Madison, WI, there was a statistically insignificant decrease in work-trip bicycle shares where new paths were implemented of −2 percent. The bicycle share was already very high, almost 6 per- cent, and the paths involved largely parallel existing bicycle infrastructure in a fully developed network. The paths provide corridor cyclists more options (Cleaveland and Douma, 2009), and may well be important for non-commuter user groups such as children and recreational cyclists. • Bicycle work-commute mode share also decreased along new paths in Salt Lake City, by a sta- tistically insignificant 24 percent. The path segments were located in low-density areas periph- eral to the core of the city. The researchers noted that both path and on-street bike facility implementation in the 1990–2000 decade was poorly publicized, and hypothesized that the improvements were hardly noticed in amongst preparations for the 2002 Winter Olympics (Cleaveland and Douma, 2009). • Areas along new trails in the Orlando region showed a statistically insignificant 21 percent decline in bicycle commute share, actually less than for Orange County, Florida, as a whole. These new trails were mainly in low density areas far from downtown, with poor connections to other bicycle facilities. As in the other urban areas examined in this study, the research approach did not allow examination of the value of these trails for walking, non-work utilitarian travel, or recreation/exercise (Cleaveland and Douma, 2009). The Ozaukee Interurban Trail, built primarily along the former electric interurban railway that once linked towns between Milwaukee and Sheboygan, Wisconsin, has not been researched in depth. Two surveys, however, provide information on what proportion of county residents report having used the 30-mile north-to-south-border trail. As can be seen from Table 16-16, 7th entry, 1 in 4 residents reported use of the trail when queried roughly 10 months after full opening (with 1 detour). Some 30 months after opening (still with 1 detour) the proportion was found to be just over 1 in 2 (Pedestrian and Bicycle Information Center, 2010, Roback, 2004), representing about 45,000 of the residents of Ozaukee County. The Hudson River Trail in New York City offers an example that is both instructive and tricky to inter- pret. The trail provides a broad promenade most of the length of Manhattan. In 2002 permanent trail replaced an interim section between West 12th and West 55th that varied between 5 and 10 feet wide, entailed 90-degree turns, and was bordered by chain link fencing and concrete barriers. “After” condi- tion peak and midday periods 6-hour NMT volumes ranged from 1,248 to 2,056 on weekdays, depend- ing on location, and 6-hour weekend volumes ranged from 3,474 to 4,498. The presence of a counted interim facility in the “before” condition gave a basis for computing growth. Volumes doubled to tripled at West 17th and quadrupled at West 34th (Chaney, 2005). Numerical details are provided in the 8th entry of Table 16-16. System interconnection effects introduce a confounding factor. The usage 16-101

increases reflect not only the direct effect of trail-improvement but also the interconnectivity effect of providing a better linkage between the preexisting fully developed sections. The 9th Table 16-16 entry simply serves to illustrate that “build it and they will come” is not always a certainty. Preliminary circa 2004 counts on the Doylestown Bike & Hike path found an average of 10 users per day on various parts of the facility (Pedestrian and Bicycle Information Center, 2010). Lacking analysis of reasons for the low usage, it may be speculated that one section next to a busy highway—shoehorned between a high concrete traffic barrier and a chain link fence—is too unattractive, or that underlying travel demand may simply be low. Alternatively, path traffic may have been slow in developing. The 10th and final table entry summarizes a review of five additional paths from the perspective of impacts on bicycling. Results are nearly a tie between studies finding no impact (2) and studies find- ing an increase in cycling (3) (Pucher, Dill, and Handy, 2010). Two of these studies are independently covered within this chapter. The case involving marketing is addressed under “Walking/Bicycling Promotion and Information”—“Transportation Mode Shift Promotions”—“Promotion of New Options” and finds cycling growth in the context of small numbers of bicyclists and a trail with some physical drawbacks (Merom et al., 2003). The other is 1 of the 2 studies finding no discernible impact on amount of cycling. The study in question examined a short paved pathway—1 mile in length—not part of a path network and not oriented to any notable destinations. The location is a residential area of West Valley City, Utah, a Salt Lake City suburb. The canal-side trail does feature sidewalk tie-ins, providing a 2.5-mile loop, and is in fact used primarily by pedestrians (71 percent) and joggers (13 percent), with relatively few cyclists (16 percent). A before-and-after study of trail-neighbor activity levels indicated that open- ing and establishment of the trail did not lead to counteraction of a downward trend in walking and bicycling activity. Trail users were mostly from outside of the trail neighborhood study area. The trail offered a new exercise route option, but for most, not an occasion to change activity mode. Among trail users interviewed in an intercept survey, 87 percent had previously engaged in their chosen activ- ity before opening of the trail. On the other hand, almost 94 percent walked, biked, or jogged to reach the trail itself. Average distances from home to the trail were 1.2 miles for walkers, 1.8 miles for bicy- clists, and 4.6 miles (sic) for joggers (Burbidge and Goulias, 2009). Five more studies involving multi-use, off-road paths, all focusing on physical activity effects, are examined under “Public Health Issues and Relationships” in the “Related Information and Impacts” section. (See Table 16-123 within the discussion “Health Benefits for Adults of Enhanced NMT Systems and Policies”—“Adult Physical Health Effects of Non-Motorized Transportation Features.”) Two of the studies, the 6th and 10th Table 16-123 entries, found additional physical activity for per- sons with good trail access in North Carolina and suburban Boston, Massachusetts. The 7th table entry simply establishes that about 2/3 of Minneapolis area trail users meet or exceed minimum physical activity guidelines through use of the trails. The 8th Table 16-123 entry summarizes a before and after study finding that two new West Virginia trails attracted both habitual and new exercisers, and that physical activity increases were seen for both groups. Only in the study entered as the 9th table entry, which examined a trail extension and suffered from unintended survey timing (2 months after trail extension opening), were findings inconclusive (see Table 16-123 for sources). Shared Use Path System Coverage The role of shared use path systems, as with other NMT facility networks, has been examined not only on the basis of facility-level effects but also with cross-sectional and comparative studies, both 16-102

aggregate and disaggregate. There have been four nationwide aggregate analyses in the 20 years from 1992 to 2011 that have investigated the impact of bike lane and path system extent on bicy- cling rates, using U.S. Census journey-to-work or comparable ACS data. These were detailed in the previous “Bicycle Lanes and Routes” subsection under “Bicycle Lane System Coverage.” The first two used combined bike lane and path measures of system extent, finding positive associa- tions of system coverage with bicycling to work. The third, on the basis of 42 large cities, found positive associations for bike lane system extent and path system extent individually, but strongly so only in the case of bike lanes. The newest and most comprehensive, covering 90 of the 100 largest U.S. cities, utilized three alter- native forms of regression analysis. Each form treated bike lane extent and path extent (relative to population) as separate explanatory variables along with proportion of students in the population; urban characteristics; region (west, southeast, etc.); weather; and (in some formulations) safety. Good statistical fits were obtained, with R2 values of 0.57 to 0.67. The relative role of bike lanes and paths varied among model formulations, but not with statistically significant differences. Both types of bicycle facilities had positive associations of system extent with bicycle commuting, with estimated elasticities of about +0.25 in each case, inelastic but statistically and programmatically significant (Buehler and Pucher, 2011). Key limitations of the national studies include lack of examination of effects on walking, irrelevant for bike lanes but important for paths, and lack of findings concerning both effect on non-work utilitarian travel and use for recreation and exercise. Table 16-19 summarizes six path system cov- erage or proximity studies that help fill some of these gaps. 16-103

16-104 Table 16-19 Summary of Research Findings on the Relationships of Shared Use Path Proximity and Prevalence with Walking and Cycling Activity Study (Date) Process (Limitations) Key Findings 1. Moudon et al. (2007) (see “Ped…cycle Friendly Neigh- borhoods” for more information) Cross-sectional analysis of walking activity, socio-demographics, attitudes, and objectively measured environmental variables covering 608 adults in King County, WA. (Cycling activity not examined.) No significant relationship found between trail proximity and overall amount of walking activity, but additional analysis suggested that proximity increased the likelihood of choosing trails for walking routes. 2. Moudon et al. (2005) (see “Ped…cycle Friendly Neigh- borhoods” for more information) Similar to Moudon et al. (2007) but focused on cycling (at least once a week versus less), with addition of perceived environmental variables. (Some evidence, for 1/3 of cyclists, of neighborhood “self-selection” for recreational facility accessibility.) A moderately strong relationship was found between measured trail proximity and overall cycling activity. Increased likelihood of trail use for recreation/exercise and use of the bicycle mode for trail access was also identified with trail proximity. 3. Krizek and Johnson (2006) Cross-sectional analysis of effects of proximity to bicycle facilities, using Minneapolis and St. Paul component of the year 2000 regional survey of weekday household travel. (Only 86 sampled trip makers, spread across the two cities, reported bike trips.) Found a partially significant positive relationship between weekday bicycle trips and bike lane proximity (strongest before considering demographics), but no significant weekday relationship with off-road trail proximity or prox- imity of both facility types together. 4. Duncan and Mummery – 2005 as summarized in Saelens and Handy (2008) Analyzed survey in Rockhampton, Queensland, Australia, of incidence of recreational walking vis-à-vis perceived and objective measures. (Few measures found significant.) Higher likelihood of recreational walking during past week with home location <0.4 km. (<1/4 mile) from a footpath. Frequent walking found to be correlated with poor perceptions of footpath conditions. 5. Brownson et al. – 2000 as summarized per SR 282 Conducted cross-sectional mail survey and logistic regression analysis for a rural community sample. (Apparently most descriptors were self-reported, with prior activity not investigated.) Asphalt surface increased incidence and frequency of use. Greater trail length and location within 5 miles increased frequency. Trail length of 1/4 to 1/2 miles was associated with greater incidence of use. 6. Lansing, Mar- ans, and Zehner – 1970 as reported in Nelson and Allen (1997) U.S. Bureau of Public Roads examined schoolchild bicycle use in new communities with differing numbers of bicycle paths. (No information on methods.) About 22% of schoolchildren walked or biked to school in new communities with no bicycle path, compared to 29% for 1 path and 49% for 2. [Paths may be a surrogate for good NMT design.] Note: Where substantial additional information on individual studies is provided in text and tables or figures, this is noted — and the location within the chapter is given — in the first column. Sources: As indicated in the first column. The notation “SR 282” is shorthand for Committee on Physical Activity, Health, Transportation, and Land Use (2005) together with Handy (2004).

The studies in Table 16-19 represent regional and community cross-sectional research, along with a comparative study, focused on learning about path system coverage effects from analysis of systems in place. The studies range in complexity from detailed statistical evaluations to a multi-community comparison of schoolchild travel modes conducted by the Bureau of Public Roads (BPR), predeces- sor to the U.S. Department of Transportation. Of the six tabulated studies, four reported significant findings of higher NMT activity with more or closer paths. These higher levels of activity, reported as statistically significant (or being obvi- ously so), involved cycling in Seattle (2nd table entry); walking in Rockhampton, Queensland, Australia (4th entry); and both walking and cycling in a U.S. rural community (5th entry) and among U.S. schoolchildren in new communities (6th entry). Of the two studies failing to identify statistically significant NMT activity, one focused on walking (Seattle, 1st table entry) and one on cycling (Minneapolis-St. Paul, 4th entry). The Seattle pedestrian study did find more trail-walking in households proximate to trails, but no statistically significant indication that it represented addi- tional walking activity overall (see Table 16-19 for sources). The research effort that failed to identify more cycling activity in proximity to Minneapolis-St. Paul trails of all types (Krizek and Johnson, 2006, 3rd Table 16-19 entry) stands in contrast to the facility- specific time-series analysis of Minneapolis-St. Paul work purpose cycling included as the 5th entry in Table 16-16 (Barnes, Thompson, and Krizek, 2006). The all-trails cross-sectional analysis may be an example of a manifestation noted in the “Analytical Considerations” discussion of the “Overview and Summary,” wherein less analytically robust research has been found less likely to show significant relationship of NMT activity to a stimulus (Ogilvie et al., 2007). It attempted to draw inferences from a small sample scattered across a broad geographic area. The trail-specific research in Table 16-16 (5th entry), which found significant shifts to bicycle commuting with the introduction of trails, had the advantage of a larger data set (the U.S. Census) and could use it essentially as a clus- ter sample, with comparative data from non-commutershed areas. Only the BPR study, the 6th entry in Table 16-19, addressed mode choice outright. The higher schoolchild NMT mode shares reported in the presence of shared use paths has, for purposes of comparison with the other research, been taken as a surrogate for more walking and cycling activ- ity. New-community schools with two paths (and perhaps better NMT design overall) had over twice the proportion of walking and bicycling to school as schools with none. Schools with one path were in-between in NMT share (see Table 16-19 for source). Not quite fitting in with the more conventional studies of Table 16-19, but instructive in its own right, is unusual research on the effects of NMT infrastructure investments in the Baltimore and Sacramento regions. Regional travel survey utilitarian walking and bicycling shares for 1993 and 1991, respectively, were determined by traffic analysis zone (TAZ). These early 1990s shares were applied to 2001 and 2000 travel by all modes, determined from correspondingly newer regional travel surveys. By this means, trend estimates of walking and bicycling trips, by TAZ, for 2001/2000 were calculated. A negative bino- mial regression was then constructed to model the actual (observed) 2001/2000 utilitarian walking and bicycling by TAZ as a function of (1) the trend estimates, (2) income changes, (3) density changes, and (4) NMT infrastructure investment during the intervening period. Research difficulties included NMT spending categories that could not be isolated and the need to rely on funding timing as a surrogate for construction timing. In the resulting Baltimore and Sacramento walk and bike trip models, income and density changes were found either to have insignificant effect or to operate in the expected direction. Income increases, where significant, were associated with walking/bicycling decline and—in the one instance of density significance—density increase was associated with increased bicycling activity in Sacramento. 16-105

Expenditures on trails showed a positive, although not statistically significant relationship with walk- ing in both cities. A positive relationship with bicycling in Baltimore was statistically significant for both 1/4 and 1/2 mile buffers along trails that were financed. In Sacramento, the relationship with bicy- cling for spending on trails was both insignificant and of an illogical sign, although there was a signif- icant positive relationship for bike lane expenditures. (Bike lane expenditures could not be examined in Baltimore.) These findings pertain to utilitarian walking and bicycling only, and the researchers point out that trail expenditures could be having significant effects on recreational/exercise activity, not addressed in the study (Ewing, Handy, and McCann, 2010). Pedestrian/Bicycle Systems and Interconnections Lack of NMT system interconnectivity forces detours on pedestrians and bicyclists and throws up barriers in their way. The inherently slower speeds of walking and cycling, relative to driving, give heightened sensitivity to route circuitry in travel between places. There is evidence pedestrians and bicyclists appreciate and respond to direct connections, and that barriers to direct pedestrian and bicycle travel deter use of active transportation. Unfortunately, cities and suburbs are full of unconnected links and physical barriers such as sidewalks that end abruptly, cul-de-sacs and dead end streets, bike paths that go nowhere in particular, and streams, rivers, busy highways, and expressways without suitable crossings (David Evans and Associates, 1992). Pertinent NMT research ranges from the common finding that minimization of time and distance is a primary objective of utilitarian walkers and cyclists, to the specific outcomes of creating con- nections in practice that are presented in this subsection. The totality of this research continues to underscore the critical role of good interconnections in encouraging choice of non-motorized modes of travel (Kuzmyak et al., 2011). Studies are even beginning to show that route directness, as compared to mere nearness, is among walking inducements (Moudon et al., 2007). Concepts useful to appreciating the role of systems and interconnections are presented in the “Underlying Traveler Response Factors” section, within the “Environmental Factors” discussion. There, in the “Systems Environment” subtopic, the relevance of accessibility and the contribution to accessibility of connectivity are developed. The “Surroundings Environment” subtopic exam- ines the influence of system link quality (“Facility Compatibility Measures”) and “Ambiance,” but with the caveat that quality of individual links can make little contribution if the links are not well joined together. System interconnections make relatively large contributions in terms of completing the pedestrian and bicycle network with fairly short physical distance linkages. Examples include pedestrian and bicycle (ped-bike) bridges across major barriers including freeways, railroads, ravines, and rivers; short connections eliminating “missing links;” and cut-throughs allowing pedestrians and bicy- cles to pass directly through discontinuous street networks, such as between ends of cul-de-sacs, through large blocks, and across traffic-calming vehicle diverters. Bridges and bridge improvements obviously represent larger capital expenditures than short seg- ments of walkway or the provision of cut-throughs. Perhaps for this reason, the demand response studies of non-bridge interconnection projects have not been as widely reported. As a result, the indi- vidual interconnections portion of this subsection necessarily focuses mainly on the response to more expensive bridge provision and upgrading projects. This circumstance should not be taken to infer that less capital-intensive connections are of little importance. Information on path gap closures, facil- ity extensions, and “Interconnections of Modest Scale” will be found toward the end of the “River Bridges and Other Linkages” discussion. 16-106

Overall Systems and System Expansions Recognizing the contribution of all interconnected NMT system elements, overall system effects— enabled in part by connectivity measures—are reviewed first. This review is accomplished by recapping and adding to key overall systems studies presented elsewhere within this chapter, either in preceding facility impact or in upcoming policy impact discussions. Key studies covered are shown in Table 16-20. 16-107 Table 16-20 Summary of Research Findings on Relationships between Pedestrian/Bicycle Facility Density/Interconnectivity and Non-Motorized Travel Activity Study (Date) Process (Limitations) Key Findings 1. Goldsmith (1992), Nelson and Allen (1997), Dill and Carr (2003), Buehler and Pucher (2011) (see “Bicycle Lanes and Routes” and Table 16-14 for more information) One descriptive analysis and three cross-sectional analyses relating various measures of bicycle facility density (bike lanes and off-road paths only) to citywide bicycle mode shares for the U.S. journey to work. (City-level aggregation, work trips only, causality not established, no explicit measure of connectedness.) • Bikeway/arterial ratio < 0.035:1 asso- ciated with 1/3 the bike share of cities with ratio > 0.035:1 (“university towns” omitted). • Each additional bikeway mile per 100,000 population associated with 0.069% more commuter cycling. • Each added bike-lane mile per sq. mi. associated with ± 1% more commuter cycling. • 10% more of either bike lanes or bike paths associated with 2.5%-2.6% more bicycle commuters. 2. Pinjari, Bhat, and Hensher (2008) (see “Underlying Traveler Response Factors” — “Choice of Neighborhood…” for more) Modeled residential location and activity time-use choices of 2,793 regional survey sample households in Alameda County, San Francisco East Bay, controlling for residential sorting (a.k.a. self-selection). Zonal- level environment variables inclu- ded bike facility densities. (Bicycle ownership levels treated as givens.) Even after controlling for residential self-selection (very significant), good bicycle facility densities were found to be associated with more physical activ- ity such as walking, cycling, and jog- ging. The model predicted that a ten- times increase in bicycle facility density would produce an overall 17% increase in time of recreational facility use. 3. Birk and Geller (2006) (see also “River Bridges…” and “[NMT] Policies and Programs”) Portland, OR, bikeways increased from 78 miles in 1991 to 256 miles in 2004, a 228% increase. Bike facilities were improved or added on 4 central area bridges. (Bicycle data only, results cannot be separated from overall auto use reduction policy effects.) An extrapolation from bridge counts suggests a 210% increase in bike trips between 1991 and 2004, eclipsing population increases. Over the 1990- 2000 decade, the citywide bike mode share for work purpose trips increased from about 1% to 3%. (See cross-refer- enced discussions for ca. 2008 data.) 4. Queensland Transport – 2007 via Davies (2008) (see this section and “NMT Policies and Programs” for more) With development starting ca. 1985, Brisbane’s shared use path system extended 7-1/2 miles from the CBD in 1 corridor by 1995 and 3 corridors by 2000, with a major new bridge in 2001 (see Fig. 16-3). (Investigation based on Australian census was limited to journey-to-work trips.) For travel to the CBD and CBD fringe from surrounding areas, walk to work shares increased almost threefold and bike shares sixfold from 1986 to 2006, reaching 17.4% walk and 3.0% bike (see Figs. 16-2 and 16-3). Housing expansion in the core area may well have contributed to increased walking. Note: Where substantial additional information on individual studies is provided in text and tables or figures, this is noted — and the location within the chapter is given — in the first column. Sources: As indicated in the first column.

Facility Density and Connectivity. The 1st entry in Table 16-20 summarizes four key studies on the relationship between extent of bicycle system coverage and choice of the bicycle mode for commut- ing to work. These studies are individually described in Table 16-14 of the “Bicycle Lanes and Routes” subsection, where they are first introduced. The accompanying text notes limitations of these studies with regard to their aggregate data, focus on work trips only, and lack of demonstration of causality. To this must be added their consideration only of bicycling and their use of facility-quantity ratios as bicycle supply measures. While the supply measures are robust as far as they go, they do not specifi- cally quantify the degree of system connectivity of interest here. All four studies looked at the contribution of both bicycle lanes and off-street paths and found pos- itive association of system extent with bicycle commuting. The newest study found the two types of facilities essentially equal in their importance to the commuter cyclist. Despite their acknowl- edged limitations, the four studies make a substantial case that system extent is a significant and positive factor in the decision to bicycle by the studied population, namely, adults choosing between bicycling and other travel modes for the work commute (Goldsmith, 1992, Nelson and Allen, 1997, Dill and Carr, 2003, Buehler and Pucher, 2011). The 2nd entry in Table 16-20 highlights a study that, while not national, took a broader analytical perspective by examining participation in all types of active transportation—including walking, cycling, and jogging—for all purposes including recreation. Working with a large survey sample for Alameda County, inclusive of Berkeley, Oakland, and other cities of the San Francisco East Bay to the south and east, it first of all identified residential self-selection effects. Individuals with higher bicycle ownership and interest in physical activity were found to locate more in neighbor- hoods with greater density of bicycle facilities. Then, with this phenomenon separately accounted for, the research proceeded to produce estimates that increasing facility density is linked with modest positive changes in individual participation in active transportation (Pinjari, Bhat, and Hensher, 2008). NMT system studies that more explicitly address system connectivity are found in the “Pedestrian/ Bicycle Friendly Neighborhoods” subsection, primarily within the “Design” subtopic, but also in the “Walk Elasticities for Land Use and Site Design Parameters” discussion and tabulation (see Table 16-42). Of particular interest under “Design” is the 8th entry in Table 16-40, covering a study notable for focusing exclusively—other than taking demographics into account—on the degree to which the street system, and by inference the pedestrian system, is tightly interconnected. The study tested a composite walkability score, assembled from three different measures of system connectivity, and related it to walking activity. A higher incidence of walking in counties with higher scores was found in this connectivity research (Saelens and Handy, 2008). System Development Integral with Policy/Program Realization. Two system expansions are included in Table 16-20 from among those examined as part of the upcoming “NMT Policies and Programs” subsection. They are further expanded upon there under “New World Program Examples”—“Portland, Oregon” and “Brisbane, Australia.” These two examples are selected for highlighting here because of their noteworthy illustration of NMT interconnected-system effects. Both are also separately examined under “River Bridges and Other Linkages” because of notable river bridge program components. Portland’s NMT system expansion program has been heavily, but not exclusively, focused on bicycle facilities. Monitoring has been primarily on the basis of bicycling data. NMT system implementa- tion has run in parallel with the more recent stages of long-established policies designed to dampen auto use, and gained momentum starting in the late 1980s (City of Portland, 2004). Miles of bikeways increased from 78 miles in 1991 to 256 miles in 2004, a 228 percent increase. Starting 16-108

in 1992, major pedestrian and bicycle improvements were made to four key central area bridges, as covered below in the “River Bridges and Other Linkages” discussion. Bikeways in Portland include bike lanes, bicycle boulevards, and shared-use paths. These improvements were accompanied by an estimated 210 percent increase in bicycle trips from 1991 to 2004, as extrapolated from river crossing counts. This extrapolation appears to be corrob- orated by citywide U.S. Census commute trip data. From 1990 to 2000 the overall bicycle share of work purpose trips increased threefold, from approximately 1 percent to 3 percent, with larger increases in the dense, flat, neighborhoods of the inner city. In the “NMT Policies and Programs” subsection, Figure 16-7 maps both the distribution of the bicycle commute mode share increases and the growth of the bikeway network (Birk and Geller, 2006). The accompanying “New World Program Examples”—“Portland, Oregon” discussion provides system extent and bridge count updates through 2009. Brisbane’s NMT system expansion has not only addressed both walking and bicycling but has also been monitored on the basis of both types of use. Figures 16-2 and 16-3 illustrate the response over time as expressed in walking and bicycling shares for trips to the CBD. Brisbane’s current system of off-road, shared use paths and on-road bicycle facilities was begun in the mid-1980s, coincident with the first mode share plot in each figure. “Bikeways” had been constructed by ca. 1995 in one radial corridor and by ca. 2000 in three corridors, as mapped in Figure 16-3.27 Shortly thereafter, the total on- and off-road bikeway network totaled over 550 km. (342 miles), complementing the more than 3,950 km. (2,454 miles) of sidewalks and other footpaths in the city. By 2008–09, the bikeway network totaled more than 760 km. (472 miles), consisting of 54 percent off-road paths and 46 percent on-road bicycle facilities (Queensland Transport, 2007, Brisbane City Council, 2009a and b). 16-109 27 Brisbane includes its shared use paths under the broad term “bikeway,” along with on-road bicycle facili- ties. There are apparently a few sections of off-road bikeways that are bicycle-only.

16-110 Figure 16-2 Work-purpose walk mode share to Brisbane CBD and CBD fringe, 1986–2006. Source: Modelling, Data and Analysis Centre, Transport and Main Roads, Queensland Government (formerly Queensland Transport), Australia [2007] via Davies (2008).

16-111 Figure 16-3 Work-purpose bicycle mode share to Brisbane CBD and CBD fringe, 1986–2006. Source: Modelling, Data and Analysis Centre, Transport and Main Roads, Queensland Government (formerly Queensland Transport), Australia [2007] via Davies (2008).

Figures 16-2 and 16-3 depict the walk mode shares and bicycle mode shares, respectively, for jour- ney-to-work trips from individual analysis districts to the Brisbane CBD and CBD fringe in 1986, 1996, and 2006. Overall shares to the CBD and fringe for these and intermediate years are tabu- lated, and additional interpretation is provided, in the “NMT Policies and Programs” subsection under “New World Program Examples”—“Brisbane, Australia” (see in Table 16-45). Work trip walk shares to the CBD from within roughly a 6 km. (3-3/4 mile) radius have increased from 5.9 to 17.4 percent over the 20 years, likely with the aid of additional downtown housing. Bike work trip shares to the CBD from within roughly a 12 km. (7-1/2 mile) radius have increased from 0.5 to 3.0 percent (Queensland Transport, 2007, Davies, 2008). The Goodwill Bridge across the Brisbane River is a key component of the NMT network. It is shown in both Figures 16-2 and 16-3. Bridge use response data are given in the “River Bridges and Other Linkages” discussion. River Bridges and Other Linkages Table 16-21 covers the primary sources of findings for river bridge NMT improvements, new ped-bike bridges, and other new linkages within pedestrian and bicycle systems. As previously indicated, the more capital intensive a project, the more likely it is that traveler response data is available. Despite that, more modest projects may well be of high importance in their own contexts. Willamette River Bridges, Portland, Oregon. A central, critical element in Portland’s bicycle facil- ity system development has been improvements to the Willamette River Bridges (1st entry, Table 16-21). The geography and overall context involved are described under “NMT Policies and Programs”—“New World Program Examples”—“Portland, Oregon.” The Willamette River sepa- rates the historic core of downtown Portland on the west from the Lloyd District on the east, in many ways an expansion of downtown functions, and extensive surrounding traditional residen- tial areas. A number of bridges span the river, but for many years, the accommodation of pedes- trians and bicycles was severely constrained. Several of the bridges, starting in the 1990’s, have undergone renovations or improvements designed primarily for the benefit of bicyclists and pedestrians. Emphasis has been placed not only on upgrading the on-bridge accommodations, but also on creating pedestrian- and bicycle-friendly approaches and expanding the feeder network of off-road trails and on-street bike boulevards and lanes. Table 16-22 and its accompanying notes list the on-structure improvements during a 12-year span and provide additional context by giving mileage by year of Portland’s bicycle facilities. 16-112

16-113 Table 16-21 Summary of Studies on the Travel Effects of Providing Pedestrian/Bicycle Bridges and Other Linkages Between and Within Ped/Bike Systems Study (Date) Process (Limitations) Key Findings 1. Birk and Geller (2006), Birk (2003) (see this section and “NMT Policies and Programs” for more) Descriptive analysis of Portland, OR, Willamette River bridge bike count changes in response to 1993 painting of bike lanes on Burnside Br., 1998 sidewalk resurfacing on Broadway Br., 1999 shared-use sidewalk widening on Hawthorne Br., and 2001 opening of Steel Br. lower-deck ped-bike crossing. (Bike counts only, extrapolated from peak, outcomes confound by multimodal program.) The 4-bridge total bicycle count, up in 1992–93, dipped in 1995, then climbed consistently upward, on through 2004, up 211% in 13 years. Bike lane effects do not stand out in Burnside Br. yearly counts. Broadway Br. count growth appears most influenced by feeder network improvements. The projects on Hawthorne and Steel Bridges were accompanied by 45% and 361% bridge- specific 2-year bike count increases. 2. Abrahams (2002) a (see this section and “Travel Behavior Shifts” under “Related Info…” for more) Surveyed weekday peak period users of Goodwill Bridge, a ped-bike facility over the Brisbane River close to downtown Brisbane, Australia, 8 months after bridge opening. A descriptive analysis was prepared. (No formal count in parallel with survey for survey control.) Queensland Government 2-week daily counts 5 months after opening ranged, excluding a rain day, from 4,726 (25% cyclists) on a Saturday to 10,854 (18% cyclists) on a Tuesday. Of ped-bike users, 40% diverted from a less-safe crossing. Another 42% made complex, often multimodal, mode shifts. 3. RTC and APBP (1998), Historical Marker Database (2010) A former railroad bridge connecting downtown fringes of Lewiston and Auburn, Maine, via a former textile mill district was restored for NMT use. (Findings limited to total use.) Three years after opening to bicyclists and pedestrians this facility over the Androscoggin River was in use by over 350 people a day. It is part of a histori- cal walk but not an overall trail system. 4. Lipton (1979), Zehnpfenning et al. (1993), Bicycle Federation of America (1993) Users of the Greenway Bridge across the Willamette River in Eugene, OR, were surveyed 2-3 months after 1978 opening. (Counter failures. Relied on behavior-change perceptions.) Of surveyed bicyclists, 14–28% were cycling because of the ped-bike bridge, and 30% thought it as quick or quicker to cycle given the bridge. Summer weekday 1982 count of 1,100 cyclists. 5. UK Department for Transport – 2004 as summar- ized in Booz Allen Hamilton (2006) The 2001 ped-bike Millennium Br., in York, England, links traffic-free path sections and walking/cycling routes across the River Ouse. (Route expansion clouds interpretation.) Use of routes on both banks grew from 1999 to 2002 by 73% for walkers, 31% for cyclists, and 59% for both together. Utilitarian trips up 141%, going from 25% to 38% of all NMT trips involved. 6. Barnes, Thomp- son, and Krizek (2006) (see “Bicycle Lanes and Routes” — “Bicycle Lane Im- plementation” for Two ped-bike bridges were opened, and bike lanes were added to 2 road bridges, crossing the portion of the Mississippi River alongside down- town Minneapolis and the Univer- sity of Minnesota. The 1990 and 2000 Census results were used to Bike shares for Minneapolis-St. Paul trips crossing that segment of the river increased by 1.20 percentage points, up 36% (from 3.34%) during this decade of bridge improvements and improved bicycle connections. Bike shares for trips not crossing the river went up just more information) examine effects. (Evaluated commute trip bike shares only.) 0.34 and 0.86 percentage points (west and east sides of river, respectively). 7. City of Vancouver (2009a and b), Mustel Group Market Research (2009) (see this section for more information) On Burrard Bridge across False Cr., into downtown Vancouver, BC, lane and sidewalk use changes separated and protected bicycle and pedestrian flows. Daily “after” counts were compared to prior information. Random telephone interviews, 300 before and 300 after, 80% focused on most affected areas, were conducted. (Prior count datanot presented, survey emphasized perceptions.) Most pedestrian feedback positive, but some objections to inconvenience of relegation to one sidewalk. No signi- ficant change in pedestrian volumes. Cyclist reaction was enthusiastic. Cycle volumes were up 26%, July 13 through September 30, 2009, especially on weekends. Women cyclists up 31% versus 23% for men. Incidence identi- fied in interviews of walking the bridge was a wash but doubled for cycling. (continued on next page)

16-114 Table 16-21 (Continued) Study (Date) Process (Limitations) Key Findings 8. Harkey and Zegeer (2004) An historic bridge across Town Lake in Austin had 3.5-foot sidewalks. NMT fatalities occurred in 1991 and 2000. A parallel high-amenity ped- bike bridge was constructed with trail connections. (No examination of diversion or induced trail use.) With the nearest alternative crossing 1 mile away, the historic Lamar Bridge had some 700 to 1,000 NMT crossings per day before opening of the ped-bike Pfluger Bridge. The new bridge was initially used by 4,000 to 5,000 NMT crossings, a number said to be rising. 9. Rails-to-Trails Conservancy (2010) The Walkway Over the Hudson, on a spectacular 1.25-mile former rail- road bridge, opened October 3, 2009, from Poughkeepsie, NY, to the west side. (Info. limited to news item.) Despite short-term lack of connection to regional trails, the shared use bridge attracted 300,000 visitors in first 1-1/2 months (including 50,000 opening day crowd) versus a 267,000/year forecast. 10. McCarthy (2009) A new cable-stayed bridge over the Cooper River and Charleston, SC, harbor opened in mid-2005 with a 12-foot wide, 2.7-mile path. Inter- views were completed in Jan.-July, 2007, with 373 local area adult users, at multiple times of day, weekdays and weekends. (No count informa- tion, lower interview success with cyclists, connections not in place.) Of users approached, 17% were tourists (not interviewed). Interviewees includ- ed 57% walkers, 26% runners/joggers, and 17% cyclists, and were 56% female and 89% white. Utilitarian trips were 10% of total, with the top-ranked rea- son for bridge path commuting “To fit exercise into the Routine.” Increased activity was self-reported by 67% of all users and 75% of regular walkers. 11. Moritz (1995 and 2005a and b) (see this section and “Shared Use, Off-road Paths and Trails” — “Shared Use Path The Burke-Gilman/Sammamish River Trails were joined into a 27- mile trail linking north Seattle and UW with multiple north King Co. suburbs. The 3-mile gap was half closed in 1988 and fully closed in 1993. Counts covering 12 hours were taken near each end of the final Tuesday bicycle volumes, 7 AM - 7 PM, rose at Sheridan Beach (closest in) from 617 (1 day) to 1,136 (2-day average), up 84%, and at Kenmore from 330 to 1,079, up 227%. Saturday volumes declined at Sheridan Beach from 2,485 to 2,260, down 9% (presumably due to drizzle), but rose from 1,803 to 2,548, up 41%, at Implementation” for more information) gap in 1990 and 1994. (1994-1995 bicycle volumes on the two trails were not sustained in 2000 or 2005.) Kenmore. Overall Tuesday volumes on the trails dropped 24% 1985-1990 but rose 134% 1990-1995 (see Table 16-17). 12. Langdon (2010), Transport and Main Roads (2004-2009), data analysis by the Handbook authors The Western Freeway and Centen- ary Bikeways in Brisbane, Australia, were separated by a “missing link” until joined into a single radial route in late 2006. Biannual 7-day counts are taken at 2 sites north and 2 sites south of the link. (The induction- loop counters used counted bicycles only and may have missed some.) Cycle traffic exhibited minimal 2003- 2006 growth at the 3 count sites closest to the “missing link.”b With connection made, 2006-2007 growth was 54% (weekdays) and 59% (weekends). The 2007-2009 annual growth was 13% and 10% per year, bringing 2009 24-hour volumes to over 200 on each side of the former gap. 13. Barnes, Thompson, and Krizek (2006) (see also “Bicycle Lanes and Routes” — “Bicycle Lane Implementation”) A Minneapolis-St. Paul study of impacts of introducing 3 major bike lane facilities and 4 major off-road trails involved experimentation to find the best facility commutershed description for analysis. (Analyzed work purpose trips only, study not focused on system interconnection.) With a 5-mile length limit imposed on trips to be analyzed, new facilities in St. Paul (both <5 miles long) showed no bike share increase in their corridors. After relaxing the limit to allow inclu- sion of multi-facility trips, the bike share for TAZs along St. Paul facilities was shown to have increased by 37%.

16-115 Table 16-21 (Continued) Study (Date) Process (Limitations) Key Findings 14. Canada Mort- gage and Housing Corporation – 2008, as summar- ized in Victoria Transport Policy Institute (2011b) (see this section for more information) Measures of walking and driving directness to nearby retail and recreational destinations were utilized to identify Seattle, WA, area neighborhoods where pedestrian system connectivity was better, equivalent, or inferior to the connec- tivity via automobile. (Research methodology/details not reported.) Where neighborhood pedestrian connectivity exhibited greater directness than vehicular connectivity, the walk mode share was 18 percent. Where pedestrian and vehicular connectivity were about the same, the walk share was 14 percent, and where pedestrian connectivity was inferior, the walk share was 10 percent. Notes: Where substantial additional information on individual studies is provided in text and tables or figures, this is noted — and the location within the chapter is given — in the first column. a Goodwill Bridge trip diversion and mode shift estimates presented here reflect adjustment by the Handbook authors for differential pedestrian versus bicyclist survey response rates (see text Footnote 69 in the “Travel Behavior Shifts” subsection of the “Related Information and Impacts” section). b The northernmost count site, farthest from the “missing link” and closest to the Brisbane core, exhibited a post-2006 growth too large in absolute terms to be attributable in any major way to joining of the paths across the missing link. It has thus not been included in the gap closure impact assessment. Sources: As indicated in the first column.

Table 16-22 also provides daily bicycle volumes on each of the four principal bicycle-carrying bridges in the central Portland area. These volumes are primarily estimated from 2-hour weekday peak period counts and thus tend to emphasize commuter use shifts. The ongoing process of access improvements, and the shifting of cyclists among bridges in response to improvements, makes interpretation of individual bridge volumes problematic. Nonetheless, the 1997 to 1999 2-year before/after increase of 45 percent on the Hawthorne Bridge in response to sidewalk widening, and the 2000 to 2002 2-year before/after increase of over 360 percent on Steel Bridge in response to a cantilevered lower-level ped-bike side-bridge, are particularly notable. What truly stands out, however, is the steady growth in the bicycle total for the four bridges, excepting only a dip in 1995 and a spurt concurrent with opening of the Steel Bridge facility. From pre-1992, with a four-bridge weekday volume total of 2,855 bicyclists, to 2004 with 8,875 total, cycling cross-river more than tripled. During the 1990s, excluding Steel Bridge, bicycle volumes went up 78 percent as compared to an 8 percent increase in vehicular traffic on the three bridges and a 14 percent growth in Portland’s population. 16-116 Table 16-22 Willamette River Bridges Daily Bicycle Traffic Vis-à-vis Improvements Year Bikeway Miles Bridge Projects a Broadway Bridge Steel Bridge Burnside Bridge Hawthorne Bridge Bicycle Total 1992 83 755 230 1,075 1,500 3,560 1993 86 Burnside b 735 220 1,010 1,920 3,885 1994 103 690 220 980 1,940 3,830 1995 113 527 200 620 1,910 3,257 1996 143 950 350 1,065 2,165 4,530 1997 166 1,205 475 1,375 2,170 5,225 1998 182 Broadway c 1,854 460 905 2,471 d 5,690 1999 213 Hawthorne e 1,476 360 920 3,154 5,910 2000 221 1,405 410 1,080 3,125 6,020 2001 234 Steel f 1,680 1,250 965 3,729 7,624 2002 250 1,712 1,891 965 3,682 8,250 2003 253 1,683 1,860 965 4,055 8,563 Notes: a Various staged bicycle access improvements on the approaches to the Broadway, Burnside, and Hawthorne Bridges are, in the interests of brevity, not listed. Certain of the volume changes appear to relate directly to these improvements. b Burnside Bridge restriped to provide on-street bicycle lanes. Original 10-foot sidewalk width unchanged. c Broadway Bridge slippery sidewalk surfaces replaced at original 10-foot width (8.5-foot clear space). d Hawthorne bridge closed to bicycles for reconstruction. Count conducted on Morrison Bridge detour. The prior and following year Morrison Bridge bicycle count was 100 cycles. e Hawthorne bridge reopened with shared-use sidewalks widened from 6 feet to 10.5 feet. f Steel Bridge 12-foot pedestrian and bicycle facility opened alongside lower (railroad) deck, connecting to an extended Eastside Esplanade and pre-existing facilities. Until this point the only NMT accommodation was one 5-foot upper (highway) deck shared use sidewalk. Daily bicycle volumes mostly extrapolated from 2-hour weekday peak period (Birk, 2003). See accompanying text for 2004-2008 summary bikeway miles and bridge bicycle traffic data. Sources: Birk and Geller (2006), Birk (2003).

The daily four-bridge bicycle crossing increase (including Steel Bridge) was 140 percent over the 12 years covered in Table 16-22. Introduction of the Steel Bridge lower-level crossing has introduced into the bridge-use mix a popular jogging and bicycling exercise loop via Steel and Hawthorne Bridges and connecting paths, likely only partially reflected in the weekday-peak-derived volumes (Birk and Geller, 2006, Birk, 2003). Bicycle route choice modeling prepared for inclusion in Portland’s regional model, based on previously discussed bicycle-rider GPS tracking, found riders to view bridge-with-bike-lane passage 22 to 41 percent more favorably (i.e., more important to route choice) than an ordinary bike lane or cycling on a quiet street. (The lower percentage pertains to commute trips and the higher percentage to non-commute utilitarian trips.) Riders viewed bridge-with-separate-bike- facility passage 41 to 81 percent more favorably (Broach, Gliebe, and Dill, 2011). Since the Steel Bridge lower-level side-bridge was the only separate facility on a bridge at the time of the 2007 GPS research, these separate-facility values do not directly pertain to cycle tracks or buffered bike lanes on bridges, although they offer hints as to likely attractiveness. As of 2008—after accelerated post-2003 growth—the Willamette River four-bridge weekday vol- ume total had reached 16,700 bicyclists, approaching a sixfold growth from pre-1992. This record was despite a slowed increase in facility mileage, which expanded only from 262 miles in 2004 to 274 miles in 2008, not quite 3-1/2 times the pre-1992 system extent of 79 miles (Gotschi, 2011). Possible explanations for the continued increase in bicyclist river crossings, including Portland’s ongoing individualized marketing program, delayed response to prior actions, and increasing gaso- line prices, are explored in the “NMT Policies and Programs” subsection under “New World Program Examples”—“Portland, Oregon.” Also provided there is discussion of an observed bicyclist-volume drop-off in 2009 that erased half the 2007–2008 growth. Goodwill Bridge, Brisbane, Australia. The Brisbane River, positioned adjacent to the historic downtown core of Brisbane, Australia, forms a barrier somewhat like the Willamette River in Portland. It has, however, fewer crossings. The Goodwill ped-bike bridge (2nd entry, Table 16-21), opened in October, 2001, providing the first direct connection for walkers and cyclists between the south end of the CBD—along with the adjoining Queensland University of Technology—to South Bank Parklands, South Bank commerce, and surrounding residential and mixed-use areas. The new connection brought an additional south-of-the-river commuter railroad station into play as a destination station serving the CBD, and similarly provided a faster link for some bus riders than taking the bus all the way in. It also provided direct access to the south end of the CBD from less expensive south-of-river automobile parking. Table 16-23 provides Goodwill Bridge count data obtained 5 months after bridge opening (Abrahams, 2002). As in most cases with multi-day, multi- week NMT count data, the variation is notable. 16-117

16-118 The data obtained in these two weeks of counts show weekly use to have averaged 57,270 pedes- trians and bicyclists. The weekday average was 8,629 with 81 percent walking and 19 percent bicy- cling. The Saturday and Sunday average was 7,063 with 76 percent walking and 24 percent cycling. These weekday averages are higher, and the weekend figures are lower, compared to counts made 1 month after opening when the facility was more of a novelty. Tables 16-24 and 16-25 provide user gender, trip purpose, and bridge use motivations for week- day peak period Goodwill Bridge users as obtained from 397 respondents to a post-graduate stu- dent research survey. The hand-out survey was administered on a Wednesday in late June, a time of mild winter weather in Brisbane, just over 8 months after the official date of bridge opening. As the data in Table 16-24 suggest, the gender distribution was typical of Australia and North America, with a fairly even distribution of male and female pedestrians and a pronounced tilt toward males among cyclists. Age distributions were likewise within the range of commonly encountered find- ings. The proportion of survey respondents who were cyclists was roughly 30 percent, higher than the actual proportion, given both the 19 percent cycling share obtained in the earlier March counts and a reported survey response rate for walkers of about 25 percent as compared to 50 percent for cyclists. Table 16-24 Goodwill Bridge Weekday Peak Periods User Gender and Trip Purposes Bridge Use Mode Survey Sample Bridge User Gender Bridge User Trip Purpose Males Females Commute Social/Shop Sport Walkers 276 50% 48% 82% 6% 16% Bicyclists 121 73% 26% 72% 10% 17% Notes: Bicyclist data includes 119 actual cyclists, 1 wheelchair user, and 1 roller-blader. Gender was not identified for 2% of bridge users. Only persons 18 and older were surveyed. Adult and late-teen school trips were included as commute trips. Trip purpose was not obtained for 2%, and another 2% gave multiple trip purposes. Source: Abrahams (2002). Table 16-23 Goodwill Bridge Daily Pedestrian and Bicycle Traffic Over the Brisbane River Day of Week March 16-22, 2002 March 23-29, 2002 Walkers Cyclists Total Walkers Cyclists Total Saturday 4,704 1,630 6,334 3,559 1,167 4,726 Sunday 5,171 1,941 7,112 7,967 2,112 10,079 Monday 7,468 1,770 9,238 8,703 1,969 10,672 Tuesday 8,852 2,002 10,854 7,699 1,958 9,657 Wednesday 7,673 2,168 9,841 7,594 1,786 9,380 Thursday 7,517 1,881 9,398 6,342 1,174 7,516 Friday 5,147 1,251 6,398 2,757 a 578 a 3,335 a Note: a Showers (all other days clear or overcast). Source: Queensland Department of State Development as presented in Abrahams (2002).

The travel purpose distributions exhibit a high proportion of commuters among bridge users. The survey protocol included as “commuters” students attending classes. The heavy commuting use clearly reflects the peak-period timing of the survey. It is also the result of bridge location, next to the Queensland University of Technology and a part of the CBD, and the linkage the bridge pro- vides with less expensive commuter parking and also commuter rail and bus services. The CBD sector involved had relatively low accessibility previously, and to public transportation in partic- ular, with the main rail terminal located at the opposite CBD fringe (Abrahams, 2002). One of the rare insight opportunities offered by this research is the separate identification and cross-tabulation of trip purpose and bridge use motive. The two were obtained in individual sur- vey questions. The bridge use motive is also the walking and cycling motive, given that the cross- ing is NMT-only. Table 16-25 summarizes bridge use motive distributions, stratified by walkers versus cyclists and separately identified for commute trips as compared to trips for all purposes. Multiple motivation responses were allowed and given. 16-119 Table 16-25 Percentage Distributions of Motivations for Weekday Peak Period Goodwill Bridge Use, by NMT Mode and All Versus Commute/Non-Commute Purposes Motivation (Reason) All Walkers Commute Walkers All Cyclists Commute Cyclists All Res- pondents All Commute All Non- Commute Quicker 52% 64% 76% 85% 59% 71% 32% Cheaper 26 34 45 49 32 39 19 Safer 11 13 40 42 19 23 13 Environment 22 27 51 54 31 36 22 Exercise 56 59 60 58 57 59 61 Fun 45 39 48 46 46 40 69 Other 3 4 7 7 4 5 4 Notes: Multiple motivation choices allowed. Motivation not obtained for 2% of survey respondents. All-mode combinations (last three columns) are affected by differential walker versus cyclist survey response rates. The motivation percentages are nevertheless presented unweighted, for lack of information to adjust all the summary categories in question. It appears that survey response rate adjustment would lower the all-mode combination motivation percentages, if at all, by 1 to 4 percentage points each, without substantively affecting relative standing among the individual motivations (i.e., motivation importance rankings). “Walkers” and “Cyclists” columns would be unaffected. “Commute” includes both work commute and student commute. Source: Abrahams (2002), with note on survey response rate effects by the Handbook authors. Time savings (“Quicker”) stand out in Table 16-25 as the most frequently cited motivation for Goodwill Bridge use by commuters. Cost savings (“Cheaper”) were cited a little more than half as much, and likely pertain in significant measure to persons using less expensive parking to the south and finishing their commute by walking over the bridge. Most notable, however, is the importance of exercise not only to all respondents but to commuters as a distinct group. Moreover, “Fun” is noted by many, even commuters, as a bridge use inducement (Abrahams, 2002). The exercise motive identification rate, at about 55 to 60 percent for both commuters and non- commuters, is another demonstration of the tendency to accomplish two things at once by exercis- ing while also getting to work or satisfying other needs and desires. The fun motive response rate

may in part reflect the relative newness of the bridge, but probably is also an indicator that bridge use is motivated in part by its being a stimulating destination in its own right and not simply a means to an end (see “Underlying Traveler Response Factors”—“Behavioral Paradigms”). The “safer” motivation—found most often with cyclists—apparently reflects comparison with the previously overcrowded NMT facilities on the next bridge over, the Victoria Bridge, where pedes- trians and bicyclists share crowded sidewalks. Of surveyed Goodwill Bridge walkers and cyclists, 40 percent previously crossed on the Victoria Bridge. That figure includes 36 percent of Goodwill Bridge pedestrians and 60 percent of cyclists. This outcome was desirable from the perspective of authorities concerned with Victoria Bridge NMT facility safety (Abrahams, 2002). Identification of Goodwill Bridge survey respondent prior modes and bridge choice is another of the rare insights offered by this research. The prior modes reported are tabulated and discussed, in adjusted form, in the “Related Information and Impacts” section under “Travel Behavior Shifts.” The location of the Goodwill Bridge relative to the CBD, the adjoining university, and various com- ponents of the transportation network, fosters its use as a link in multimodal trips. The particular circumstances involved may not be common to many other urban scenarios, but the bridge user survey data nonetheless vividly illustrate the complexity of multi-mode travel that may compose substantial components of an urban NMT facility’s usage. Only 3 percent of interviewed cyclists were in the process of making multi-mode trips, all involving two modes, but 52 percent of sur- veyed walkers were making a two-or-more-mode trip. Ignoring multi-mode bicycle trips and con- solidating two- and three-mode walk trips,28 it is found that 50 percent of multi-mode walkers used auto for their motorized link, 28 percent used train, 17 percent used bus, less than 1 percent each used ferry or taxi, and 4 percent used two motorized modes including auto/train, bus/train, and auto/bus (Abrahams, 2002). Other River Bridges. Information on other river bridge provisions for pedestrians and/or bicyclists, though less complete, presents a broader range of circumstances and outcomes. Eight different river crossings are covered in the 3rd through 10th entries in Table 16-21. The first of these, the Lewiston- Auburn Railroad Bridge in Maine (3rd entry), is an example from the other end of the volume scale from the large-city river bridges of Portland, Oregon, and Brisbane, Australia. The short shared use rail-trail over the converted Lewiston-Auburn bridge crosses the Androscoggin River, connecting a city park at the south end of the Auburn downtown with the Lewiston-Auburn Railroad Park on the other side. Through the latter, and a former textile mill district, the south end of the Lewiston down- town may be reached (Historical Marker Database, 2010). The combined population of the two cities at the time of the 1995 bridge conversion was about 58,000. Within 3 years of opening as a new ped- bike river crossing, facility use was 350 people a day or more (RTC and APBP, 1998). The Greenway Bridge in Eugene, Oregon (4th Table 16-21 entry), was opened across the Willamette River in 1978, providing a ped-bike-only linkage between a major shopping complex to the north- east and a residential area to the southwest. It also connects shared use trails on each side of the river. The CBD is about 2 miles away along the river to the south (Lipton, 1979). The new link reduced travel time and distance for many non-motorized travelers, such that when surveyed, approximately 30 percent of cyclists thought it as quick or quicker to make their trip by bicycle via the bridge as compared to driving an automobile (Zehnpfenning et al., 1993). Survey results sug- 16-120 28 Only four surveyed cyclists made multi-mode trips, three involving commuter train use and one auto use. Most of the 14 percent of multi-mode walkers who reported use of three modes simply reached that total by reporting a walk at each end of their motorized link. Thus the summarization focuses on walkers and lumps two-mode and three-mode multi-mode walk trips together.

gested that 14 to 18 percent of weekday users and 28 percent of Saturday users would not have made their trip by bicycle without the Greenway Bridge. These trips were most commonly recre- ational trips; nevertheless, a reduction of more than 500 automobile trips per week by 1978 bridge users was estimated. April–May 1978 Greenway Bridge weekday cyclist trip purposes were 32 to 41 percent recreation, 46 to 41 percent work and school, 10 to 12 percent shopping, and 11 to 7 percent personal business and other. On the same 2 days pedestrian trip purposes for three Willamette River crossing oppor- tunities combined, from the Greenway Bridge south to the CBD and the University of Oregon, were 35 to 63 percent recreation, 28 to 16 percent work and school, 15 to 10 percent shopping, and 22 to 12 percent personal business and other (Lipton, 1979). Some 1,100 summer weekday cyclists were counted on the Greenway Bridge in 1982 (Bicycle Federation of America, 1993). This volume was likely substantially higher than the bicycle traffic during the 1978 surveys, taken when the bridge was new. The Millennium Bridge in York, England (5th table entry), was built over the River Ouse in 2001, near the University of York campus and roughly 1 mile from the city center. Its location saves walkers and cyclists up to about 1-1/4 miles maximum. Annual usage of connecting bicycle and pedestrian paths and other routes at each end of the crossing increased from 430,000 walkers to 740,000, from 220,000 cyclists to 290,000, and from 650,000 overall to 1,030,000. These increases reflected both presence of the new ped-bike bridge and further development of the feeder route system. As indicated in the 5th entry to Table 16-21, utilitarian trips increased 141 percent compared to 59 percent for all trips. The increase in annual utilitarian trips was from 160,000 to 390,000, thus over one-half of all new trips were to and from destinations such as workplaces and shops (Booz Allen Hamilton, 2006). The 6th study listed in Table 16-21 included, in its longitudinal impact analyses of various Minneapolis bicycle facilities, an examination of work trip mode shifts to bicycling likely to have been magnified by improvements to Mississippi River crossings northeast of downtown Minneapolis. Two separate former railroad bridges were converted to ped-bike bridges, and bicycle lanes were provided on two other bridges, while NMT provisions were unchanged during the decade on an additional two bridges. The increment of growth in percentage points for bicycle shares of work trips went up substantially more, as quantified in the table, for trips crossing the river relative to other trips within Minneapolis and St. Paul (Barnes, Thompson, and Krizek, 2006). Extrapolations to 24-hour volumes for 2007 from mostly 12-hour counts give scale to total weekday cross- Mississippi bridge usage near the Minneapolis downtown. Starting north of downtown and mov- ing toward the southeast, bridges and their NMT volumes were 1,560 pedestrians and 1,200 cyclists on the Hennepin Avenue bridge, 690 pedestrians and 490 cyclists on the 3rd Avenue bridge, 2,120 pedestrians and 940 cyclists on the ped-bike Stone Arch Bridge, 940 pedestrians and 990 cyclists on the 10th Avenue bridge, and 250 pedestrians and 130 cyclists on ped-bike Bridge #9 (City of Minneapolis, 2007).29 The Burrard Bridge trial reallocation of space among bicyclists, pedestrians, and motorized traffic, the 7th case listed in Table 16-21, is unusual both in the low-cost approach to enhancing ease of bicycling and increasing bike and pedestrian safety, and in the information it provides on bicyclist response to 16-121 29 These counts were taken 41 to 57 days following the collapse of the I-35W freeway bridge, located between the Stone Arch and 10th Avenue bridges, with unknown effects on NMT volumes. (The Stone Arch and 10th Avenue bridges would make exceptional viewing platforms for “sidewalk superintendents.”) Bicycle volumes were up 76% over 2003 on the Hennepin Avenue bridge, 96% on the 3rd Avenue bridge, and 34% on the Stone Arch Bridge. Comparisons were not published for pedestrians (City of Minneapolis, 2007).

the safety improvements in particular. Burrard Bridge is the westernmost of three crossings of False Creek from southern city neighborhoods into downtown Vancouver, British Columbia, Canada. In years immediately before the trial, the bridge carried nearly 2,400 pedestrians and 3,500 cyclists per day. Approximate mode shares for the 8,000 to 9,000 per-hour peak-period bridge users were 65 per- cent auto, 20 percent transit, 10 percent walk or bicycle, and 5 percent unspecified or rounding error. The starting condition was six traffic lanes and two 2.6-meter (8.5-foot) sidewalks, each shared by pedestrians and cyclists. Space allocation was changed to have two southbound (outbound) traffic lanes, one southbound barrier-protected bicycle lane next to the west sidewalk (essentially a cycle track), two-way pedestrian-only flow on the west sidewalk, three northbound traffic lanes as before, and northbound bicycle-only flow on the east sidewalk. Bridge approach changes including signal- ization and bike lane adjustments accompanied the bridge modifications. The revised traffic patterns were in place starting July 13, 2009, after a weekend of implementation (City of Vancouver, 2009b). Helped by transit priority enhancements, bus travel times were affected very little, if any, and over- all vehicle volumes and travel times exhibited no appreciable change. The largest traffic impact has been a peak-period travel time increase averaging 1-1/2 minutes for one particular traffic movement. Vehicular traffic diversion to next-over Granville Bridge was only a temporary phenomenon, while pedestrian and bicyclist diversion was not studied (City of Vancouver, 2009a). Hospital emergency visits for Burrard Bridge cycling crashes during 20 summer weeks studied by the University of British Columbia dropped from four in 2008 to one in 2009 (City of Vancouver, 2010). Walkers and cyclists reported feeling safer and more comfortable with the changes, cyclists especially so. Counts showed no significant change in bridge pedestrian volumes, despite some complaints about being routed onto one sidewalk. Bicycle counts went up 26 percent, including a 40 to 70 percent increase in weekend use. (Post-Labor Day gains were more muted.) As indicated in Table 16-21, cross-bridge cycling went up more substantially for women (31 percent) than for men (23 percent). Anecdotal reports were suggestive of more cross-bridge cycling by children. In the market survey interviews noted in Table 16-21, one question asked whether the resident had walked or cycled over the bridge at least once in the previous month. Incidence of reported cross- bridge walking went up 61 percent for the Downtown neighborhood while biking incidence was up 92 percent. Walking incidence for the Near Westside neighborhood, south of the bridge, went down 44 percent but biking incidence was up 119 percent. The net effect for all interviewees was a 6 percent decline in reported incidence of cross-bridge walking in the last month (16 percent before, 15 percent after) versus a doubling of reports of cycling across (9 percent before, 18 percent after) (City of Vancouver, 2009a, Mustel Group Market Research, 2009). The 8th and 9th entries in Table 16-21 cover a crossing of Town Lake (an impounded section of the Colorado River) in Austin, Texas, and The Walkway Over the Hudson at Poughkeepsie, New York. Roughly a fivefold increase was seen in on-site Town Lake crossings when the ped-bike Pfluger Bridge opened parallel to the historic Lamar Street bridge with its dangerously narrow 3.5-foot sidewalks (Harkey and Zegeer, 2004). The Hudson River crossing was new as a ped-bike facility, although the former railroad bridge had been built in 1888 (Rails-to-Trails Conservancy, 2010). Initial counts on both bridges averaged on the order of 5,000 persons per day (visitor counts in the case of the Hudson River bridge; see Table 16-21). The Pfluger Bridge with its scenic connecting Town Lake Hike and Bike Trail system and The Walkway Over the Hudson itself may reasonably be adjudged recreational des- tinations in their own right—the Hudson River, “world’s tallest pedestrian bridge,” spectacularly so. As such, these bridges presumably reflect the role of conventional economic demand in explaining usage of highly attractive ped-bike facilities, as contrasted to the derived-demand theoretical basis for most urban trip making. These different demand concepts are discussed under “Behavioral Paradigms” in the “Underlying Traveler Response Factors” section. 16-122

The 10th entry in Table 16-21 involves another ped-bike crossing serving as an attraction in its own right, this one a 2.7-mile sidewalk/path constructed as an integral part of a new South Carolina bridge over the Cooper River and Charleston harbor, connecting Mt. Pleasant and Charleston. Its role as a tourist attraction was underscored when 17 percent of path users approached for inter- views were found to live more than 20 miles distant. Neither of the pair of bridges replaced had safe NMT provisions. Interviewees, limited to users living within 20 miles, included 57 percent walkers, 26 percent runners and joggers, and 17 percent bicyclists. They were fairly well distributed among age groups. Women constituted 56 percent of users and were more likely to be regular bridge walkers than men. Men were more likely to be regular bridge runners and much more likely to be bridge bicyclists. A motor vehicle was the mode of access of 73 percent of all users. Non-whites, mostly African American, were 11 per- cent of users and self-reported increased physical activity in 85 percent of all cases. Among white bridge users, 64 percent reported more physical activity because of the new facility. Of all interviewees, 10 per- cent were making work trips or running errands. Percentages varied by gender, with 6 percent of women and 15 percent of men reporting commuting. The top-ranked reason for bridge path commut- ing, at 4.8 on a 5-point scale, was “To fit exercise into the Routine” (McCarthy, 2009). Path Gap Closures. Closure of the 3-mile “Missing Link” across the top of Lake Washington, between Seattle’s Burke-Gilman Trail and the Sammamish River Trail in northeast suburbs includ- ing Redmond, affords a trail-gap-closure mini-study. Joined, the trails form an inverted broad- based “U” shape. As indicated in the tenth entry of Table 16-21, the gap was half closed in 1988. The new off-road trail segment substituted for need to use a heavily trafficked four-lane state high- way with poor shoulders. The gap was fully closed in 1993, with installation of a tunneled grade separation eliminating routing via a busy industrial street (Moritz, 1995). The Table 16-21 “Key Findings” entry focuses first on 1990-1994 before-and-after bicycle statistics, noting Tuesday increases of 84 percent at Sheridan Beach (closest-in and busiest side of the new link) and 227 percent at Kenmore, and drizzle-affected Saturday changes of −9 percent and +41 per- cent, respectively. Corresponding Tuesday pedestrian count changes were −19 percent (closest-in location) and +163 percent, while Saturday changes were −11 percent and +150 percent (Moritz 1995 and 2005b). The pedestrian count outcomes seem to reflect an overall increase combined with redistribution to the new and to the previously less accessible sections of the combined trails. Although the 1993 final gap closure received the most attention, the partial closure of 1988 can also be examined using 1985 and 1990 counts at the same Sheridan Beach and Kenmore locations. In this instance Tuesday bicycling counts changed by −28 percent and +61 percent, respectively, while increasing by 48 and 147 percent on Saturday. All pedestrian counts were up, with particularly sharp percentage increases at the least used end of the gap at Kenmore, where 1985 Tuesday and Saturday walkers from 7:00 AM to 7:00 PM totaled 10 or less, as compared to roughly 70 in 1990 and on the order of 200 in 1995. Across the 1985–1995 decade spanning the two-stage gap closure, Sheridan Beach total NMT usage went from 1,208 to 1,645 on the weekday and from 2,026 to 2,964 on the weekend day. Kenmore total usage increased from 210 to 1,540 on the weekday and from 743 to 3,204 on the weekend day (Moritz 2005b). Sorting through these variations, which may be only partially attributable to the gap closure, the over- all trail traffic growth stands out clearly. Growth was more modest at the end of the gap with the higher initial volumes, with little change in the cycling/walking ratio. Volumes at the end of the gap with the lower initial volumes rose sharply, with a somewhat larger gain for pedestrians, to nearly match counts at the higher-volume end. The observed weekday growth seems most related to the final gap closure. Weekend growth, on the other hand, seems more evenly divided between the partial and 16-123

final gap closure periods. The findings of huge positive impact are muddied somewhat by inability of the combined trails to sustain the trail traffic levels achieved in 1995 on through the 2000 and 2005 count years, an outcome discussed more extensively in the previous “Shared Use, Off-Road Paths and Trails” subsection under “Shared Use Path Implementation”—“Seattle Urban/Suburban Trails.” A second example of path interconnection is provided by the joining of the Centenary and Western Freeway Bikeways in Brisbane, Australia, in late 2006 (12th entry, Table 16-21). Brisbane’s “miss- ing link” was a little over a kilometer of difficult terrain. When joined, a shared use through facil- ity was formed linking southwest Brisbane and the central area to the north. The count station immediately south of the missing link showed the largest changes. Weekday 24-hour bicycle counts averaged 62 in 2004 through 2006 with no growth. Closing the gap led to a 142 percent growth from 2006 to 2007, followed by growth averaging 24 percent per year from 2007 through 2009. Weekend cycle volumes were 60 in 2004, grew about 9 percent per year through 2006, gained 164 percent from 2006 to 2007 with gap closure, and had growth averaging 14 percent per year from 2007 through 2009 (Langdon, 2010, Transport and Main Roads, 2004–2009). The 2009 24-hour bicycle volumes at this site just south of the former missing link averaged 205 weekdays and 239 weekends. Weighted average results for this count station combined with the stations roughly 2-1/2 km. north and south are given in Table 16-21. The count station 5 km. north and closer to the CBD recorded 2009 biannual 24-hour volume averages of 640 bicycles on weekdays and 512 on weekend days (Transport and Main Roads, 2004–2009). Although the annual bicycle counts were not accompanied by pedestrian counts, importance of the interconnected path to walkers may be inferred from circa 2010 staged construction to widen and upgrade congested sections of the overall “bikeway” from a shared use path to separate pedestrian and bicycle pathways. Manhattan’s Hudson River Trail provides an example of an intermediate path segment so vastly upgraded that the improvement plays a major interconnectivity role. NMT volumes on the link dou- bled to quadrupled, depending on location and day of week (Chaney, 2005), as more fully detailed in the above-mentioned “Shared Use, Off-Road Paths and Trails” subsection under “Shared Use Path Implementation”—“Other Path Information,” including the 8th entry of Table 16-16. Facility Extension Effects. The methodology development stage of Minneapolis-St. Paul research described under “Bicycle Lane Implementation”—“Longitudinal Commute Mode Share Research” in the “Bicycle Lanes and Routes” subsection provides special insight into the difference between open- ing an isolated bicycle facility segment and making extensions to an existing bicycle network. The relevant research steps, and the unexpected findings, are summarized in the 13th entry of Table 16-21. In effect, the results of testing the initial research approach showed that two new facilities in St. Paul apparently generated no net increase in work commute bicycling self-contained within the new- facility corridors themselves. The entire commuter cycling increase within the facility corridors of 37 percent (an 0.45 percentage point mode share increase) was introduced by trips traveling beyond the new St. Paul facilities via (or within the corridors of) the pre-existing bicycle facility network. That network connects with the University of Minnesota and downtown Minneapolis (Barnes, Thompson, and Krizek, 2006). Another way of looking at this outcome is that if the two new St. Paul facilities had been built as iso- lated segments, one a shared use trail 1.9 miles long and the other bicycle lanes 4.6 miles in extent, the impact on commute trip bike shares would have been negligible. Since they were built as part of an interconnected system, their effect was substantial. Interconnections of Modest Scale. The introduction to this “Pedestrian/Bicycle Systems and Interconnections” topic notes the paucity of impact studies of lower-cost NMT connections, as impor- 16-124

tant as they may be. Three sets of observations offering clues as to the importance of local-scale inter- connections are presented here. The most definitive, the Seattle-based research included as the 14th entry of Table 16-21, is presented last. Paired community research in the San Francisco area sheds some light, although only through inference, on the importance of such connections. The analytical comparison in question, covering the Rockridge and Walnut Creek communities, is more fully reported in the “San Francisco East Bay Pedestrian Versus Auto Oriented Neighborhoods” case study of Chapter 15, “Land Use and Site Design,” and is also sum- marized in the upcoming “Pedestrian/Bicycle Friendly Neighborhoods” subsection. Rockridge has a traditional neighborhood design overall. The gridlike street network has irregular- ities, however. The pedestrian network makes up for the irregularities with fairly small blocks and also a number of pedestrian path interconnections. While the effect of these interconnections cannot be isolated, they presumably contribute to the relatively high walk and bike mode shares in Rockridge tabulated in the Chapter 15 case study. The Walnut Creek comparison neighborhood has a more auto-oriented design, but similar demographics and regional rail rapid transit service. Both have substantial commercial development, and work-purpose-trip rail mode shares are almost identical. However, for all walk/bike and bus transit categories of travel, Rockridge NMT and bus mode shares are over 2 times—and up to 7 times—those found in Walnut Creek (see Table 16-39, 14th entry) (Cervero and Radisch, 1995). Without the path interconnections, it is doubtful that Rockridge’s level of active transportation usage could be attracted. Within the earlier “Street Crossings” subsection, the 4th Table 16-5 entry records a dramatic pedes- trian flow increase from roughly 50 persons/day to 1,000/day at a Ft. Pierce, Florida, intersection converted to an urban traffic circle with extensive pedestrian safety features and amenities. The inter- section forms the gateway between the historic downtown and the beachfront. As noted in the table, the intersection improvement was part of an overall program to slow traffic, widen sidewalks, enhance the pedestrian environment, and revitalize the downtown. Supported by the increased foot traffic, new pedestrian-oriented retail has opened in previously vacant spaces. A key element of the pedestrian improvements was redeveloping part of a parking lot into beachfront-park pedestrian and bicycle access from the gateway intersection (Harkey and Zegeer, 2004). Pedestrian travel patterns were not determined, but an obvious supposition is that the pedestrian flow increase through the upgraded intersection reflects strengthening of a pedes- trian linkage between the business area and the beachfront. This is a linkage that had previously been severely degraded by inhospitable intersection traffic conditions and a beachfront parking lot barrier effect. NMT and motorized-travel levels of connectivity are often not the same, and may be manipulated to favor walking and bicycling. Unfortunately, limited-access-highway and major-arterial barrier effects, along with sidewalk deficiencies, more often result in the opposite condition. There are enhanced subdivision and new-town designs that use pathway connections and NMT connectiv- ity via small- and medium-sized parks to restrict through traffic while allowing relatively direct pedestrian and bicycle flow (Victoria Transport Policy Institute, 2011b, Stover and Koepke, 2002). Neighborhood traffic calming designs utilizing traffic diverters and other barriers to through traf- fic provide an equivalent higher-NMT-connectivity condition if passage for bicycles is provided and sidewalks are good. Walk mode choice effects of differing relationships between motorized and NMT connectivity were explored by the Canada Mortgage and Housing Corporation (see the 14th entry of Table 16-21). The actual research was conducted in urban neighborhoods of varying character in the Seattle, 16-125

Washington, region. Utilizing measures of walking and driving directness to nearby retail and recre- ational destinations, they determined that when pedestrian and vehicle connectivity were both high, the pedestrian mode share was about 14 percent. Where pedestrian connectivity offered greater directness than vehicular connectivity, the walk mode share was higher (18 percent) and where pedestrian connectivity was poorer it was lower (10 percent). A “Fused Grid” layout, with cul-de- sac or “U”-shaped-street loops made continuous for pedestrians and cyclists via public squares, was calculated to provide a 10 percent increase in relative connectivity for pedestrians. The study esti- mated that this would raise the odds of walking by almost 10 percent, produce a 23 percent decrease in local VMT, and increase the odds of meeting recommended physical activity levels through local walking by about 25 percent (Victoria Transport Policy Institute, 2011b). Finally, there is the advantage to transit service of having NMT interconnections that make up for street and sidewalk system discontinuities. Resolving such discontinuities increases transit service effective coverage and thus ridership, and supports choice of walk and bicycle modes for transit access. This role for interconnections of modest scale is discussed below in the “Pedestrian/Bicycle Linkages with Transit” subsection (see “Non-Motorized Access to Transit”—“Pedestrian Access and Egress,” espe- cially the 4th and 7th entries in Table 16-26 and associated discussion). Pedestrian/Bicycle Linkages with Transit Two primary aspects of NMT access and egress treatments for transit stops and stations are cov- ered here. First addressed are considerations involved in getting to and from transit by walking and bicycling. Of interest are the effects of improved NMT access on both transit ridership and the decision about what travel mode to use for transit access and egress. Modes of transit access include not just walking and bicycling but also motorized options, including driving and parking, getting dropped off in an auto, and connecting to a bus if available. Improved NMT access can come through either pedestrian and bicycle facility improvements, including bicycle storage provisions, or through alternative land development designs that place more residents and businesses within easy walking distance. Thus, the “Non-Motorized Access to Transit” discussion is immediately followed by a review of “Transit Oriented Development” findings. The other aspect addressed is the on-vehicle accommodation of bicycles to allow transit-riding cyclists to take their bicycles with them for use after alighting. The outcomes of various bicycles-on- transit programs are examined under “Bicycles on Transit Vehicles.” Such programs give bicyclists a flexibility of transit use conceptually equal to that afforded walkers, who inherently have full flex- ibility to walk at both ends of a transit trip. The flexibility to bicycle at both ends of a transit trip expands the effective service area, however, given the longer distances it is reasonable to cover in accessing and egressing transit by bicycle. Both aspects of NMT access and egress involve and may affect “mode of access” choice or share. It is important to differentiate mode of access share from mode share and sub-mode share. Mode share refers to choice of primary travel mode between a trip’s origin and its final destination. For exam- ple, a trip starting with driving alone to a light rail transit (LRT) station, followed by an LRT ride terminating a quarter-mile from the final destination, and concluding with walking to get there, would be classed as an LRT trip for purposes of mode share calculation. For a trip to be classified as a walk trip in this “mode share” context would require that the entire origin to destination dis- tance be walked, with no other mode involved. A mode share proportion, in most newer studies, is expressed as a percentage of all travel by all modes in the travel category of interest. Mode share is sometimes referred to as “prime-mode share” to clearly distinguish it from sub-mode share or mode of access share. 16-126

Less often encountered is sub-mode share, the proportion of transit trips using a particular form of tran- sit, such as local bus or heavy rail transit (HRT). A true sub-mode share is expressed as a percentage of all transit travel in the category involved. Of critical interest in examining NMT linkages with transit, especially in the context of local area traf- fic, parking, and environmental concerns, is mode of access share. This share describes the proportions among means of getting to and from the primary mode. The access and egress modes in the mode share example given above would be, respectively, drive-alone (to the station) and walk (from the station). As fully detailed in the “Related Information and Impacts” section (see “Extent of Walking and Bicycling”—“Extent of Walking”), the 2001 and 2009 NHTS surveys show 16 percent of all walk trips in the United States to be the access/egress component of transit trips.30 They compose 1.7 percent of all trips by any mode (Agrawal and Schimek, 2007, Kuzmyak et al., 2011). NMT access and egress to/from transit service is an important contributor to physical activity for transit riders, with an estimated 29 percent of transit users achieving the 30 minutes or more of physical activity a day recommended by the Surgeon General solely by walking to and from transit (Besser and Dannenberg, 2005). Further infor- mation on this benefit is provided in the “Related Information and Impacts section” (see “Public Health Issues and Relationships”—“Baseline Walking and Bicycling Activity” and also Table 16-123). Non-Motorized Access to Transit The quality of NMT connections to public transit may affect the overall choice to use or not use transit, thus affecting prime mode share, and is an important determinant of the choice of access mode, such as walking versus driving to the station. Even motorists who choose to drive to tran- sit stops must eventually leave their automobiles and walk to the boarding point, and are highly likely to walk to their final destination. A fall 1992 survey of San Francisco BART HRT riders found that walking accounted for more than 75 percent of all BART egress trips (Loutzenheiser, 1997). Similarly, a Chicago intercept survey reported 80 percent of Metra commuter rail riders and 73 per- cent of Chicago Transit Authority HRT riders walking to their final destination (Wilbur Smith and Associates et al., 1996a). More transit access and egress data are provided or cross-referenced in the discussion which follows. Walk and bicycle access to transit are examined separately within this NMT access to transit topic. The “Pedestrian Access and Egress” discussion below is immediately followed by discussion of “Bicycle Access and Egress.” Pedestrian Access and Egress. Transit riders are usually thought of as willing to walk about 1/4 mile, or 5 minutes at 3 miles per hour (mph), to a regular bus stop and about 1/2 mile to a rail transit stop. These rules of thumb have been generally confirmed by numerous evaluations, although some transit riders will walk farther (Ewing, 1996). Examples found further on in this subsection suggest that while the 1/4 mile value for bus riders applies to the majority, the 1/2 mile value for rail riders applies more to the median if measured along the walking route. Of equal importance is the finding of many transit rider surveys that transit mode share and walk to transit share both have a strong inverse relationship to the distance from the stop or station. A classic example addressing likelihood of using local bus transit is a study focused on a typical 16-127 30 The walk access and walk egress components of an individual trip via transit are not counted separately from each other in the NHTS survey, the primary source of U.S. national data on walking to and from tran- sit stops and stations. The walk access and walk egress are counted as one trip.

Hartford, Connecticut, bus route. Riders were surveyed to obtain demographic and travel pattern information. The bus riders and their trips were then classified by car ownership status and walk- ing distance, based upon over 350 usable survey responses. The trips were next compared to the number of dwelling units in each strata, similarly classified. A series of “ridership penetration curves” were developed, relating bus rides per 100 dwelling units to automobile ownership and distance from the nearest bus stop. At 200 feet from the nearest bus stop the zero-car ownership penetration ratio was 65 rides per 100 dwelling units, for one car it was 55 rides, and for multiple cars it was 50 rides. A decline in transit use with increasing walking distance to bus stops was found for each level of car ownership. The curves, in a range from 200 to 1,000 feet walking distance, show a drop of about five weekday rides per 100 dwelling units for each 100 feet in added walking distance for households within any one of the three car ownership categories. For example, between 200 and 1,000 feet from a bus stop, the penetra- tion ratio for single-car owners dropped from about 55 rides per 100 dwelling units to 15 rides per 100 dwelling units (Levinson and Brown-West, 1984). A 1996 Chicago study of HRT (subway/elevated) and commuter rail access provides a rail transit example expressed in terms of likelihood of choosing walk access. It demonstrates the relationship of shorter access distances to larger proportions of walk access and also the longer walks found in rail transit access as compared to local bus access. Overall, 84.1 percent of surveyed rail transit users within about 1/2 mile of the station chose to walk to it, 46.9 percent of users between about 1/2 mile and 1 mile chose to walk, 12.4 percent between 1 mile and 1-1/2 miles, 3.4 percent between 1-1/2 miles and 2 miles, and practically no users from farther than 2 miles (Wilbur Smith and Associates et al., 1996b). After distance, the four most prevalent reasons given by survey respondents for not walking to transit stations were “inadequate sidewalks, weather, not dressed appropriately, and dangerous traffic intersections” (Wilbur Smith and Associates et al., 1996a). Another study cites “danger from auto traffic, no sidewalks, and inadequate lighting” as the chief reasons for not walking by potential walkers (Replogle and Parcells, 1992). Figures 16-4 and 16-5 provide further background by illustrating the drop-off in walking with distance to BART HRT stations, from home, in the case of work-purpose trips utilizing this Metro-type San Francisco Bay Area system. The graphs show the proportion of work trips by each access mode, dur- ing a weekday, for different distances from the stations. Both figures present averages for a group of non-CBD stations, but Figure 16-4 pertains to urban BART stations such as Mission-16th Street (San Francisco), Berkeley, and Lake Merritt (Oakland) while Figure 16-5 pertains to suburban center stations. Note that for urban stations walking takes place for longer distances from the stations, with at least 50 percent walk access for slightly over 1/2 mile. The primary alternative access mode is bus transit. For suburban center stations walking maintains at least a 50 percent walk access mode share for only up to 3/8 mile, and the primary alternative is park-and-ride (Parsons Brinckerhoff et al., 1996b). A sur- vey of Mountain View, CA, Caltrain commuter rail station area resident users showed a walk access from home mode share pattern close to midway between those illustrated in Figures 16-4 and 16-5, but for commuter rail trips of all purposes (Park and Kang, 2008). The illustrations and the Mountain View observations are all based on airline distances (Parsons Brinckerhoff et al., 1996b, Park and Kang, 2008). Table 16-26 summarizes a number of studies that, from various points of view, investigated the importance of good transit service to walking activity, or the role of pedestrian access distance or facility availability on transit mode choice, or choice of the walk mode for transit access. The research covered in the 1st and 2nd table entries found, respectively, that quality of transit service is positively associated with choice to walk in general (Committee on Physical Activity, Health, Transportation, and Land Use, 2005) and that transit users are more likely to participate in at least a moderate level of walking than non-transit users (Moudon et al., 2007). 16-128

16-129 Figure 16-4 Commute trip mode of access from home to urban BART stations. Source: Parsons Brinckerhoff et al. (1996b). Figure 16-5 Commute trip mode of access from home to suburban center BART stations. Source: Parsons Brinckerhoff et al. (1996b).

16-130 Table 16-26 Summary of Research Findings and Other Studies on Relationships of Transit Service Levels and NMT Access Quality with Walk and Walk/Transit Activity Study (Date) Process (Limitations) Key Findings 1. Cervero and Gorham – 1987 as summarized per SR 282 Tabulated 1990 Census travel data for 14 income-matched “transit” and “auto” neighborhood pairs in San Francisco Bay Area and 12 in Los Angeles region. (Work trips only.) Transit neighborhoods in S. F. area had 1.2 to 13.4 percentage points higher walk mode share. In L.A. area, the transit neighborhoods had 1.7 to 24.6 percentage points higher walk share. 2. Moudon et al. (2007) & Moudon et al. (2005) (see “Ped…cycle Friendly Neigh- borhoods” for more information) Cross-sectional analyses of walking and cycling activity, socio- demographics, attitudes, and objectively measured environmental variables covering 608 adults in King County, WA. (Extent of walk and bike activity self-reported.) Transit users (at least once a week) were found to be much more likely to walk moderately (odds ratio of 4.4) or to walk sufficiently (Surgeon General’s recommendations) (odds ratio of 6.3) relative to their odds of being a nonwalker. Cycling was also positively related to transit use. 3. Besser and Dannenberg (2005) (see also “Under- lying Traveler Res- ponse Factors” – User Factors”) Descriptive statistics were calculated from the 2001 National Household Travel Survey (NHTS) covering the walking activity involved in access- ing U.S. public transit. (Trips with a 2 nd access mode such as auto, 5% of the transit trips, were excluded.) During their survey day, 3.1% of NHTS respondents walked to/from transit, averaging 19 minutes total daily walk time. Highest odds for being transit walkers were found among lower income, less educated, and non-white populations, and in denser urban areas. 4. Investigation by Handbook Authors – 2008 (see Montgomery Co. case study un- der “Results - Path Connection…”) Prior to mid-1980s the Garrett Park, MD, MARC commuter station was separated by private property from Randolph Hills. A park was created and an 800 ft. paved path built, and later illuminated. In 2008 a 1-day count was made of PM alighting riders. (Prior conditions anecdotal.) Prior to path completion, rail ridership from Randolph Hills was negligible. In the “after” condition, out of 33 alight – ing passengers on 6 outbound trains, 24% walked toward Randolph Hills, 42% walked away into Garrett Park and vicinity, and 33% drove away in cars parked at the station. 5. Project for Public Spaces (1998) Sidewalk area and curb extensions were added along NW 23rd Avenue in Portland, OR, to help address pedestrian congestion at bus stops. (No investigation of travel effects.) Despite interference of street furniture with full use, the bus stop improve- ments (including shelters with seating, trash receptacles, newspaper boxes) received high marks in a user survey. 6. Wilbur Smith and Associates et al. (1996a) Behavioral models describing transit use and transit access mode were developed on the basis of preference and intercept surveys for Chicago’s Metra (commuter rail) and CTA (HRT). (Bus not addressed.) Most pedestrian improvements estima- ted to have positive impact on walk mode access. Significant prior access modes included auto (Metra) and feeder bus (CTA). Estimated up to 7.2% more walk access for Metra. 7. Hsaio et al. (1997) (see this section for more information) Orange County, CA, 1990 on-board transit rider survey used, in conjunc - tion with GIS evaluation of street pattern and land use effects, and 1990 Census journey-to-work mode shares, in a descriptive analysis of walk distance effects on bus mode share. (Observed static situation.) Established that 80% of bus riders walk 1/4 mile or less to/from bus stops. Found that as the proportion of an area within 1/4 mile decreases from 80% to 20%, work purpose trip transit mode share declines from 7.9% to 0.5%. From this result the importance of improved pedestrian linkages was 8. Zhao et al. (2002) (see this section for more information) Miami Dade County transit onboard survey sample used to improve on use of 1/4 mile buffer to describe walk to transit accessibility for resi- dential population. (Static situation, no accessibility analysis differentia- tion between Metrorail and bus.) Found a decay function to best repre- sent transit stop accessibility, applied (up to 1/2 mile) to street network distance from the residence, taking walk barriers into account. Relative to transit trips at 300 ft., siting at 1,200 ft. produced 21% as many, at 2,400 ft., 4%. inferred.

16-131 Table 16-26 (Continued) Study (Date) Process (Limitations) Key Findings 9. Weinstein et al. (2007) (see also “Under- lying… Factors” – Trip Factors”) AM surveys of riders walking into 5 U.S. West Coast rail transit stations asked respondents to trace their route from their origin on maps and inquired about route choice factors. (Observed static situation.) The 25th percentile walking distance was 0.27 miles, the median was 0.47, and the 75th percentile was 0.68 miles. Top ranked route choice factors were distance minimization, followed by safety factors and sidewalk condition. 10. Park and Kang (2008) (see this section both under “Pedestrian Access and Egress” and “Bicycle Access and Egress” for more information) Self-administered survey was hand- ed to transit users entering Moun- tain View, CA, Station from 5:30 to 10:30 AM. Origin, route to station, travel, and socio-economic status information was obtained with a 62% response rate. Binomial logit mode of access models were devel- oped. The walk versus auto model covered commuter rail users living within 1.5 airline miles and the bike model covered users within 2.0 miles. (Walk vs. bike vs. bus trade- offs excluded from consideration, only one station area site studied.) Caltrain user mode of arrival (all dis- tances) was 17% walk, 11% bike, 2% bus, 50% drive alone, and 20% carpool and drop-off. Variables remaining in the final walk model (R2=0.54) were access distance, work purpose, car availability, race (Asian), auto friendly street close by, and (the only positive) 4-way intersection density. Negative variables in the final bike model (R2=0.21) were distance, car availa- bility, and auto friendly street. Posi- tives were male gender and white race. Neither model found significance for income, age, or national origin. Note: Where substantial additional information on individual studies is provided in text and tables or figures, this is noted — and the location within the chapter is given — in the first column. Sources: As indicated in the first column. The notation “SR 282” is shorthand for Committee on Physical Activity, Health, Transportation, and Land Use (2005) together with Handy (2004). The 3rd entry in Table 16-26 provides information supportive of the walking sufficiency findings described in the 2nd table entry. It highlights a study, of particular importance to public health practi- tioners, that found users of transit service in the United States to average 19 minutes total of daily walk- ing time in the course of getting to and from transit stops and stations. As noted already, this same study determined that 29 percent of transit users meet or exceed the recommended 30 minute walk exercise minimum while accessing and egressing their bus or train service (Besser and Dannenberg, 2005). The 4th Table 16-26 entry presents a small count-based analysis to help demonstrate the obvious yet often overlooked role of NMT connections in facilitating transit ridership by expanding service area. The 5th table entry offers no actual travel demand outcome information, but serves to under- score the contribution made to transit walk and bike accessibility by proper bus stop design (Project for Public Spaces, 1998). Construction of suitable bus stop provisions combined with crit- ical links of sidewalk have been shown in specific cases to be quite cost effective when they allow access to conventional transit service by people with disabilities who otherwise would require expensive-to-provide and often time-constraining Americans with Disabilities Act (ADA) door-to- door paratransit service (Goodwill and Carapella, 2008). (See “Related Information and Impacts”— “Economic and Equity Impacts” for cost recovery calculations.) Behavioral model research and application, rather than empirical findings, forms the basis of the 6th entry in Table 16-26. Application of mode of access share behavioral models developed for the purpose, in this Chicago area Metra commuter rail and Chicago Transit Authority (CTA) access study, produced estimates that most pedestrian improvements would have a positive impact on choice of walk mode access. The model results for Metra stations indicated that many of new walk- ers would have previously driven to the station. For rapid transit stations, some of the new walkers

16-132 Table 16-27 Walk Access Versus Journey-to-work Mode Share, Orange County, California Percent of Population within Walking Distance Total Workers Bus Riders Drive Alone Carpool Number Percent Number Percent Number Percent 80 – 100 129,629 10,278 7.9 82,683 63.8 26,958 20.8 60 – 80 213,088 7,013 3.3 160,934 75.5 31,645 14.9 40 – 60 276,417 7,908 2.9 215,343 77.9 38,551 13.9 20 – 40 223,432 3,088 1.4 187,073 83.7 25,154 11.3 10 – 20 166,012 872 0.5 143,160 86.2 16,797 10.1 Total 1,008,578 29,159 2.9 789,193 78.2 139,105 13.8 Source: Hsaio et al. (1997). would come at the expense of the feeder bus service. Extensive pedestrian improvements tested for the commuter rail stations were estimated to induce up to 7.2 percent more riders to choose walking for access (Wilbur Smith and Associates et al., 1996a). The 7th entry in Table 16-26 encapsulates an analysis of the Orange County (California) Transportation Authority’s 1990 on-board survey along with 1990 U.S. Census journey-to-work mode shares by Census tract. The analysis found that more than 80 percent of bus riders were walking up to, but no more than, 0.25 miles to or from the Authority’s bus stops. Using this 0.25 mile threshold as a definition of accessibility, researchers looked at how differences in street patterns or land use char- acteristics impacted pedestrian accessibility to transit. Two areas with irregular street patterns and lower-density land use were compared with two mature suburban areas with regular grid street patterns and higher-density mixed residential and commercial land uses. Bus stops and residences were pinpointed in a geographic information system and distances were measured along road cen- terlines. About 56 percent of the population was determined to be “transit accessible” in the two areas of irregular streets compared to 75 and 81 percent in the two gridded-street areas. Further Orange County analysis found that as the pedestrian accessibility level for an area decreased from 80 to 20 percent, using the 0.25-mile threshold and the street centerline measurement approach, the journey-to-work mode share for transit usage decreased correspondingly from 7.9 to 0.5 percent. This relationship, along with the prime-mode shares for driving alone and carpool- ing, is illustrated in Table 16-27. The analysis was done at the Census tract level of trip data aggre- gation. It was concluded that providing additional pedestrian linkages to enable more direct access to transit “would logically result in increased ridership” (Hsiao et al., 1997). The 8th and 9th entries in Table 16-26 provide two more studies along the same vein. The Miami-based analysis, the 8th entry, found walking to drop off sharply with distance from a bus stop or Miami Metrorail station (the transit system and the study are bus-dominant). It was determined that if transit ridership generation at 300 feet from a bus stop or station is indexed at 100 percent, rider- ship is only 21 percent at 1,200 feet and 4 percent at 2,400 feet (Zhao et al., 2002). The 9th entry illus- trates, on the basis of walking patterns to five rail transit stations in the San Francisco Bay Area and Portland, Oregon, that riders tend to be willing to walk farther to access urban rail stations than bus stops. This study found the median walking distance to urban rail to be almost 1/2 mile (Weinstein et al., 2007). Information from both of these two studies is used in a comparative analy- sis with bicycle access distance within the “Bike-on-Bus Programs” discussion (see Table 16-35).

The 10th entry in Table 16-26 introduces a mode of access modeling research effort that included a review of past mode of access models. The review is summarized in Table 16-28 along with the researchers’ own Caltrain commuter rail access model, derived from their survey of Mountain View, California, station arrivals. Except as noted with respect to race, socio-economic variables proved significant in only one modeling effort—the third utilizing San Francisco BART HRT sta- tion access data—and then only in one of three formulations reported on. In the Caltrain model, Asian race was a negative for walk access and white race was a positive for bike access. Income, age, and United States versus foreign birth were all specifically found not to be significant in the Caltrain research. Auto availability, however, was a negative factor in three of the models, and sta- tion parking supply was a negative in two, indicating that ease of driving dampens non-motorized access choice (Park and Kang, 2008). 16-133 Study/System Korf, Demetsky, and Hoel – 1979 BART Cervero – 1995 BART Loutzenheiser – 1997 BART Cervero – 2001 Wash. Metro Park and Kang (2008) Caltrain Factors Socio-economic variables ±W — — Race (Asian, white) — ±W,B Gender (male) +W +B Trip purpose (work) –W Auto availability –W –W –W,B Station auto parking supply –W –W * Station access distance –W –W –W,B Density/compactness +W +W Land use mix/retail access +W +W Intersection density +W +W Sidewalk/street miles ratio +W Auto-friendly street nearby –W,B Notes: “+” indicates positive factor, “–“ indicates negative factor, “±” indicates sign depends on variable or model. “W” indicates inclusion in walk model, “B” indicates inclusion in bike model, “—“ indicates specifically-reported lack of significance. Bike model significant factors reported only for Park and Kang (2008). * Neither station-auto- nor bike-parking supply could be modeled in a single-site study. Source: Park and Kang (2008), with elaboration by the Handbook authors. Table 16-28 Factors Found in Revealed Preference Modeling to Influence Walking and Bicycling Mode of Access Shares at Rail Stations Station access distance once again shows up as important, being so identified in a majority of the mode of access model sets examined in Table 16-28. Also, all but the earliest model—which was lim- ited by data availability—found at least two indicators of pedestrian/bicycle friendly environment to be significant. In the case of the Caltrain model, one of the indicators is a negative, namely, living within 250 feet of an auto-friendly street—defined as having a posted speed of 35 mph or higher. The

16-134 researchers note that such a street can be either a deterrent to walking or bicycling, or an encourage- ment to drive, or both. The other significant Caltrain model indicator of pedestrian/bicycle friendly environment is the number of four-way intersections per square mile in the home Census tract (Park and Kang, 2008). This value is a surrogate for good connectivity. Also, in Mountain View, the higher densities of four-way intersections occur where traditional neighborhood design, the historic down- town, and a complete grid system of streets and sidewalks are found. The relationships between distance to a transit stop or station and the amount or percentage of walk- transit trip making are so strong that it seems reasonable, even without much before-and-after data, to assume at least some relationship transferability to assessment of effects of shortening transit access distance. Shortening pedestrian access distances to transit through introduction of good pedestrian linkages should logically increase walk-transit trip making.31 Bicycle Access and Egress. Bicycle trips taken in conjunction with transit use constituted, circa the year 2000, roughly 1/10 of 1 percent of all trips taken in the United States and perhaps 10 percent at most of all bicycle trips. The derivation of this estimate is presented (in conjunction with Tables 16-88 and 16-89) in the “Related Information and Impacts” section under “Extent of Walking and Bicycling”—“Extent of Bicycling.” Relatively little empirical study has been done on U.S. bicycle access and egress to/from transit in and of itself. There is additional information, however, on response to allowing bicycles on buses and rail transit vehicles, presented under “Bicycles on Transit Vehicles” following the “Transit Oriented Development” discussion. Table 16-29 presents a modicum of evaluation findings concerning bike-to-transit activity. In addi- tion to Table 16-29, the 2nd and 10th entries in Table 16-26 included bike-to-transit research in con- text with walk-to-transit findings. These research efforts are reexamined from a bicycle perspective after review of Table 16-29 entries. The 1st Table 16-29 entry covers a stated preference experiment aimed at determining the relative effectiveness of bike lanes, lockable covered parking, and bike lockers as means for attracting more usage of bicycling for transit access. Lockable covered parking was estimated to be 40 percent as effective as bike lockers, which were felt by frequent cyclists to be more important than either wide curb lanes or bike lanes on access routes. Infrequent cyclists, however, felt the access improve- ments were more important than lockers (Taylor and Mahmassani, 1996). 31 This is essentially the thought process employed by the authors of the Orange County, California, analysis described above in deriving their conclusion about the benefit of pedestrian linkages (Hsiao et al., 1997). The primary limitation in this argument is the extent to which those desirous of using transit may have deliber- ately “self-selected” their residence location to be close to a stop or station. The phenomenon of self selec- tion is examined in the “Underlying Traveler Response Factors” section under “Choice of Neighborhood/ Self-selection.”

16-135 Table 16-29 Summary of Research Findings and Other Studies on Relationships of Bicycle Access Quality and Bicycle Parking with Bike-Transit Activity Study (Date) Process (Limitations) Key Findings 1. Taylor and Mahmassani (1996) (see “Point-of- Destination Facilities” for more information) Conducted a stated preference experiment with hypothetical transit access scenarios for commuting to work and developed a nested logit model from the auto-only, auto- park-and-ride, and bike-and-ride preferences expressed. (Conven- ience sample, mostly avid cyclists.) Provision of a bike lane slightly more important than bike lockers for infre- quent cyclists; for frequent cyclists either wide curb or bike lane was 30% as important as lockers. Lockers esti- mated to encourage bike-and-ride over driving to transit or all the way. Lock- able covered parking 40% as effective. 2. Wilbur Smith and Associates et al. (1996a) (see this section for more information) See Table 16-26 for survey/ forecast process description. (In behavioral model development, the influence of bike paths, lanes, and routes was found not to be statistically signifi- cant — possibly reflecting in part baseline bicycle-friendliness of the station areas. Significant effects were estimated for “wide curb lanes” but these were undefined in the survey, reducing outcome significance) Wide curb lanes and most especially secure bike parking were determined to have a positive influence on bike access choice. For access trips from home originating within 2 miles of stations, improvements were predicted to increase bike shares from 2.1% to 3.2% for Metra and from 0.5% to 6.5% for CTA. The overall Metra/CTA bike access share was estimated to increase from less than 1% to nearly 1.5%. 3. Replogle and Parcells (1992) (see this section for additional inventory example) In July 1979, bicycle racks for 457 bikes were added to 9 commuter rail stations near Chicago to help miti- gate Edens Expressway traffic. Bike counts were made at 88 stations in 1990. (No in-depth analyses.) In August 1979, 222 bicycles were parked in the new racks. In 1990, 88 Metra stations had bike parking. Parked there in designated locations were 564 bicycles, plus 245 more were seen locked to poles, trees, signs, etc. 4. RTC and APBP (1998), Bikestation (2003) Long Beach, CA, has a Bikestation on the transit mall at the end of the Blue Line LRT to Los Angeles with free valet parking for 150 bicycles, rentals, repairs, etc., convenient to over 30 miles of bike paths/lanes. (See next column for limitations.) In its first 18 months of operation, the facility parked about 1,500 bicycles per month, increasing at a rate of about 10% per month. (Proportion of Bike- station use associated with LRT or local bus system usage was not reported.) 5. Replogle (1993) (see this section for more information) Concurrent with suburbanization, Japan constructed thousands of miles of bicycle paths and lanes and millions of public and private bike parking spaces at rail stations, totaling 2.77 million spaces in 8,735 facilities by 1989. (Few details.) Despite rising auto ownership, use of bicycles to access rail stations grew from some 300,000 in 1975 to more than 3,000,000 in 1989. Ten percent of rail riders bike to transit stations, with some stations experiencing as much as 50 percent bicycle access. Note: Where substantial additional information on individual studies is provided in text and tables or figures, this is noted — and the location within the chapter is given — in the first column. Sources: As indicated in the first column. Chicago area behavioral modeling of rail transit access choices, based on surveys of existing options, conditions, and choices, is summarized in the 2nd Table 16-29 entry. Again, parking pro- visions exhibited primary importance, with the estimated opportunity to raise overall Metra com- muter rail and CTA subway/elevated bike-to-rail shares by almost 50 percent (Wilbur Smith and Associates et al., 1996a). Bicycle count comparison data presented as a supplement to the 4th entry in Table 16-29 suggest that Chicago’s Metra has accomplished more than that in 10 to 15 years.

Additional details of the Chicago area rail transit access behavioral modeling are of interest for the insights provided on primary mode and access mode shift potentials. The separate models for Metra and CTA, when applied for improved bicycle parking and access, predicted Metra bike mode of access choice to increase from 2.1 percent to 3.2 percent for home-based trips originating within 2 miles of the station. For CTA stations, the predicted increase was from 0.5 percent to 6.5 percent. The analysts did not find that the improvements would affect prime mode choice, that is, induce new transit riders. Instead, the changes in access mode share were estimated to accrue from diversions from other modes of access. For Metra, which has extensive drive access (68 per- cent at the time), about one-half the diversion was from drivers. For CTA, with primarily walk access (44 percent), the majority of the diversion (about 80 percent) was from walkers. The poten- tial for diversion diminished as the origin distance from the station increased. Although for many reasons it is desirable to attract bicyclists from the drive access mode, the study found that bicy- clists generally were more likely to be diverted from pedestrian access, there being more in com- mon between these two modes than between bicycling and driving a car (Wilbur Smith and Associates et al., 1996b). The 3rd and 4th entries in Table 16-29 offer usage-count evidence that moderately-sized bicycle parking facilities will receive worthwhile usage when provided at U.S. rail stations and transit ter- minals, but the information provided does not lend itself to more expansive conclusions. However, when the report that the Metra system in 1990 parked just over 800 bicycles at stations offering spaces—with 30 percent of the bikes not in designated spots (Replogle and Parcells, 1992)—is con- trasted to information that Metra circa 2008 provided 4,257 bicycle spaces (Pucher and Buehler, 2009b), strong continued bike-to-Metra growth is indicated. With respect to bike stations, the subject of the 4th table entry, 5 of the 10 bike stations in the United States as of 2009 were at San Francisco Bay Area rail transit stations. They provide a total of 433 spaces at three BART stations and 226 spaces at Caltrain commuter rail stations. Reported utilization rates range from over 100 percent at the BART station in Berkeley down to 11 percent at the Caltrain station in Palo Alto. The Berkeley bike station was slated for a tripling of capacity (Pucher and Buehler, 2009b). Another example of inventory-type information was developed for Miami Metrorail HRT stations and bus park-and-ride lots, in the 2001–2002 period, as a precursor to improvements. At the 21 Metrorail stations, a 3-day average of 122 bicycles system wide was observed in racks or informally parked. In addition, 53 bike lockers were rented, relative to 111 undamaged lockers available out of a total of 246 mostly installed in 1986. It was judged that 170 to 180 patrons used Miami Metrorail bicycle park- ing daily. A grand total of two parked bicycles were observed at the 10 bus park-and-ride lots. A survey, offered in three languages, was given to all the Metrorail bicycle parkers who would vol- unteer and 72 responses were obtained. Racial minorities and persons of Hispanic origin made up 48 percent. Males were over 85 percent of the total, median age was in the 40 to 59 bracket, and almost 60 percent who gave their income earned less than $30,000 per year. Interestingly, the other heavily represented income grouping, at over 20 percent, was persons earning more than $70,000 per year. Bike-transit riding by the Miami Metrorail survey respondents was very regular, with a median frequency of 5 days per week. Some of this usage was bike-on-transit, with 40 percent of respondents sometimes taking their bike on Metrorail (allowed only off-peak) and just over 20 per- cent reporting use of the Bikes-on-Bus program. Having had a bicycle parked at Metrorail stolen and/or vandalized was reported by over half and the highest level of importance was given to hav- ing all bicycle parking in view of staff/security (Hagelin, 2002). BART 2009 inventory data illustrate the other end of the U.S. HRT spectrum from the early 2000s Miami Metrorail situation. In addition to its 246 bike lockers, not all usable, Miami Metrorail pro- 16-136

vided racks at a number of its stations: 22 racks in total (Hagelin, 2002). Assuming a rack capacity of a dozen bicycles each, Miami Metrorail would thus have had a theoretical total of about 500 bicy- cle parking spaces at its 21 stations, an average of 24 per station. BART in 2009 had over 4,300 bicy- cle parking spaces in total located at most of its 43 stations, an average of 100 per station, 1/4 in secure lockers. Adding in bike-station spaces brings the 2009 BART average to 110 per station (Pucher and Buehler, 2009b). To some extent this is a result, of course, of the much higher BART ridership per station. The 5th and final Table 16-29 entry illustrates the substantially greater growth and usage (in absolute terms) of transit-related bicycle facilities in Japan, with a 10 percent reported bicycle access mode share for rail stations in 1989, and some stations experiencing as much as 50 percent bicycle access (Replogle, 1993). In the Netherlands, some 35 percent of transit patrons overall use bicycles for access. The typical station with over 5,000 boardings per day might have 2,000 guarded bicycle parking spaces versus only 250 spaces for automobiles (Replogle and Parcells, 1992, Wilbur Smith and Associates et al., 1996a). Turning back to bicycle access information included in with the “Pedestrian Access and Egress” discussion, the 2nd entry in Table 16-26 provides an example of overall cycling levels found to be related, positively, to higher use of transit, in this instance in the greater Seattle area (Moudon et al., 2005). The 10th entry in Table 16-26 is of special interest as a rare example of a reasonably successful attempt to model bicycle mode of access to transit with revealed preference survey data, notwithstanding the somewhat disappointing explanatory value of the model (R2=0.21) compared to the companion walk mode of access model (R2=0.54). The model development was based on surveyed Caltrain commuter rail rider access to the Mountain View, California, station from surrounding residences. In terms of socio-economic variables, the research found no significance for income, age, or national origin, although male gender and white race were positive factors. Higher car availability was a negative indicator for bike access. The greater the distance of the home from the station, the lower was the probability of bike mode of access choice. One built environment variable was found to be significant. Although residence within 100 feet of a bike lane was one of several built environment variables that had to be dis- carded, living within 250 feet of a street with fast-moving traffic was a significant negative factor for bike mode of access share. The constraints of the single-site study were such that effects of sta- tion automobile parking supply could not be examined (Park and Kang, 2008), and neither would modeling of bike parking supply importance have been possible. Figures 16-4 and 16-5, presented earlier, illustrate the drop-off in bicycling mode of access shares with distance using BART HRT data for weekday work-purpose trips. Figure 16-4 pertains to non- CBD urban BART stations in San Francisco, Berkeley, and Oakland. The primary market for bicy- cle access under these conditions can been seen to be between 3/8 and 1-1/2 miles from the station. Figure 16-5 pertains to suburban center stations and shows the primary bicycle access market there to be between 1/4 and 1-1/8 miles from the station. The bicycle access mode share picks up again after 2-1/4 miles, but applies to a diminishing absolute number of rail transit riders that far out from the station (Parsons Brinckerhoff et al., 1996b). The Mountain View Caltrain survey, cover- ing all commuter rail trip purposes, found bicycle mode of access shares to be above 10 percent (considering walk, bike, and auto only) between 1/2 mile and 2 miles of the station. The bike share peaked between 1 and 1-1/4 miles at about 38 percent, and was no less than 5 percent for a dis- tance of 5 miles out from the station (Park and Kang, 2008). In the 1990s and first decade of the 21st Century, many U.S. transit operators—bus operators in particular—placed more emphasis on allowing patrons to put bicycles onto transit vehicles than 16-137

16-138 on parking them at ordinary transit stops. A return-on-investment analysis and related studies have found that despite considerable bike-on-transit activity, investment in bike parking at bus stops and even many rail stations has been low. Bike parking is receiving renewed interest, how- ever, at agencies where bike-on-transit ridership is exceeding the practical bike capacity on bus bike racks and train cars. Between 2006 and 2008, the number of bike parking spaces at rail stations and bus stops grew by 26 percent in the United States and 67 percent in Canada, with the 2008 U.S. inventory totaling 24,178 at rail stations, 9,005 at bus stops, and 176 at ferry terminals (Hagelin, 2005, Pucher and Buehler, 2009b). Bike-on-transit experiences are presented immediately follow- ing the “Transit Oriented Development” discussion. Some of these experiences suggest that oppor- tunities for shifting riders from bike-on-transit to use of station and stop bike parking facilities may have their limits, particularly when both ends of the trip are in spread-out suburban development. Transit Oriented Development Perhaps the best way to enhance pedestrian and bicycle access to transit is by placing greater num- bers of residences and activities within reasonable walking distance of a transit station or stop, all in a pedestrian-friendly environment. This is the thrust of transit oriented development (TOD), which generally refers to higher density development, with pedestrian priority, located close around a major transit station or stop. TOD should be good for bicycle access as well, although the short access distances offered tend to make walking the dominant TOD transit access mode by far. TOD is primarily covered within Chapter 17, “Transit Oriented Development,” of this TCRP Report 95: Traveler Response to Transportation System Changes Handbook. A summary of key comparative observations from Chapter 17 pertaining to impacts of TOD on amount of active transportation is provided in Table 16-30. Prime-mode-share findings are addressed first in discussion of the table, followed by examination of mode-of-access share observations.

Prime Mode Share Observations. The 1st and 2nd studies listed in Table 16-30 are each compar- ative in structure. They are divided as to whether NMT-only travel is more prevalent in TODs than other areas, but note that the 1st study addresses work purpose travel only, and the 2nd addresses non-work travel only. The Pleasant Hill, California, transit-adjacent project comparison with nearby conventional suburb Walnut Creek illustrates an instance where, for work-purpose trips, there is little difference between a TOD and conventional suburbs in selection of the walk-only 16-139 Table 16-30 Summary of Primary Comparative Observations from Chapter 17 on Impacts of Transit Oriented Development (TOD) on Non-Motorized Travel Activity Source (Date) Process (Limitations) Key Findings 1. Lund, Cervero, and Willson (2004) and 2000 U.S. Census SF3 data (see Chapter 17, “Response by…” — “…Regional Context” — “Sub- urban TODs”) Current mode shares and mode of access shares were obtained from self-administered surveys at 4 Pleas- ant Hill (PH) station area projects. Mode share for Walnut Creek, CA, as a whole was from U.S. Census. (Comparative commute data sources, methods, and definitions not fully consistent.) Work purpose walk-only prime mode share of 2.3% for PH projects was little different from Walnut Creek overall. However, rail-transit PH-project mode share of 44.3% versus 13.5% for Walnut Creek (bus use nominal), with 96% walk mode of access to rail for PH projects, suggests multiple times as much walking as part of the commute. 2. Derived from data in Evans and Stryker (2005) (see Chapter 17, “Response by…” — “…Regional Context” — “City Center Versus Suburban TOD Comparisons”) Analysis of non-work travel from Portland Metro 1995 regional travel survey. TOD-like traffic analysis zones (TAZs) were identified using professional judgment. Most 1995 TODs were bus-only. (Demographic effects, not isolated out, pertained mostly to household size. Too few central area non-TOD TAZs for direct central area comparisons.) Home-end walk/bike-only non-work shares of 33% for central area TOD and 14% for outlying TOD vs. 8% for non- TOD (mostly outlying). Non-home- end NMT shares of 18% for both cen- tral area and outlying TOD vs. 8% for non-TOD. Non-work transit shares (generally associated with walking) of 7% to 8% for central area TOD and 2% for outlying TOD vs. 1% for non-TOD. 3. Lund, Cervero, and Willson (2004) (see Chapter 17, “Response by…” — “Response to TOD by Land Use Mix” — “Residen- tial”) Self-administered survey responses were obtained from 624 households in 26 station-area projects in CA. For 15 projects throughout the state, commute mode shares were compared with Census data for the surrounding 1/2 to 3 mile “donut.” (Survey response rate was 13%.) Station-area resident transit use for commuting was 20% above that for surrounding “donuts,” with a station- area range of 36% above to 8% below the surrounding area. Over 90% of station area residents in the overall survey reached their neighborhood rail station by walking. 4. S.B. Friedman & Company et al. (2000a and 2000b) (see Chapter 17, “Underlying… Factors” — “Land Use…” — “…Sup- portive Design”) Self-administered passenger surveys were obtained at 6 high-ridership Chicago-area commuter rail stations, with an overall survey response rate of 32%. (Station access trip length was self-reported.) From 0.0-0.5 miles to station, 82% walked to train; from 0.5-1.0 miles, 41% walked; from 1.0-2.0 miles, 8% walked; from more than 2.0 miles, 1% walked. High ridership apparently related to good pedestrian environment and concentrated station-area development (including stores). 5. Dill (2006a and b) (see Chapter 17, “Related Info…” — “Pre- and Post- TOD Travel Modes” — “Disag- gregate Mode Shifts…”) Self-administered survey returns from residents at 8 different TOD and transit-adjacent developments at 4 stations on Portland’s LRT Blue line, with individual-development survey response rates ranging from 24 to 43%. (Questions about travel mode at prior residence were dependent on respondent recall.) Walk-only commute mode shares increased by 38% (2+ percentage points), with no change in bicycling, upon moving to TOD and transit- adjacent developments. Transit commute shares increased by 156% (16 percentage points). Of transit commuters, 69% to 100% walk to the station, depending on the location.

commute mode. In the Pleasant Hill case, a full range of TOD amenities is apparently found only for the specific projects constructed on former park-and-ride lot land areas. The Portland, Oregon, analysis (the 2nd study listed) examined 1995 non-work travel rather than commute travel, and found major differences in the relative walking/cycling prime mode shares between TOD and non-TOD areas. For example, in outlying Portland TOD areas choice of walk- only and cycle-only modes for non-work travel was roughly twice as likely as in non-TOD areas, taking into account the travel associated with both residences and other land uses. It is important to emphasize that potential socio-demographic effects were not explored in either of these analyses. (See Table 16-30 for sources.) A different approach to assessing TOD impact on choice of the walk-only and bike-only travel modes, and one that has lesser (or at least largely different) methodological issues, is to find out how TOD dwellers have changed their travel mode choices relative to the choices they made at their prior abode. Chapter 17 summarizes the four such studies known to be available of prime- mode shifts by residents upon moving into TODs. (See “Pre- and Post-TOD Travel Modes” under “Related Information and Impacts” in Chapter 17.) The newest and most comprehensively reported of these is encapsulated as the 5th entry in Table 16-30. In this study of 8 Portland TODs, walk-only mode commute shares increased by 38 percent, from 4.7 to 7.0 percent of work-purpose trips. Bicycling for the commute remained unchanged at 1.4 percent (Dill 2006a and b). A smaller Portland study reported on mode shifts upon moving into mostly below-market-rate, seniors-oriented housing in the Center Commons TOD. The proportion of bike- and walk-only trips to work actually dropped from 9 percent to 3 percent, apparently reflecting a decrease in the proportion of work trips under 5 miles in length. Bike- and walk-only trips for non-work purposes increased from 5 percent to 6 percent (Switzer, 2002). The other two studies, covering transit adja- cent development in California, including but not exclusively reflecting development with full TOD characteristics, appear to suggest that net effects on prime-mode-choice NMT use for com- muting are not significant (Cervero, 1993, Lund, Cervero, and Willson, 2004). This mix of circumstances and findings provides insufficient basis for the drawing of strong conclu- sions. The effects of TOD projects on enhancing walk-only and bike-only travel mode shares appear to be highly dependent on TOD land uses, design, and stage of development, and on whether or not new residents are attracted from neighborhoods where they had been within walking distance of work and other activities. However, since well executed TODs are pedestrian-friendly by design, the conclusions reached from more extensive research on NMT activity presented in the “Pedestrian/ Bicycle Friendly Neighborhoods” subsection below can (with reasonable caution) be taken to apply for prime-mode walking and cycling in TODs as well. Doing so, the possible tendency seen for higher walk-only mode shares in the majority of the individual and collective California and Oregon TODs covered in Table 16-30 entries gains additional support. Mode of Access Share Observations. Up to this point in discussion of Table 16-30, the focus has been on walk-only and bike-only trips viewed from the perspective of prime mode choice. Focusing exclu- sively on the prime mode overlooks the crucial walk activity gain and auto use reduction impact of shifts to walking as the transit access mode for most TOD residents using transit. Such shifts are only identifiable if the access mode component of the transit trip is looked at separately. The 1st, 3rd, and 5th studies summarized in Table 16-30 illustrate substantially more transit use in TODs than in nearby non-TOD areas or in the prior-to-TOD residential locations. These same studies also found very high walk access-to-transit shares. The Pleasant Hill station area development projects showed over 3 times the BART rail commute mode share as the comparison area, and a 96 percent walk access share to the station. The average walk access share found within 1/2 mile for 26 California station 16-140

areas was over 90 percent, and the walk access shares for eight Portland TODs ranged from 69 to 100 percent (Lund, Cervero, and Willson, 2004, Dill, 2006b).32 The 4th study in Table 16-30 illustrates, using Chicago commuter railroad passenger survey responses, how the inherent transit adjacency of TOD produces enhanced walk access shares. The walk proportion found at the six stations studied increased from 8 percent for a 1.0 to 2.0 mile dis- tance from the station, to 41 percent for 0.5 to 1.0 miles, and 81 percent for under 0.5 miles (S. B. Friedman & Company et al., 2000a and 2000b). Figures 16-4 and 16-5, presented above, illustrate sim- ilar relationships for BART HRT in the San Francisco region. TODs thus have enhanced walking activity and diminished auto use for two reasons in addition to pedestrian friendliness: 1. Shifting from auto to transit use, which in turn comes with substantially higher levels of walking for transit access in the transit-adjacent TOD context. 2. Shifting from motorized means of transit access to walk access for those who would be using transit even if they were not located in a TOD. The second-listed phenomenon is particularly well illustrated in a previously mentioned compara- tive study of the BART HRT tributary areas of the pedestrian-friendly Rockridge neighborhood in Oakland, California, and the more auto-oriented nearby city of Walnut Creek. The commute travel share attracted by BART was essentially identical in the two BART station areas (21 versus 20 per- cent, respectively). The BART-access walk share of 31 percent in Rockridge was, however, well over twice the share in Walnut Creek (13 percent) (Cervero and Radisch, 1995). (For more on this study see Table 16-39, 14th entry, in the “Pedestrian/Bicycle Friendly Neighborhoods” subsection.) Bicycles on Transit Vehicles There is no recent assemblage of nationwide U.S. comparative statistics on the transporting of bicy- cles on urban public transit vehicles (Pucher and Buehler, 2009b). It is, however, informative to examine indications that can be extracted from an advocacy-inspired comparative tabulation based on U.S. data from the 1996–2004 period, roughly halfway along in the development to date of bike- on-transit service. In this data set, bicycle carriage data were seasonally adjusted, and gaps were filled by use of similar-city bike-share analogies (Steve Spindler Cartography, 2010). The top performer in terms of bike-on-transit share of unlinked trips was Caltrain, the commuter railroad serving the San Francisco Peninsula including Silicon Valley, at 6.2 percent in 1997.33 (Another source, covered below in examining Caltrain results, puts the 1997 proportion at 7.5 per- cent.) The 75th percentile bike share performer, at 1.8 percent in FY 2000, was the Chittenden 16-141 32 These are not walk access shares for all users of the HRT, LRT, and commuter rail stations involved. They are the walk access shares for station use by persons living within 1/2 mile (26 California station areas study) or within specific projects or TODs adjoining or surrounding the stations. 33 An “unlinked trip” is a trip maker’s ride (or carriage of a bicycle) on an individual transit vehicle between the boarding point and the alighting point, even if the boarding and/or alighting point represents no more than a transfer from/to another train or bus. Since some bike-on-transit trip makers are using the service to avoid transfers, unlinked-trip bike shares may slightly understate bike-on-transit usage from a “linked trip” perspective, which “links” trips at points of transfer to allow examining whole one-way trips from point of first entry onto the transit system to point of final exit from the system.

County Transportation Authority bus system in Burlington, Vermont—illustrating that location in the sunbelt is not a prerequisite for above-average attraction of bike-on-transit riders. The two operators at the 50th percentile (median) bike share of 0.7 percent, for which there were hard data, were Delaware Transit Corporation’s Resort Bus Service (2001 data)—operating 4 months of the year—and King County Metro (no date given), greater Seattle’s public bus service provider. The 25th percentile system, at 0.8 percent in May, 2003, was PACE, the Elgin, Illinois, bus operator serv- ing the exurbs of Chicago (Steve Spindler Cartography, 2010). About the only service-area type that has seen no penetration of bike-on-bus is the most dense of cities, where walking is the pri- mary access and egress mode, such as for bus service internal to San Francisco proper and New York City (Pucher and Buehler, 2009b). Another perspective is offered by looking at the systems with the largest numbers of bike-on-transit trips reported. Table 16-31 lists the top U.S. carrier of bicycles for each primary transit mode in the 1996–2004 data listing described above. The two highest carriers of bike-on-bus trips (with hard data) are both included to provide a bus and LRT comparison within one service area, Santa Clara County, California, served by the Valley Transit Authority (VTA) of San Jose. In this tabulation the dominant bike-on-transit carriers are all found to be in the Sunbelt. This may be as much a reflec- tion of the spread-out nature of the major Sunbelt cities or suburbs as it is a reflection of the warmer climate the Sunbelt is famous for. 16-142 Table 16-31 Top U.S. System(s) in Bicycle Boardings per Each Transit Mode, Circa 2000 Agency Headquarters Mode Date Monthly Boardings Bike-on- Transit Share Valley Metro Phoenix, AZ Bus FY 2002 85,000 2.0% VTA San Jose, CA Bus n/a 77,800 2.0% VTA San Jose, CA LRT 2001 21,200 3.2% BART Oakland, CA HRT (Metro) n/a 36,800 0.5% Caltrain San Carlos, CA Commuter Rail 1997 45,000 6.2% Notes: A possible top performer, the Los Angeles Metropolitan Transit Authority bus operation, is omitted because the listed underlying bike share had to be estimated by analogy. The Valley Transit Authority (VTA), Bay Area Rapid Transit District (BART), and Caltrain data were estimated on the basis of one-day bike-on-transit counts and annual unlinked trip reportings, with applicable conversions. n/a = Date not given, but presumably from the 1996-2004 period. Sources: Steve Spindler Cartography (2010), Hagelin (2005). It is relevant to note that BART, in the San Francisco region, does not allow bicycles on board peak- period, peak-direction trains (Pucher and Buehler, 2009b). More recent statistics for Caltrain of the San Francisco Peninsula, and Valley Metro of Phoenix, are found in the bike-on-bus and bike-on- rail discussions to follow. Table 16-32 summarizes information on these and other bike-on-transit programs selected for their experiences offered and information availability. In reviewing Table 16-32, note that the 1st, 6th, 7th, and 8th table entries, taken together, suggest that where both bike- on-transit and bicycle parking offer viable options, from three out of four to nine out of 10 persons arriving by bicycle apparently prefer bike-on-transit over leaving their bicycles parked at stops and stations (NuStats, 2009, Eisen⎟ Letunic and Fehr & Peers, 2008: Pucher and Buehler, 2009b).

16-143 Table 16-32 Summary of Before and After Studies and Research Findings on Relationships of Accommodations for Bicycles on Transit Vehicles with Bicycling Activity Study (Date) Process (Limitations) Key Findings 1. Zehnpfenning et al. (1993), Doolittle and Porter (1994), RTC and APBP (1998), NuStats (2009), Valley Metro (2010) (see this section for more information) Valley Metro in Phoenix transition- ed directly from a 1991 bike-on-bus demonstration, with surveys, to systemwide implementation by mid-1992. Annual boarding counts are readily available from FY 2000- 2001 to present. Origin-destination (O-D) survey taken in 2007. (Available bike-on-bus rider analysis largely limited to demonstration period.) Early full-system usage was some 1,000 bicycles boarded/day, roughly 1% of boarding passengers. Usage increased to 3,200/weekday (2.0%) 10 years later in FY 2002-03, and over 4,600/weekday (2.2%) in FY 2008-09. The O-D survey found 4% bike access among bus riders, with 3 out of 4 bike access trips reporting bike egress (presumably via bike-on-bus with the rest bike-and - park). 2. Newman and Bebendorf (1983), Coverly (2010) (see this section for more information) A Santa Barbara, CA, bus bicycle- trailer demonstration program with data collection and various surveys was initiated in 1978. Subsequently, bike-on-bus service was off-and-on or partial until 2001. (Late 1970s bus riding increases were attributed mainly to the concurrent gas crisis.) The college-focused bus bicycle-trailer services carried 42,463 bicycles in FY 1980-81, 60% students. Primary travel effects were access mode shifts from walk to bike access and mode shifts from cycling all the way to bus bicycle- trailer use. FY 2001-02 front-mounted bus rack usage was 52,736 bicycles. 3. Hagelin (2005) (see this section for more information) An examination of 15 Bikes on Bus (BOB) programs included obtaining performance measures for 3 Florida systems and user surveys for 2 of these plus 1 other. (User survey success rates were 11% to 14%.) The BOB share of all unlinked transit trips ranged from 0.25% to 1.61% for the 3 systems (Table 16-34). Among BOB users 1 in 4 was new to transit, and for over 80% of these, BOB service availability prompted the switch. 4. American Public Transit Association (1997), SunLine (2003) SunLine operates a regional transit service (“SunBus”) over nearly 1,000 sq. miles in California’s Coachella Valley. In 1997, 2-bike racks were in place on the front of all 40 buses. (No formal analysis.) The system began carrying 6,000 bikes per month. Riders were a combination of both commuters and visitors. In 2002, SunBus operated 46 buses on 13 routes and carried some 3,700,000 riders and 61,300 bikes (1.7% share). 5. Boyle (2002), Hagelin (2005) (see this section for more information) Denver’s Regional Transit District did a study of bicycles-on-bus in the summer of 1999 that included a user survey. Most RTD buses were equipped with front-mounted bike racks by that time. (Summer focus.) There were some 2,300 bike-on-bus boardings each weekday (about 1.4% of all summer bus boardings). The most popular routes linked to the City of Boulder, home of the University of Colorado. (See text for survey results.) 6. Tannen (2010), Eisen Letunic and Fehr & Peers (2008) (see this section Caltrain commuter rail service from San Jose to San Francisco initiated bike-on-rail with its 1992 demon - stration. Stowage was expanded in 1994 with onboard racks. Counts of In September 1997, 1,960 riders (7.5%) boarded bicycles. Mode of access in February 2007 was 7% bike-on-rail and 1% bike-and-park. Prior to ±10:30 AM, 500 NB and 424 SB bikes were boarded. for more information) 1997 and 2007 have been analyzed. (No mode choice effects analysis.) Onboard demand over capacity, but station bike parking 55% of capacity. 7. Doolittle and Porter (1994), TriMet (2009), Pucher and Buehler (2009b). (see this section for more information) In early 1990s, Portland, Oregon’s TriMet equipped buses with bicycle racks systemwide and allowed bikes on its single LRT line. By 2009, with 4 LRT lines, a low-floor car with 4 bike hooks was on each LRT train (not the case at first). Bus and LRT bike use is typically reported in combination but LRT sample counts are taken. (Limited information.) Early in the program about 80,000 bus and LRT bicycle boardings/year were observed, with bike-on-LRT counts in the range of 70/day in Sept. 1993 and 60/day in Feb. 1994. As of 2009 staff estimated 2,100 daily bicycle boardings on LRT versus 200 parked at stations. A TriMet survey found 76% of cyclists unwilling to park their bikes at transit stops even with security and shelter. (continued on next page)

Bike-on-Bus Programs. The Phoenix, Arizona, area bike-on-bus program, overseen by the Valley Metro Regional Public Transportation Authority, is of special interest given both data availability and the consistent offer of bike-on-bus service for almost two decades. In addition, based on circa 2000 national statistics, it has been identified as one of if not the largest carrier of bicycles of any U.S. transit system (Steve Spindler Cartography, 2010). The 1st entry in Table 16-32 addresses the Valley Metro experience. The predecessor to the Valley Metro Regional Public Transportation Authority in Phoenix began their bikes on buses involvement in 1991 with a demonstration program. This pilot project put front-mounted racks on 40 buses operating on three routes. Over 5,500 bicycles total were carried during the 6-month test. In surveys, some cyclists indicated they would not otherwise have ridden the service. The successful test led to system-wide implementation, complete by mid-1992, and a reported 1,000 bicycle boardings out of a total of 104,000 passenger boardings each day. Bike on bus was found to be used primarily for commuting to work, but multiple trip purposes were reported. The average origin-to-destination trip length was 7 miles, including a 9 minute bicycle ride to the bus, a 41 minute bus ride, and an 8 minute bicycle ride to the final destination. (The essentially equal bike-ride times at each end of the trip conflict with Florida access versus egress distance findings discussed in connection with Table 16-35 and may be an artifact of analy- sis in origin-destination trip format). Non-bike-riding passengers were found to be not unduly inconvenienced: 89 percent reported seeing no delays on routes with bicycle racks. Experience from elsewhere suggests that most delays associated with bike-on-bus are caused not by the load- ing and unloading of bicycles from the racks, but from disputes over how to handle situations when the rack is already full (Zehnpfenning et al., 1993, Doolittle and Porter, 1994, RTC and APBP, 1998). Table 16-33 provides Phoenix bike-on-bus and total ridership data for the initial decade of the 2000s, with abbreviated data for alternate years. 16-144 Table 16-32 (Continued) Study (Date) Process (Limitations) Key Findings 8. Pucher and Buehler (2009b), Steve Spindler Cartography, (2010) Bay Area Rapid Transit (BART) focuses heavily on providing bike parking at stations, with 1,010 bike lockers and 3,303 other BART bike spaces at 43 stations in 2009. Bikes may be brought on-board except on peak-direction trains during peak hours. (Limited information.) Data from 1996-2004 placed BART at the top of U.S. HRT Systems for bike- on-rail riders, with 36,800/day and 1/2 of 1% of passengers bringing a bike on board. Despite the heavy emphasis on bike parking, a 2008 survey indicated that 72% of bike-and-ride passengers bring their bike on the train. 9. Pucher and Buehler (2009b), Handbook authors’ comments The New York MTA in theory allows bikes on its subway trains at all times, but access at 84% of all stations requires carrying the bike on stairs. (No usage data.) Bike-on-rail use of New York subways is probably on the order of 0.02% of ridership based on comparisons avail- able in the previously cited 1996-2004 data listing utilized for Table 16-31. Notes: No system charged for bicycle carriage except for pre-1977 experimentation in Santa Barbara. Where substantial additional information on individual studies is provided in text and tables or figures, this is noted — and the location within the chapter is given — in the first column. Sources: As indicated in the first column.

The data in Table 16-33 is indicative of a plateauing from 2002 to 2006 in growth of the percentage of Phoenix passengers boarding with bikes. This plateauing, at about 2 percent, is in turn sugges- tive of either full market penetration or capacity constraint effects. The uptick in FY 2007–2008, continuing into the following year, could possibly reflect easing of capacity constraints with intro- duction of three-bike racks. Santa Barbara, California’s mid-1970s bike-on-bus pilot-then-demonstration project, the 2nd entry in Table 16-32, tested the adding of bicycle trailers to minibuses in a college-town context. The full-scale demonstration involved six trailers deployed on various routes, with emphasis on service to colleges and universities (Coverly, 2010). Not all bus trips on all routes were served. Bicycle trailer scheduling was hourly except for added service toward the end of the demonstration on the route with highest usage. Most successful was an express route serving a residential community, the downtown transit center, and the University of California Santa Barbara with 11 designated bike bus stops in approxi- mately 17 miles. Usage was boosted by high gasoline costs related to the late 1970s fuel crisis (Newman and Bebendorf, 1983). Overall bus ridership increased from 153 to 487 weekday passengers (up 218 percent) on the bicycle- trailer routes during the time period from November 1978 to November 1979. Bicycle-trailer services increased by 205 percent over the same period and the number of passengers with bicycles increased from 11 to 76 per weekday (up 591 percent). In the next 12 months, between November 1979 and November 1980, the level of bicycle-trailer service was increased by another 23 percent, while the number of passengers with bicycles increased from 76 to 174 per weekday (up 129 percent). Passengers with bikes as a percent of total ridership over the full demonstration period averaged 20 percent except for lower usage between December and March (Newman and Bebendorf, 1983). Annual Santa Barbara bike trailer usage in FY 1980–1981, around the end of the demonstration, was 42,463 bicycles (Coverly, 2010). The demonstration included placement at stops of bicycle racks and lockers, in addition to the bicycle- trailer service. Regular-rider bicycle mode of access to bus stops on the bicycle-trailer routes increased 16-145 Table 16-33 Passenger and Bike Unlinked Trip Boarding Statistics by 2000–2009 Fiscal Year for Valley Metro, Phoenix Fiscal Year Annual Passenger Boardings Annual Bike Boardings Percentage Boarding with Bikes Weekday Bike Boardings Fiscal Year Percentage Boarding with Bikes 2008-2009 65,670,807 a 1,465,980 2.23% 4,657 2007-2008 2.17% 2006-2007 58,019,812 1,175,935 2.03% 3,800 b 2005-2006 1.99% 2004-2005 c 56,358,335 1,168,805 2.07% 3,700 b 2003-2004 c 2.04% 2002-2003 c 50,319,003 1,000,050 1.99% 3,200 b 2001-2002 c 1.94% 2000-2001 c 40,011,099 748,124 1.87% 2,400 b Note: a “Metro Rail” LRT was opened in Phoenix during FY-2008-2009. The FY 2008-2009 Annual Ridership Report does not include bike-on-rail counts, therefore rail boardings have been removed from the Annual Boardings statistic for comparability. b Pre-FY-2008-2009 average weekday bike boardings are estimated by the Handbook authors based on the FY-2008-2009 ratio of average weekday to annual bike boardings. c Data from two small operations components lost. Source: Valley Metro (2010), with elaboration by the Handbook authors.

during the demonstration along with trailer use. The percentage of riders using a bike to access the bus was 1.5 percent in 1978, 12 percent in 1979, and 23 percent in 1980. Walk mode of access decreased, from 80 percent in 1978 to 63 percent in 1979 and 54 percent in 1980. In addition to the shift from walking to cycling access, prime mode shifts occurred, with the majority of diverted users having previously biked all the way. Some 60 percent of transit and/or local bike facility users were students, as compared to 23 percent students in the service area population. At a time when usage of transit for shopping, social, and recreational purposes was increasing, the bicycle- transit service attracted a higher percentage of work and recreational trips than did conventional transit. It helps in understanding the scale of the Santa Barbara operation to compare the reported level of bus bicycle-trailer usage with metropolitan area bikeway usage. By the beginning of the demon- stration, 44 miles of on- and off-road bikeways had been opened, and 600 to 700 bicyclists per weekday used these facilities. That is roughly one bicycle-trailer user at the end of the demonstra- tion per earlier-observed bikeway user (Newman and Bebendorf, 1983). Bicycle-trailer service had to be terminated in 1982, when larger buses obtained in response to ridership increases could not legally use the trailers. In 1984, two-bike rear-mounted racks were made available on five routes. The rear mounting had poor visibility from the bus, which con- tributed to crashes and bike theft. Total annual usage never exceeded 1,200 bikes, and rear-rack use was ended in 1987 (Coverly, 2010). Seattle and other properties that tried rear mounting encountered similar operational problems (Federal Highway Administration, 2010). After nearly a decade, demonstration of two-bike front-mounted racks on three routes was initi- ated in 1996, producing a 4-year average usage of 18,600 bicycles/year. All 14 routes operating full- size buses (and capable of supporting the racks) were then equipped in 2000–2001, and FY 2001–02 saw 52, 736 bicycles carried. This result was in the context of system ridership that had grown sub- stantially in the 1990s. The modest increase in bicycles carried compared to FY 1980–81 reflects effects of two-bike rack capacity limitations compared to the circa 1980s 14-bike trailers (Coverly, 2010). July 2003 bike-on-bus usage was 1.0 percent of overall ridership, which was then 7,070,700 unlinked trips annually, with 60 percent of all buses equipped with racks (Steve Spindler Cartography, 2010). A Florida Bikes-on-Bus (BOB) return on investment analysis, the 3rd entry in Table 16-34, exam- ined 11 systems in Florida and four elsewhere in the United States, most of which had been started in the 1994–98 period and had equipped all buses with racks. BOB performance measures were assembled for three Florida systems as set forth in Table 16-32. No BOB user permits were ever required by 11 of the operators surveyed, including all four agencies outside Florida. At the time of the study, the first two systems listed in Table 16-34 plus Jacksonville required user permits, and Miami-Dade Transit had just eliminated their permit requirement (Hagelin, 2005). 16-146

The two larger of these systems, HART (Tampa region) and PSTA (St. Petersburg-Clearwater region), together with Miami-Dade Transit, provided permit-holder data bases that allowed ran- dom samples for user surveys. The rate of successful telephone calls for the HART and PSTA sur- veys was 11 percent, and the mail-out survey return rate for Miami-Dade Transit was 14 percent, together yielding 220 completed surveys. The survey found the BOB service to have prompted shifts from other modes to transit use, as quantified in Table 16-32, and self-identified increased frequency of use by 72 percent of the 3/4 of BOB users who were prior transit riders. BOB users were found to be highly regular riders, nearly 70 percent having used the service for 1 year or more, and 65 percent using the service 4 or more days a week on average. Racks often or always full were a problem for 20 to 30 percent of users, yet 1/3 of all BOB users were unwilling to park their bike at a bus stop bike rack whether or not they would otherwise have to wait for another bus. Nearly 2/3 of BOB user survey respondents were between 25 and 44 years of age, and over 90 per- cent were male. Work purpose trips accounted for 72 percent of BOB trips taken. Non-whites and Hispanics constituted 52 percent of BOB survey respondents willing to provide their ethnicity, about 4 percentage points higher than for Miami Metrorail bike-to-rail riders. Of respondents will- ing to give their income, 78 percent made less than $30,000 in self-reported income per year (2004 dollars), and there was no evidence of a significant higher-income ridership component as had been found among Miami Metrorail bike-to-rail riders (discussed above under “Non-Motorized Access to Transit”—“Bicycle Access and Egress”). One-half reported no working vehicles in the household and only 18 percent had more than one working vehicle (Hagelin, 2005). The BOB survey obtained self-reported bike access and egress distances. A full tabulation is pro- vided in the “Underlying Traveler Response Factors” section under “Trip Factors”—“Bicycle Trip Distance, Time, and Route Characteristics”—“Bicycle Access to Transit” (see Table 16-66). A sum- mary for Miami-Dade transit, contained within Table 16-35, provides a comparison between bike- on-bus access and egress distances (1st versus 2nd rows of mileage entries). These Miami-Dade survey data, either taken independently or combined with HART and PSTA survey data, depict bike egress distances that are substantially shorter than bike access distances. 16-147 Table 16-34 Bikes-on-Bus (BOB) Annual Statistics for Three Florida Transit Agencies Agency Annual Statistic 2000 2001 2002 2003 Hillsborough Area Regional Transit (HART) BOB boardings 54,000 55,200 57,600 68,400 Unlinked passenger trips 9,219,738 9,761,011 9,390,575 9,185,410 BOB share 0.59% 0.57% 0.61% 0.74% Pinellas Suncoast Transit Authority (PSTA) BOB boardings 45,600 111,480 133,800 152,400 Unlinked passenger trips 9,360,135 9,372,832 10,118,769 9,487,531 BOB share 0.49% 1.19% 1.32% 1.61% Tallahassee Transit (TalTran) BOB boardings 15,708 12,636 11,568 10,860 Unlinked passenger trips 3,922,150 3,934,447 4,140,250 4,372,762 BOB share 0.40% 0.32% 0.28% 0.25% Source: Hagelin (2005).

Table 16-35 also provides a comparison between Miami-Dade bike-on-bus access distances and walk- to-transit access distances (1st versus 3rd rows of mileage entries—rows not shaded). A 4th mileage- entry row pertaining exclusively to walk access to HRT and LRT stations, based on San Francisco Bay Area and Portland, Oregon, data, has been added for comparison with Miami-Dade bike access (to bike-on-bus) and walk access (to both bus and Miami Metrorail, but with bus dominant). These two comparisons illustrate the effective expansion of transit system coverage for those willing and able to use bike access. (Distance differentials are probably exaggerated somewhat by the trip purpose differences documented in the table notes.) This effect of longer bike-on-bus access distances also holds when comparing the bike access distances (1st row of mileage entries) against walk access trips to urban rail (4th row of mileage entries), despite the confirmation that walk access trips to rail tend to be longer than walk access trips to service that is predominantly bus. PSTA ridership responded in an instructive manner when gasoline prices first exceeded $3.00 per gal- lon in 2006. Total FY 2006–07 boardings went up by 260,000, or 2.3 percent, from about 11.3 million in FY 2005–06. Within that number, however, BOB boardings grew by 38 percent to over 300,000 annu- ally, representing 2.6 percent of total boardings. Anecdotally, it was reported that ridership stayed firm, at least in the short term, when gasoline prices subsequently declined (Silva, 2007). It would appear that the effective PSTA tributary area expanded with an assist from the BOB program. 16-148 Table 16-35 Comparisons, by Percentiles, of Florida Bike-on-Bus Access and Egress Distances and Florida and West Coast Walk-to- Transit Access Distances Systems Mode 25th Percentile 50th Percentile 75th Percentile 90th Percentile Miami-Dade a, b Bike access to bike-on-bus 0.50 mile 1 mile 2 miles 3 miles Miami-Dade a, c Bike egress from bike-on-bus 0.25 mile 0.25 mile 0.50 mile 1 mile Miami-Dade d, e Walk access to transit 0-0.06 mi. 0.06-0.11 mi. 0.11- 0.17 mi. 0.17-0.23 mi. West Coast f, g Walk access to urban rail 0.27 mile 0.47 mile 0.68 mile Not reported Notes: These comparisons do not have full trip purpose consistency. The Miami-Dade bike access and egress data are for work purpose trips only. The Miami-Dade walk access data are for all purposes of transit travel. The West Coast walk access to rail data are for all purposes of rail transit travel in the morning peak period — presumably work-purpose dominant. a Self-reported distances in 1/4 mile and 1 mile increments gave “lumpy” results. b Same results were obtained for three Florida systems in total (Hillsboro Area Regional Transit, Miami-Dade Transit, and Pinellas Suncoast Transit Authority combined), except 90th percentile fell in 2 mile increment. c Same results were obtained for three Florida systems in total. d Miami-Dade Transit operates both bus and rail, but bus is dominant (Zhao et al., 2002). e Distances measured along the street network with barriers accounted for (Zhao et al., 2002). f One Bay Area Rapid Transit (BART-HRT) station in El Cerrito, CA, one LRT station in San Jose, CA, and three LRT stations in Portland, OR (Weinstein et al., 2007). g Distances per actual routes traced by survey respondents (Weinstein et al., 2007). Sources: Hagelin (2005) and Zhao et al. (2002), with elaboration by the Handbook authors, and Weinstein et al. (2007).

The 4th and 5th entries in Table 16-32 serve to give some additional breadth to the coverage of bike- on-bus, illustrating the range operations with examples from California’s Coachella Valley (1.7 per- cent bike-on-bus in 2003) to Denver, Colorado. Denver is of particular interest because of their survey distributed to the users of on-bus bicycle racks in the summer of 1999. Riders were asked what they would have done had the bus not been equipped with a rack, with multiple answers allowed. Many respondents indicated alternatives not involving transit, with 37 percent listing bike all the way; 27 percent, drive all the way; and 6 percent, drive to a park-and-ride. Alternatives involving transit use included walking to the bus at 34 percent and locking the bike at the bus stop at 22 percent. Users gave a variety of reasons for using the bus-on-transit combination including “to cover a greater dis- tance” (65 percent), “to have a bicycle at the destination” (64 percent), “quicker than walking” (61 per- cent), “avoid transfers” (28 percent), “foul weather/breakdowns” (28 percent), and “avoid parking hassles” (11 percent). A majority (57 percent) stated they used the service 3–5 days a week (Boyle, 2002). Bicycle on Rail Programs. As with bike-on-bus, bicycle on rail programs expand the number of destinations within quick reach for a public transit user. Perhaps in no environment is this as important as in the service area of North America’s leading bike-on-rail provider, the Caltrain com- muter rail service on the San Francisco Peninsula. Caltrain and its predecessor, Southern Pacific Railroad’s “Peninsula Service,” for a century operated a conventional rural-then-suburban rail passen- ger operation focused predominantly on carrying passengers to and from downtown San Francisco. Indeed, in 1967 it was possible to develop a robust commuter railroad mode share estimating relation- ship for the Peninsula Service based on nothing more than distance of a suburban station from down- town San Francisco and the Census-based proportion of downtown workers in the suburban workforce residing adjacent to the station (Alan M. Voorhees & Associates, Inc., 1968). This conventional-commuter dominance began to change with the evolution of today’s Silicon Valley, with its high-tech industrial and office parks spread throughout much of the old Peninsula Service tributary area. Suburban residents who work for Silicon Valley employers, and San Francisco resi- dents who reverse-commute—all faced with intensely heavy highway traffic—are desirous of using Caltrain despite the dispersion of Silicon Valley worksites and the limitations of suburban bus con- nections. The placement of the San Francisco terminus, 1 mile from the city’s transit spine on Market Street, has also been a factor (Tannen, 2010). This context is essential to understand when deriving lessons from the extraordinary success of the Caltrain bike-on-rail program.34 Caltrain trialed bike-on-rail for 4 months in 1982, but the present operation dates from a larger demonstration in 1992. In 1994 bike stowage capacity was increased by replacing some seats with bike racks (Tannen, 2010). As of 2007 each train had one car or sometimes two cars accommodat- ing 16 or 32 standard bicycles each. In February 2007, from 4:30 AM to about 10:30 AM, 500 north- bound (intra-Silicon-Valley and conventional San Francisco commute) and 424 southbound (reverse commute and intra-Silicon-Valley) passengers with bikes were boarded, representing 7 percent of all AM riders prior to midday (Eisen Letunic and Fehr & Peers, 2008). This was essen- tially the same proportion as 10 years earlier, in 1997, when the bike-on-rail proportion was reported as 7.5 percent (Tannen, 2010). Caltrain AM mode of access as determined in the February 2007 passenger survey was 7 percent bike-on-rail, 1 percent bike parked at the station, 29 percent walk, 19 percent transit, 8 percent free shuttle, 27 percent drove car, and 9 percent drop-off/pick-up. Caltrain bike-on-rail use had been 16-149 34 The Caltrain experience will be of primary relevance to urban rail planners interested in development of reverse and intra-suburbs ridership in areas of significant suburban employment.

at or above capacity for some time, with just under 2,000 daily trips. A special all-day survey in September 2007 counted 51 bicycles “bumped” (turned away), with 55 percent of the bumps occur- ring in the morning at the San Francisco terminal and the 22nd Street station in San Francisco, exclusively affecting reverse riders. These two stations attracted 17 percent of AM system total boardings and 26 percent of AM bike-on-rail boardings. Of 10 major stations accounting for 3/4 of both total boardings and total bike-on-rail boardings, all except the San Francisco terminal han- dled 3 to 9 percent of system bike-on-rail boardings each, with zero to eight bumps at individual suburban stations (Eisen Letunic and Fehr & Peers, 2008). Subsequent Caltrain service changes with introduction of “Baby Bullet” express trains are not reflected in any of these data. Among “Bike + Caltrain” users taking a 2007 online survey, 1/4 said they ceased using the com- bined mode of travel because they got tired of being bumped (Caltrain Bicycle Master Plan Technical Advisory Group, 2007). Nevertheless, rack, locker, and other-facility bicycle parking at stations was underutilized. Except for 100 percent utilization at Redwood City, half-way down the Peninsula, utilization ranged from 25 percent (22nd Street) to 75 percent, including 66 percent at the San Francisco terminal with its high bike-on-rail demand. Overall, station bicycle parking capacity was 55 percent utilized (Eisen Letunic and Fehr & Peers, 2008). Clearly bicycle parking at stations did not meet the needs of most Caltrain bike-on-rail commuters, although locker and bike station/shed fees may be a factor compared to the free bicycle carriage onboard. Although the 2007 online bicycle survey was not statistically controlled, the 13 percent of Bike + Caltrain current users who reported parking at a station, as compared to taking their bike onboard, matched official counts. Bike-on-rail rider responses to “Why do/did you bring your bike on board?” support the importance to them of the service. The top six responses were: “Having my bike with me gives me flexibility” (58 percent), “I need to have my bike with me” (37 percent), “I bike the other way for exercise” (32 percent), “Transit/shuttle connections don’t work for me” (31 percent), “Bike parking options are unsatisfactory” (18 percent), and “Saves money over connect- ing transit or renting a bike locker” (15 percent). Another question of both current and former users found that 80 percent of Bike + Caltrain riders rode Caltrain without their bike on board only 0 to 10 percent the time (Caltrain Bicycle Master Plan Technical Advisory Group, 2007). The final three entries in Table 16-32 present perspectives gained from additional bike-on-rail expe- riences and current operations. The Portland TriMet entry provides an LRT example. The Portland and San Francisco BART entries make clear the preference of a majority of bike-using riders for bike-on-rail over bike-and-park. Although “Portland does not provide much parking at train and bus stops,” with 670 spaces total at transit centers and LRT stops, BART places strong emphasis on bike parking as indicated in Table 16-32. The BART and New York Metropolitan Transportation Authority (MTA) entries span the spectrum from late 20th Century (and newer) Metro-type HRT to subway/elevated HRT systems whose core elements reflect the bicycle-unfriendly station access standards of the early- and mid-20th Century and also tend to serve dense and highly walkable urban development (Pucher and Buehler, 2009b). The BART and New York City systems also span the range of crowding encountered, with BART normally having a low proportion of standees and New York typically experiencing packed rush hour subway trains. As noted in the Portland TriMet entry, bike-on-transit statistics are often presented with LRT and bus information combined. This adds interest to the VTA San Jose data, presented earlier in Table 16-31, separately covering the Santa Clara County, California, bike-on-bus and bike-on-rail programs. As seen there, VTA LRT attracted a bike-on-transit share of boarding passengers about 50 percent higher than did bus, at 3.2 percent versus 2.0 percent circa 2001, although the LRT system’s absolute numbers were 16-150

smaller (Steve Spindler Cartography, 2010). VTA serves essentially the same suburban environment as Caltrain, although more focused on the southern sector of Silicon Valley. Bicycle on Ferry Programs. In addition to bike-on-bus and bike-on-rail programs, there are programs to prioritize the accommodation of bicycles on ferries. (The actual carriage of bicycles on ferries has been commonplace from the widespread introduction of bicycles in the late 19th Century.) Particularly notable is the approach developed by the Washington State Ferries for their Seattle area commute period runs. Loading-area bicycle lanes, auto deck arrangements, and loading and unloading proce- dures afford on-time bicyclists first-loading and first-off priorities in addition to lower fares than autos. The bicycle surcharge on the passenger fare is waived entirely for $20 annual permit holders (Washington State Department of Transportation, 2011). Washington State Ferries were delivering 295 bicycles to the downtown Seattle terminal in the 2007 AM 3-hour peak period. This was 13 percent of the bicycle volume at the 29 count stations along the downtown cordon, and reflected a growth since 1992 of just slightly less than the 106 percent for the cordon as a whole. For context, the four highest AM 3-hour counts out of the 29 cordon sta- tions were the Dexter Avenue bike lanes (connecting with multiple shared use trail approaches to the north) at 318 bicycles, the ferry terminal at 295, and the Elliott Bay Trail from the northwest at 218 and from the south at 220 bicycles (City of Seattle, 2008). Given the limited numbers of urban ferry routes in the United States, the traveler response to specific actions has not been investigated for TCRP Report 95; however the bicycle priority and pedestrian-handling strategies used with fer- ries are important to NMT encouragement where they are relevant. Point-of-Destination Facilities Point-of-destination facilities are provided at a workplace, school, shopping area, or other attrac- tion to make it more feasible or easier to use non-motorized transportation. The obvious example is bicycle parking. Other point-of-destination facilities, conditions, and services examined here are showers and changing facilities, overall destination ambiance, walkable accessibility to multiple stores and services, and bikesharing. The quantitative evidence of effect on active transportation choice is limited and not in consistent measures. Nevertheless, an overall importance of destina- tion facilities for encouraging more utilitarian bicycling—and also walking—comes through. Although bicycle parking and bikesharing are facilities unique to the bicycle mode of active trans- portation, other destination features bear as much or more on the choice to walk. Bicycle Parking and Changing Facilities The few research efforts pertaining to bicycle parking and changing facilities, including showers and lockers, have tended to focus on the commute to work and on bicycling. Although some pedes- trians may avail themselves of changing facilities, especially if they run or jog to work, the primary market for such facilities is bicyclists. One researcher noted, “arriving at work sweaty from exer- tion with nowhere to shower and change discourages all but the most die-hard cyclists” (David Evans and Associates, 1992). A Bicycling Magazine Harris Poll in April 1991 found that “showers and storage” would encourage 17 percent of all adults and about 44 percent of “active riders” to ride a bicycle to work (Goldsmith, 1992). Such surveys identify intent, not actual response, but they are suggestive of the role of point-of-destination facilities. Bicycle parking and changing facilities are considered by most NMT planning practitioners to be basic necessary conditions for bicycle commuting. They may be provided separately or together 16-151

and on-site or nearby. The facilities may be exclusive to the purpose or may have dual use. For example, many developments provide access to changing facilities through an on-site health club. Buildings may have a bicycle room or the developer may set aside space in a parking garage for bicycle parking (C.R.O.W., 1993). A survey of employers participating in Southern California’s trip reduction program found that 45 percent provided bicycle parking and 26 percent provided chang- ing facilities (Litman, 1994). A September 1993 survey of trail use in the Baltimore-Washington area found that among bicycle commuters, 75 percent reported availability of bike racks or showers at their destination (Guttenplan and Patten, 1995). In terms of impact on NMT travel choice, the consensus on the need for good bicycle parking and associated facilities is stronger than the research on the question, perhaps because the need seems so obvious. It is not clear whether bicycle parking tends to be provided in response to increasing demand for it, or comes first and encourages more cycling (Pucher, Dill, and Handy, 2010). Table 16-36 identifies the findings of a few studies that have actually attempted to assess the effects of parking provisions and showers/lockers on bicycling, or bicycling and walking, along with what is available on outcomes of specific parking and shower/change facility actions. Key find- ings are expressed in a number of different metrics, making quantitative synthesis unfeasible. All programs were found, however, to have had a positive impact or to have shown indications of pop- ularity. Together they lend support to the importance of bicycle parking and related amenities. The 1st and 2nd entries in Table 16-36 report research that isolated positive effects on bicycling or cycling and walking levels of bicycle parking or shower provisions. The 2nd of these studies obtained its bicycling and walking measurements in the context of employers subject to a Southern California vehicle trip reduction ordinance (Pucher, Dill, and Handy, 2010, Comsis, 1993). The 3rd Table 16-36 entry focuses on a specific Southern California suburban employer faced with trip reduction ordinance requirements. In this example, the employer chose to combine bicycle storage and changing facility provisions with both a financial incentive and bike maintenance assistance. A 10 percent bicycle commute mode share was achieved, 10 times the regional average (RTC and APBP, 1998). 16-152

16-153 Table 16-36 Summary of Studies on the Travel Effects of Providing Bicycle Parking and Shower/Change Facilities Study (Date) Action Key Findings 1. Noland and Kunruether – 1995 as summarized in Pucher, Dill and Handy (2010) The researchers estimated the effect, on commuter choice to bicycle, of having safe bike parking available at the workplace. A significant rise in perception of cycling convenience and an increased likelihood of cycling to work was identified in connection with safe bicycle parking provisions. 2. Comsis (1993) A 1990s survey of employers parti- cipating in Southern California’s trip reduction program was utilized in California Air Resources Board TDM impact modeling to assess ef - fectiveness of individual strategies. Provision of either bicycle racks or showers in the workplace was found to be associated with discernibly higher levels of bicycling and walking to and from work. 3. RTC and APBP (1998) Fleetwood Enterprises in suburban Riverside, California, installed bike lockers, provided changing facilities, and offered access to tools for cycle repairs. Financial incentives were offered to bicycle commuters in the form of a point/reward program worth about $2.00 per day cycled. This builder of recreational vehicles and manufactured homes, with about 650 employees, ran its bicycle commut- er program between the late 1980s and early 1990s as part of a trip reduction ordinance compliance effort. The 10% bicycle commute mode share achieved was 10 times the regional average. 4. Hunt and Abraham (2007) Utilized a stated preference experi– ment based in Edmonton, Canada, to estimate effects of providing se– cure bicycle parking and showers at the trip destination. (Convenience sample: survey attached/handed to parked/passing bikes/cyclists.) Found statistically significant effects on bicycling, large for parking provisions (equivalent to a reduction of en route cycling time in mixed traffic of 26.5 mi - nutes) and small for showers (equiv. to 3.6 minutes). Parking effect more for younger cyclists, less for older groups. 5. Wardman, Tight, and Page – 2007 as summarized in Pucher, Dill, and Handy (2010) Used the U.K. National Travel Sur- vey and stated preference data, in a multivariate analysis, to examine effects on bike commute shares of different degrees of workplace bike parking and facilities provision. With a base work trip bicycle mode share of 5.8% given no special provi- sions, estimated that bike share would increase to 6.3% with outdoor parking, 6.6% with indoor secure parking, and 7.1% with that plus showers. 6. RTC and APBP (1998) and Herman (1993) (see this section for more information) The City of Chicago embarked in 1992 on a large-scale effort to install bicycle parking throughout the city. The first 1,100 racks were located per suggestions from cycling advocates and city planners. The racks, first located at public places, neighborhood retail, and the CBD, proved popular with cyclists and — after initial objections — businesses started to request them. By the end of 1997, 4,250 racks had been installed. 7. Taylor and Mahmassani (1996) (see “Pedestrian/ Bicycle System Linkages with Transit” for more information) Utilized a stated preference experi- ment with hypothetical work trip transit access scenarios to estimate auto-only, auto-park-and-ride, and bike-and-ride preferences. (Convenience sample, mostly avid cyclists; no reported investigation of bicycle locker space pricing.) Bike lockers identified as a significant incentive to bike-and-ride instead of driving to transit or all the way. Lock - able covered parking 40% as effective. Bike lane more important than lockers for infrequent cyclists; lockers more important for frequent cyclists. Results for showers at work were illogical. 8. Urban Transportation Monitor (2009) Bike stations in 8 U. S. and British Commonwealth cities ranging in size from Boulder, CO, to Chicago, IL, responded to a survey on bike station characteristics. Usage data was given for 7 stations in 6 cities. Consolidated parking was offered in 8 stations (7 cities) ranging in capacity from 40 (Aukland) to 300 (Chicago). Average percent occupancy, where known, ranged from 28% (Seattle) to 88% (San Francisco, Caltrain terminal). (continued on next page)

Canadian and U.K. research on importance of bicycle parking and showers is covered in the 4th and 5th table entries. A stated preference study in Edmonton, Alberta, slanted toward more fre- quent cyclists, found significant but small importance for showers and a large importance for park- ing. Secure parking was, in the research model, equivalent in benefit to avoiding 26.5 minutes of en route mixed-traffic cycling time (Hunt and Abraham, 2007). The research in the United Kingdom estimated that, starting with a 5.8 percent bicycle mode share, outdoor bike parking was associated with 0.5 additional percentage points of bicycle share (9 percent more cycling), secure indoor parking with 0.8 additional percentage points (14 percent more cycling), and secure indoor parking plus showers with 1.3 additional percentage points of bicycle mode share (22 percent more cycling compared to the starting share) (Pucher, Dill, and Handy, 2010). The 6th Table 16-36 entry offers circumstantial evidence of bicycle rack installation effectiveness in Chicago (RTC and APBP, 1998, Herman, 1993). A separate analysis of concurrent Chicago bicycle lane impacts found a bicycling commute mode share increase from 0.28 percent in 1990 to 0.50 percent in 2000 in the bike lane corridors, a 78.6 percent increase in a small starting bicycle share of journeys to work. Three major forces are likely to have affected this gain: implementation of bicycle lanes on at least five radial arterials, extensive promotion of cycling, and the bicycle rack program (Cleaveland and Douma, 2009). The 7th table entry reports on stated preference work that found bike lockers to be over twice as well regarded as ordinary lockable covered parking for transit station use. Frequent cyclists ranked the bike lockers higher in importance than infrequent cyclists, who identified bicycle lanes as being more important to bike-and-ride mode choice selection (Taylor and Mahmassani, 1996). The 8th and 9th entries in Table 16-36 address bike hubs, stations, or the equivalent. The survey form- ing the basis for the first of these two entries employed a voluntary convenience sample of bike hubs/stations to identify bicycle parking occupancy rates. They were found to range, under widely varying circumstances, from 28 to 88 percent occupancy (Urban Transportation Monitor, 2009). It is dif- ficult to conclude much from the bike hub/station usage statistics other than that some locations/ operations are more successful than others, illustrating the importance of location, approach, and exe- cution. The last (9th) table entry provides results from the earlier development of four bike storage and changing/shower facilities in downtown Portland, Oregon, along the lines of what is now termed “bike hubs” or “bike stations.” Users were charged a modest fee. Rates of initiation of bicycle commuting were not given, but for Bike Central users who already commuted from time to time by bicycle, a fivefold 16-154 Table 16-36 (Continued) Study (Date) Action Key Findings 9. RTC and APBP (1998) and City of Portland (2001) In the context of an aggressive pro- gram of providing bicycle parking while gradually restraining CBD auto parking, Portland created “Bike Central” — four locations offering showers, changing facilities, and bicycle storage. Monthly or daily passes allowed use at a modest fee. A before-and-after study found users of the service increased their frequency of commuting by bicycle from 3.1 days/ month before to 15.5 after; driving or taking transit less. First year estimates were 14,600 bicycle trips generated and 46,400 VMT, 23 tons of CO, and 360 pounds of hydrocarbons eliminated. Note: Study methodology, where available, is summarized in the “Actions” column. Analysis limitations were not reported except as provided (in parentheses). Where substantial additional information on individual studies is provided in text and tables or figures, this is noted — and the location within the chapter is given — in the first column. Sources: As indicated in the first column.

increase in days per month of bicycle commuting was seen (RTC and APBP, 1998, City of Portland, 2001). The overall context of auto parking restraint in the downtown is described in the “CBD Parking Supply Management in Portland, Oregon” case study of Chapter 18, “Parking Management and Supply.” Other Destination Amenities There is more at trip destinations than just bicycle parking and showers, or lack thereof, that may be of importance to the choice of travel via walk, bicycle, or transit/walk. Destination amenities include the variety of shopping and services available close by, allowing more daily needs to be met at the work- place or other destination, and friendliness of the destination walk environment. Both land use variety and walk environment friendliness are factors that spread over into other topics, most particularly “Pedestrian/Bicycle Friendly Neighborhoods” in the subsection to follow. Nevertheless, there are some findings of traveler response studies that may be productively considered by taking a trip destination perspective. A sampling of these “destination amenities” findings is presented here. A study of workplace destination amenity effects on combined walk and bike work trip mode shares was carried out in the early 1990s using a sample of 330 employment sites in Los Angeles County subject to vehicle trip reduction requirements. Observed NMT mode share averages for the various worksite-environment groupings examined were within or close to the range of 2 to 4 percent. All five amenity measures examined were associated, when indicative of conditions postulated to be favor- able to walk and bike commutes, with higher NMT mode shares. For example, high walking accessibility to convenience services around the workplace was asso- ciated with NMT work trip shares that were higher by 0.7 percentage points (employers without TDM financial incentives) and 1.1 percentage points (employers with TDM financial incentives) compared to sites with low workplace walking accessibility. High appearance of safety around the workplace was associated with NMT shares higher by 1.5 percentage points (without or with incentives) compared to sites with a low appearance of safety. High workplace and vicinity aes- thetic appeal was associated with NMT work trip shares that were higher by 0.7 percentage points (without TDM financial incentives) to 1.3 percentage points (with incentives) compared to sites with low aesthetic appeal. (These examples are derived from Table 19-28 in the “Land Use and Site Design” subsection within the “Underlying Traveler Response Factors” section of Chapter 19, “Employer and Institutional TDM Strategies.” A fuller description of the study may be found there.) Oddly, the only characteristic that achieved statistical significance in the final explanatory model was aesthetic appeal, leading the researchers to speculate that it was acting as a surrogate for other unidentified factors (Cambridge Systematics with Deakin, Harvey, Skabardonis, 1994). City and county of San Francisco travel demand modeling results covered in Chapter 15, “Land Use and Site Design,” extend the Los Angeles findings into a fuller array of trip purposes and destination types. Besides topography, which in San Francisco tends toward the dramatic, the most influential des- tination pedestrian environment characteristic among those scaled for model use by a Delphi panel proved to be urban vitality. The urban vitality characteristic, both in the case of work and in the case of other trip purposes, was an indicator of higher mode shares for walking, walk-transit, and also (“other” trip purposes only) bicycling. Only destination (non-home) pedestrian environment factors proved useful in estimating choice of mode in the San Francisco context (Cambridge Systematics et al., 2002). (In Chapter 15, see “Response by Type of Strategy”—“Site Design”—“Transit Supportive Design and Travel Behavior”—“Pedestrian/Transit-Friendliness,” including Table 15-44, for further detail.) Finally, it is appropriate that attention be drawn back to the Austin, Texas, research reported on in this “Response by Type of NMT Strategy” section under “Sidewalks and Along-Street Walking”— 16-155

“Sidewalk Coverage and Traffic Conditions” (see 1st entry in Table 16-2 and related discussion). The Austin findings strongly suggest that completeness and quality of commercial area sidewalks and sidewalk connections to stores is of major importance for attracting more persons to the walk mode for shopping trips (Cao, Handy, and Mokhtarian, 2006). Indeed, these factors appear—under light residential traffic conditions—to likely be more critical for utilitarian walking than complete- ness of the residential area sidewalk system. Bikesharing Bikesharing, involving the shared use of a publicly available bicycle fleet, may be considered both an origin and a destination facility and service. A subscriber obtains a standardized bicycle at a bikeshare docking “station” close to his or her trip origin or mode change point and after using it turns it in, taking care of parking in the process, at a bikesharing station close to the destination. The intended use is short-term, and pricing is often set to strongly discourage longer-term use. Bikesharing Development. Bikesharing is in the developmental stage of being scaled up into major programs from the predecessor small-scale applications that had been gradually evolving since 1965. Evolution has taken place in three stages. In the first generation distinctly-painted bikes were made available in selected local areas, typically tourist areas and downtowns, for free use. Some attempts failed because of theft, but others survived and evolved. The second generation involved a shift to coin-deposit systems requiring deposits averaging roughly US $4.00 to unlock a bike, with the deposit being remitted with return of the bicycle. Theft remained a major problem. The earlier programs tended to be too small or unreliable for citywide impact. The third and present generation, greatly facilitated by the arrival of information technology (IT), includes the bicycles (still distinctive), docking stations, a user interface for check-in and check-out (kiosk or technology), and IT with GPS to facilitate reservations, pick-up and drop-off management, and location tracking. Modest single-use, day/week-pass, and/or annual subscriber fees are typi- cally assessed. Some systems offer a brief initial or quick-trip period of use at no rental cost. The Paris Vélib day pass in 2009 was just under US $1.50, and the Washington, DC SmartBike and Paris annual passes were US $40.00 or thereabouts (Shaheen, Guzman, and Zhang, 2010). Theft and vandalism have been reported as growing problems in Paris. Some 80 percent of the initial fleet was stolen or damaged in the initial 1 to 2 years, and satisfaction with condition of the bikes declined from 55 per- cent in 2008 to 46 percent in 2009 (Erlanger and De La Baume, 2009). On the other hand, only one or two bikes out of 700 were lost in the initial year of Nice Ride Minnesota’s operation in Minneapolis, and only three had vandalism damage costs of over $100 (DeMaio and Meddin, 2010, Dossett, 2011). Nice Ride Minnesota requires a $50.00 deposit when taking out a one-day subscription to the service. Subscription fees themselves are $5.00 for 24 hours, $30.00 for 30 days, and $60.00 for 1 year ($50.00 for students). There is no trip fee for using a bike for 30 minutes or less. Beyond that, trip fees are $1.50 for up to 60 minutes, $4.50 for up to 90 minutes, and $6.00 for each additional 30 minutes, up to $65.00 maximum per day. The fee system is clearly set to encourage short-term use, and the Nice Ride web- site links prospective users to local bike rental services for longer-term uses (Nice Ride MN, 2011). When tabulated circa 2009, a number of area-wide systems were already in operation, as shown in Table 16-37. They were predominantly in Europe, although the most bikes deployed were in China. Not included in this table, or this discussion, are “closed” systems on university and workplace campuses and the like. As of 2009 there were over 65 such closed systems in the United States alone, and 10 more were expected in 2010. Also not included are more traditional bicycle rentals such as are typically arranged by bicycle shops (Shaheen, Guzman, and Zhang, 2010). 16-156

The market penetration of public bikesharing systems is rapidly evolving, making tabulations such as that presented in Table 16-37 only a snapshot in time. Technology adoption lifecycle research by Everett Rogers and colleagues in the last half of the 20th Century identified a course of innova- tion adoption that follows a “bell curve” or normal distribution. Innovators and early adopters “buy in” on the upswing, which in isolation, takes on the appearance of an exponential curve. Early-majority adopters take the curve up to its apogee, with late-majority and laggard adopters on the mirror-image downswing (Wikipedia, 2011). A rough tally of bike-sharing services worldwide (quite likely not using the same criteria as Table 16-37) identified 11 in 2004, 60 in 2007, 92 in 2008, 160 in 2009, and 238 in 2010 (DeMaio and Meddin, 2010). These data points approximate an exponential curve, particularly if one accounts for fleet growth of services already established, placing the third generation of bikesharing in the innovator or early adopter phases of technology adoption. Compared to the one U.S. system of Table 16-37 in 16-157 Table 16-37 Public Bikesharing Programs Listed by Country in Decreasing Order of Numbers of Bicycles Deployed, Circa 2009 Country Number of Programs a Number of Bicycles Number of Stations China 3 65,000 2,522 France 21 34,898 2,797 Spain 21 11,080 842 Germany 3 6,069 128 b Canada 1 5,000 400 Italy 16 3,392 361 Denmark 3 2,513 277 Sweden 3 2,125 171 Taiwan 2 2,000 31 Norway 1 1,660 154 Austria 3 1,500 82 United Kingdom 2 1,410 809 Belgium 1 1,000 100 Ireland 1 450 40 South Korea 1 430 20 Luxembourg 2 370 40 Finland 1 300 26 Brazil 2 232 26 New Zealand 1 175 11 Switzerland 1 120 11 United States 1 120 10 7 Other Countries c 9 311 d 263 Total 99 140,155 d 9,121 b Notes: a Within each country, each system is counted as one program even though it may serve multiple cities. “Closed” systems, such as a number of campuses have, are not included. b Number of stations does not include “flex station” drop-off points used in 5 German cities. c Countries reporting 100 or fewer bicycles or for which bicycle totals could not be confirmed. d Does not include the bicycle totals for Mexico (1 program, 12 stations) or The Netherlands (1 program, 200 stations) because the researchers could not confirm bicycle numbers. Source: Shaheen, Guzman, and Zhang (2010).

2009, three large-scale U.S. services were implemented in 2010 (DeMaio and Meddin, 2010), and at least 10 U.S. metropolitan areas had major systems in the process of implementation or actually oper- ating in mid-2011 (DeMaio, Simmons, and Meddin, 2011). Bikesharing Operational and Impact Statistics. Given the relative newness of third generation bikesharing programs, statistics—especially relevant travel demand statistics—are sparse. Among available operating statistics, Table 16-37 indicates an average per-docking-station deployment of 15 bicycles. The range of per-country-averages is 9 to 26 bicycles per station, excluding one outlier (Taiwan). With 20,600 bicycles and 78,000 trips on an average day, Vélib in Paris achieved a daily turnover rate of 3.8 uses per bike. Hangzhou, China, reported a turnover rate of 6 daily uses per bike for their 40,000-bicycle system (circa 2009), the world’s largest (Shaheen, Guzman, and Zhang, 2010). There are also reports of Velo’v bicycles in Lyon, France, being used 6.5 times per day (DeMaio, 2009) and a small Dublin, Ireland, Dublinbikes service reaching 10 trips per day (DeMaio and Meddin, 2010). Other statistics offered in parallel are actually subscribers per bike; not the same thing as actual turnover. Viewed in the context of typical U.S. short-term parking turnover rates, the Paris bicycle turnover rate of 3.8 seems a more conservative representation of likely citywide bikesharing usage patterns in western countries. Indeed, Capital Bikeshare January through July 2011 statistics for Washington, DC, and Arlington County, Virginia, indicate a 3.3 rentals-per-bike winter-into-summer average. Turnover increased from 1.3 in January to just under 5.0 in June, with nearly identical turnover in July, at which time there was a 3 percent bike availability constriction relative to June. June 2011 saw 140,400 rentals, an average of 4,680 per day, with 943 bicycles in service. January 2011 was only the fifth month of operation (Capital Bikeshare Dashboard, 2011). The relative effects of season, numbers of tourists, and system newness are unknown. First-year (2010–2011) Capital Bikeshare statistics include an average trip length of “around 1.79 miles per trip” (David C., 2011). Internationally published mode shift data, while not documented or defined in detail, seems to indicate notable increases in cycling in response to the major citywide programs. Bicycling in Lyon, France, is reported to have increased 44 percent in the first year of their Velo’v program. All but 4 percent were “new users who had not previously bicycled in the Lyon city center.” Reports of bicycle riding increases in Paris in response to deployment of their Vélib system range from 70 per- cent (Shaheen, Guzman, and Zhang, 2010) to 250 percent, the latter measured as a 1.5 percentage point shift starting from about a 1 percent share in 2001 and increasing to 2.5 percent in 2007, the year of system implementation.35 The bicycle share in Barcelona, Spain, increased from 0.75 per- cent in 2005 to 1.76 percent in 2007, again the year of implementation, representing a 235 percent increase and a 1.0 percentage point shift (DeMaio, 2009). The starting shares were low in Paris and Barcelona, and in both cities bicycle facility improvements were made during the years in question. In Lyon, user survey respondents reported replacing about 1/2 of their transit trips with Velo’v bicycle trips. The trade-off is the advantage for accessing transit, possibly compensating to some degree for loss in transit riding. In Paris, 21 percent of 2008 survey respondents reported use of Vélib for transit access and 25 percent reported Vélib use for transit egress, with 28 percent over- all reporting reduced use of their private vehicle during the day. The 2009 survey found 28 per- 16-158 35 Some of these bicycle mode share increases possibly pertain to the city center rather than the city as a whole and/or seem likely to incorporate overall trends.

cent of respondents indicating use of Vélib “to begin and to end their multi-leg transit trip,” with 48 percent overall reporting reduced private vehicle use (DeMaio, 2009). In 2010, the first season of the Nice Ride Minnesota operation (it shuts down for the winter), most subscribers were for 24-hours only. This pattern is shifting. The 1,300 annual-subscriber base of December 2010 grew to 3,200 in June 2011, with the system poised for expansion from Minneapolis into St. Paul (DeMaio and Meddin, 2010, Dossett, 2011). A 2010 Nice Ride subscriber survey obtained 685 responses, a return of 53 percent compared to the December subscriber base. Respondents came from a fairly broad range of ages, but with only eight younger than age 18 or older than 64, and with the most in the 25 to 34 age range (39 percent). Aside from only 6 percent reporting household annual incomes below $20,000, the income spread was remarkably even. On the other hand, only 15 percent reported less education than a full four-year college degree. Students accounted for 19 percent of subscribers. The gender split was typical of U.S. bicycling, 63 percent male after correction for non-responses to the question. Before Nice Ride, 1/3 rode a bike less than once a month, while almost 1/2 rode at least once a week (Nice Ride MN, 2010). These and other responses, given the newness of the system, may be judged representative of inno- vative and early-adopter subscribers more than would be seen with a more mature operation. The survey results provide further insight into usage patterns and the relationships between bikeshar- ing and transit use. However, all but one of the relevant survey questions were couched in terms of pri- mary use rather than a specific trip. Primary use by purpose category was 37 percent commuting to work or school, 30 percent errands and meetings—transportation around downtown, 5 percent each for social riding and exercise, 3 percent shopping, 13 percent going to events and eating/drinking estab- lishments, and 6 percent other. Self-reported changes in use of transit (bus, LRT, and commuter rail) identified 10 percent who increased use, 17 percent who decreased use, 58 percent whose use stayed the same, and 14 percent who were not transit users. Some 32 percent primarily used Nice Ride to con- nect with bus or rail transit. The trip-specific question elicited responses that if Nice Ride Minnesota had not been available for the respondent’s most recent trip, it would have been made by walking (38 percent), personal bicycle (8 percent), public transit (20 percent), driving (19 percent), getting a ride (1 percent), taxi (3 percent), other (2 percent), or would not have been made at all (9 percent) (Nice Ride MN, 2010). Pedestrian/Bicycle Friendly Neighborhoods Some of the more notable travel impacts of urban land use structure and design relate to their effect on “active transportation”—the decision to walk, to cycle, or to do one or both in conjunction with taking public transportation. The travel demand, land use, and urban design interrelationships involved have attracted a significant proportion of research on effects of the physical environment on walking and cycling. This circumstance leads to a modified approach for presentation of avail- able findings. The presentation approach is also shaped by the fact that this “Pedestrian/Bicycle Friendly Neighborhoods” subsection supplements the full presentation and interpretation of the travel behavior effects of urban form in Chapter 15, “Land Use and Site Design.” It is also aug- mented by the “Transit Oriented Development” discussion within the “Pedestrian/Bicycle Linkages with Transit” subsection above, which is drawn from Chapter 17, “Transit Oriented Development.” Both Chapters 15 and 17 serve as primary cross-references and should be referred back to for additional context and insights, particularly on the broader topic of urban form’s influ- ence on all modes of travel as well as the underlying role of socio-demographic factors. Accordingly, the following discussions are oriented to findings of five major NMT syntheses (including a meta-analysis) that are newly available since Chapter 15 was prepared, taken together 16-159

with an NMT-specific extraction from the Chapter 15 material itself. The five syntheses primarily address land use, transportation, and public health. Information from these sources is leavened with additional or confirmatory insights from selected individual studies, but without any attempt at completeness in terms of individual study inclusion. The numerous findings in Chapter 15 con- cerning land use effects on transit use per se are not repeated here, but elasticities for transit use are reported alongside walk elasticities in the “Walk Elasticities for Land Use and Site Design Parameters” discussion at the end of this subsection. As emphasized elsewhere, significant walk- ing is involved in most transit trips. The presentation uses four summary tabulations of findings, Tables 16-38, 16-39, 16-40, and 16-41. Table 16-38 draws from the five NMT syntheses. Table 16-39 extracts from Chapter 15 material specifically pertaining to walking and bicycling, Table 16-40 is a selection of individual studies with information about impacts on primarily adult active transportation, and Table 16-41 is a selec- tion focused on child walking and bicycling activity. In addition, meta-analysis built-environment elasticity results are provided in Table 16-42. Some summaries in these tables are allowed to speak for themselves without much if any further elaboration. Studies addressing choice of mode for access to transit stops and stations are, however, all discussed further in the text. The same is done for studies focusing on bicycling relationships with land use and site design. This approach is to compensate for the general lack of coverage of transit access mode choice effects and bicycling-specific effects in the five major NMT syntheses. 16-160

16-161 Table 16-38 Relationships between Pedestrian- and Bicycle-Friendly Neighborhood (NBH) Characteristics and Walking/ Bicycling, Summarized from Key Syntheses Study (Date) Process Key Findings 1. Ewing and Cervero (2010) This research used meta-analysis of over 50 quantitative studies to derive and interpret new elasticities for an array of land use and site design parameters. Many additional studies were used in the overall synthesis. VMT, walking, and transit use elasticities were derived. Built environment descriptors found most closely related to walking were intersection/street density, diversity (all measures), and local access to jobs. Land use mix, the neighborhood design measures, and distance to a stop were important to transit use. All elasticities were in the lower inelastic range. 2. Saelens and Handy (2008) This synthesis was designed as a broad update to both SR 282 (see below) and the child-focused work by Davison and Lawson (2006) (see below). It covers 9 reviews published 2002-2006 and also indi- vidual adult- and child-focused papers published 2005-May 2006, almost all on cross-sectional studies. Found greater density, higher mix of land use, aesthetics, street connectivity, enhanced accessibility or proximity, traditional NBH design, and related infrastructure and conditions such as sidewalks and safety to be positively correlated with walking or walk+bike activity. For children the list was short- er, with distance to school critical. 3. Committee on Physical Activity, Health, Transpor- tation, and Land Use (2005), Handy (2004) TRB Special Report 282 (SR 282) ex- amined the connection between U.S. physical activity levels and the built environment, employing synthesis of both transportation and physical activity research results, along with 7 specially commissioned papers including Handy (2004). Similar to other contemporary reviews, found the basic density, diversity, and proximity measures positively related to NMT travel, as were traditional, transit-served, and walkable NBHs. Relationships with total physical activity were mostly limited to NBH pleasant-environment measures. 4. Davison and Lawson (2006) Researchers at the University at Albany – SUNY, in New York State, prepared a synthesis on the same questions as SR 282 but focused on school access commute modes and physical activity of children. Of 3 studies of school access distance, all found use of active modes for access to be correlated positively with shorter distances. Less consistent research, on balance, found street connectivity and good access to destinations positively related with student physical activity. 5. Moudon, Stewart, and Lin (2010) This Washington State Department of Transportation (WSDOT) assess- ment undertook a comprehensive international literature review with interpretation of Safe Routes to School (SRTS) programs and outcomes and related information. All 19 studies that examined distance to school found it inversely related to use of active transportation to school (ATS). Increased distances from 1969 to 2001 explained 47% of the drop in ATS. NBH characteristics were mostly logically or insignificantly related. Note: Findings of each of these syntheses are discussed further within this “Pedestrian/Bicycle Friendly Neighborhoods” subsection. Sources : As indicated in the first column. The organization of this subsection revolves around the various “D’s” describing the built envi- ronment. First come the “original” 3D’s of density, diversity, and design. Those are followed by effects of destination accessibility and distance to transit, and then all built environment factors operating in conjunction. Two other “D’s” sometimes included in land use and transportation dis- cussions are demographics and demand management (Ewing and Cervero, 2010). Demographics receive limited attention in the “Response by Type of . . .” subsections of this “Traveler Response

to Transportation System Changes” Handbook because they are treated as a given. Socio-economic effects on walking and bicycling are primarily examined in the “Underlying Traveler Response Factors” section under “User Factors.” The positive effect on bicycling of one example of work- place travel demand management (TDM) was covered in the “Point-of-Destination Facilities” sub- section under “Bicycle Parking and Changing Facilities.” Density Density is a measure of concentration of population, dwelling units, employment, or other vari- ables of interest per unit area. Historically, this measure has often been used in aggregate, simple relationships that cause higher densities to act as a stand-in for many other closely linked charac- teristics—such as closeness to the CBD, better transit service levels, lower auto ownership, and higher parking costs—and to thus exhibit very strong associations with NMT and transit use and lower VMT (e.g., Dunphy and Fisher, 1996, Table 16-39, 1st entry). Newer research typically exam- ines density independent of such influences. Findings concerning magnitudes of density effects thus cover a broad spectrum. In interpreting research that addresses density in isolation, it is important to take into account that certain associated factors are directly affected by density. Population and employment densities are intermediate variables often expressed through other variables with effects stronger than den- sity itself. Included among other variables positively affected by higher densities are not only tran- sit service intensity (more people available to support good service) and auto ownership (lower auto ownership where auto availability need is less and parking is more difficult and costly) but also NMT accessibility (more activities within a given walking or cycling distance) (Ewing and Cervero, 2010, Schneider, 2010). On the other hand, other influences that often historically accom- pany density such as orientation to the CBD, greater land use mix, grid street patterns, and lower incomes are not caused by density. They should properly be considered exogenous influences whose effects ought not to be attributed to density in and of itself. A weak association was found between density and vehicle miles of travel (VMT), walking, and transit use in meta-analysis derivation (Table 16-38, 1st entry) of built environment elasticities. (See also Table 16-42 under “Walk Elasticities for Land Use and Site Design Parameters.”) This may be taken to infer that when other factors are controlled for the direct effect of more residents or jobs per unit area only slightly increases walk and transit use activity or mode shares (Ewing and Cervero, 2010). As already noted, however, there are other positive influences on active transportation that draw support from higher densities. To elaborate on the example of transit service, where higher densi- ties move the number of transit riders upward beyond basic-service-level bus transit capacity thresholds or rail transit investment thresholds, more intensive transit service is the result, provid- ing service frequency and speed benefits. These benefits attract more transit riders on their own, in line with service and ridership relationships not much built into independently derived density elasticities such as those in Table 16-42.36 However, an important “take-home” lesson from the low 16-162 36 There are several parts of Chapter 15, “Land Use and Site Design,” that address these important relation- ships, including “Density”—“Density Related to Transit Use”—“Density and Transit Choice” within the “Response by Type of Strategy” section, “Transportation Service Levels” in the “Underlying Traveler Response Factors” section, and “Transit Service Feasibility Guidelines”—“Density Thresholds for Transit Service” under “Related Information and Impacts.”

elasticities for density per se is that density not well integrated into the urban fabric—such as apart- ments in the middle of auto-oriented suburban sprawl—will not offer a full measure of beneficial effects on VMT, transit use, or walking and bicycling for transportation. Three of nine reviews published between 2002 and 2006 (Table 16-38, 2nd entry) identified density as being linked to added walking. The researchers conducting the overall synthesis suggest that this out- come reflects the fact that higher density tends to make destinations more proximate. Indeed, five of the nine reviews—representing “the most consistent set of conclusions”—found accessibility based on closeness to destinations to be associated with additional walking. Individual studies reviewed, mostly newer, supported these conclusions but only for utilitarian walking. Little or no evidence was found for relationships between recreational walking and density or non-residential destina- tion proximity (Saelens and Handy, 2008). The term “recreational walking” generally includes walking for exercise. TRB Special Report 282: Does the Built Environment Influence Physical Activity? Examining the Evidence (SR 282) (Table 16-38, 3rd entry), in general anticipates the results of the Saelens and Handy synthesis, par- ticularly as they pertain to travel undifferentiated by utilitarian versus recreational purposes. Unlike the Saelens and Handy 2002–2006 synthesis and the Ewing and Cervero meta-analysis, SR 282 includes consideration of bicycling. Bicycling is included, however, only in combination with walking and not independently. SR 282 specifically offers the observation that in those studies which examined both density and accessibility, only accessibility was found to be significant as a predictor of walking and bicycling. A likely explanation offered is that “density may serve as a proxy for accessibility, which provides a more direct explanation for travel behavior” (Committee on Physical Activity, Health, Transportation, and Land Use, 2005, Handy, 2004). Along the same vein, a review of infrastructure, programs, and policies to increase bicycling notes that one of many probable reasons for higher cycling rates in northern Europe is the general restriction of low-density, auto-oriented land uses. The resulting compact, mixed-use development supports shorter trip distances more readily covered by bicycle (Pucher, Dill, and Handy, 2010). Individual research efforts covered in Chapter 15, extracted in Table 16-39, and the additional indi- vidual studies assembled in Table 16-40, add more texture to the synthesis study findings. However, the individual studies that address density also address diversity, design, and/or acces- sibility, typically with more notable results. Thus Table 16-40, covering additional individual stud- ies, is introduced later on. 16-163

16-164 Table 16-39 Summary of Primary Comparative Observations from Chapter 15 on Impacts of Density, Diversity (Mix), and Design on Walking/Bicycling Study (Date) Key Observations 1. Density: Dunphy and Fisher (1996) (within Chapter 15, “Land Use and Site Design,” under “Response by Type of Strategy,” see “Density” — “Density as Prime Indicator at the Behavioral Level” — “Density Inclusive of Related Phenomena”) Active transportation becomes more significant at higher densities. Nationwide, at population densities of 2,000 to 5,000 persons/sq. mi., 7% of daily trips are made by walking or biking, versus 28% at 10,000 to 49,000 persons/sq. mi., and 46% at over 50,000 persons/sq. mi. These percentages are in response not only to density but also to all the urban characteristics that usually accompany it, including greater land use mixing, shorter distances between attractions, and better pedestrian accommodations. 2. Density: Kockelman (1996) (in Chapter 15 see “Diversity (Land Use Mix)” — “Accessibility, Entropy and Other Measures” — “Accessibility and Land Use Mix”) Evaluation of NMT choice relative to density in its purest form found no significant direct den- sity effect on the basis of detailed San Francisco area data. There was a small positive effect channeled through reduced auto ownership. 3. Density: Frank (1994) (in Chapter 15 see “Density” — “Density related to Transit Use” — “Density and Transit Choice”) Seattle area research estimated that higher popu- lation density of 10 persons more per acre at origin and destination was associated with ~8% higher NMT share, with 10 more employees per acre adding another 1% to 2% of NMT share. 4. Density: Parsons Brinckerhoff et al. (1996b) (in Chapter 15 see “Density” — “Density related to Transit Use” — “Density and Means of Transit Access”) For access to rail transit service, walking nor- mally predominates for only up to 1/2 to 3/4 of a mile, though a peripherally located downtown commuter rail terminal can push the envelope up to 1-1/2 miles. Population density higher by 1% is associated with 1 to 2 percentage points higher choice of walking to rail transit in Chicago and to the Bay Area Rapid Transit (BART) system in the San Francisco area. Associated auto use is lower by about 2 percentage points for more-suburban systems, and bus use for rail access drops by about 1 percentage point on the urban rail systems. 5. Diversity: Kockelman (1996) (in Chapter 15 see “Diversity (Land Use Mix)” — “Accessibility, Entropy and Other Measures” — “Accessibility and Land Use Mix”) The research on San Francisco area data that did not turn up a direct density effect on NMT choice estimated walk/bike elasticities of +0.23 and +0.22 relative to land use balance and accessibility, respectively. 6. Diversity: Frank (1994) (in Chapter 15 see “Density” — “Density related to Transit Use” — “Density and Transit Choice”) In the Seattle area evaluations, choice of walking for the work trip was the only instance where land use mix proved to be statistically significant as a research model variable. 7. Diversity: Steiner (1998) (in Chapter 15 see “Diversity (Land Use Mix)” — “Land Use Mix and Transit Use” — “Mix and Pedestrian Access”) Some 20% to 38% of weekday shoppers at highly walk-accessible San Francisco East Bay shopping centers were observed walking to shop (more on Saturday), though the result at the more popular centers was not less parking demand, but rather more shopping activity. 8. Diversity: Parsons Brinckerhoff et al. (1996a) (in Chapter 15 see “Diversity (Land Use Mix)” — “Land Use Mix and Transit Use” — “Mix and Mode Choice”) An 11-city study found proximity of retail to housing most important for NMT choice, with — depending on density — a 15 to 17 percentage point gain in walking and cycling for trips 1 mile long.

16-165 Table 16-39 (Continued) Study (Date) Key Observations 9. Diversity: Parsons Brinckerhoff et al. (1996b) (in Chapter 15 see “Diversity (Land Use Mix)” — “Land Use Mix and Transit Use” — “Mix and Means of Transit Access”) Estimated elasticities for mode of access/egress to Bay Area Rapid Transit (BART) stations, quantifying response to an index of mix, were +1.1 for walk and -1.3 for auto, both reflecting elastic (very sensitive) travel demand behavior. 10. Diversity: Cervero (1988) (in Chapter 15 see “Site Design” — “Suburban Centers” — “Suburban Employment Centers”) Study of suburban employment centers (SECs) identified mix of uses within SECs as having a small but positive effect on the incidence of walking trips. Houston’s SECs had 20% of all trips being made by walking despite long blocks, limited crossings, and disconnected sidewalks. The non-work walk share was 22%. Of all walk trips 1/3 were between 11 AM and 2 PM. 11. Diversity: Rutherford, McCormack, and Wilkinson (1997) (in Chapter 15 see “Site Design” — “Community Design and Travel Behavior” — “Paired TND and CSD Communities”) Higher walk mode shares were found in mixed use locales with gridded streets in Seattle (18% walk versus 9% for the whole of North Seattle), and in the Seattle suburbs (8% for a town with mixed land use and partially gridded streets versus 3% for the inner ring overall). 12. Design: Parsons Brinckerhoff (1996) (in Chapter 15 see “Site Design” — “Suburban Centers” — “Worksites with Travel Demand Management”) Worksites with an “aesthetic” setting obtained employee commute NMT shares 25% higher than other worksites, except this relationship held only in the presence of TDM programs with financial incentives. 13. Design: McNally and Kulkarni (1997) (in Chapter 15 see “Site Design” — “Traditional Neighborhoods versus Hierarchical Planned Unit Developments” — “Community Design and Travel Behavior”) Southern California comparisons found pedestrian shares in traditional neighborhood design (TND) communities ranging from 17% less to 53% more than in conventional planned unit developments (PUDs). 14. Design: Cervero and Radisch (1995) (in Chapter 15 see “Site Design” — “Community Design and Travel Behavior” — “Mixed Use Communities versus Surrounding Areas” and “Paired TND and CSD Communities” for more information, and see also “Case Studies” — “San Francisco East Bay Area Pedestrian versus Auto Oriented Neighborhoods” in Chapter 15 for an expanded description with additional travel data) A paired community analysis in the San Fran- cisco East Bay showed the TND neighborhood with fine-grained land use mix and integrated sidewalks and paths to engender a 31% walk share for rapid transit station access, compared to 13% for the community with a conventional suburban design (CSD) environment and a coarser land use mix, mostly stand-alone auto- oriented retail, large blocks, and a substantial commuter parking lot. (The rail transit stations are centrally located in both communities and had 21% and 20% rail mode shares, respectively, for work trips.) Also found was a large differ- ence in walk/bike choice for non-work travel: 10% NMT share in the TND neighborhood versus 2% in the CSD area. The corresponding walk-only shares for work purpose trips were 7% (TND) versus 1% (CSD). 15. Design: Kitamura, Mokhtarian, and Laidet (1994) (in Chapter 15 see “Site Design” — “Community Design and Travel Behavior” — “Traditional Urban Neighborhoods versus Newer Suburbs”) A 5-neighborhood San Francisco Bay Area study concluded that attitudes are more important in NMT choice (though not exclusively so) than either household or urban form characteristics. Urban location and presence of sidewalks were isolated as significant built environment factors. (continued on next page)

16-166 Table 16-39 (Continued) Study (Date) Key Observations 16. Design: Cambridge Systematics, Putman Associates, and Calthorpe Associates (1992), Cambridge Systematics et al. (2002) ( see “Site Design” — “Transit Supportive Design and Travel Behavior” — “Pedestrian/ Transit-Friendliness,” still within Chapter 15) Good pedestrian environment was found to be positively related to higher NMT shares in Portland, OR, and City/County of San Francisco travel demand modeling applied research. In San Francisco, where gridded streets with side- walks predominate, urban vitality and amenable topography were the strongest indicators. Note: The location in TCRP Report 95, Chapter 15, “Land Use and Site Design,” where the full discussion is provided (in all cases within the “Response by Type of Strategy” section) is noted in the first column. Sources: As indicated in the first column. The 1st entry in Table 16-39 sets the stage by illustrating how much NMT activity varies in accor- dance with residential density when density is allowed to act as a surrogate for all the urban and socio-demographic characteristics that historically accompany it. The 2nd and 3rd entries are illus- trative of the range of results obtained when density effects have been estimated on the basis of research constructed using individual-study disaggregate data sets and modeling. The 4th entry in Table 16-39 is one of a handful of studies that address the effect of the built environ- ment on the choice of whether to access transit service via walking or bicycling or some motorized mode. Differentiation between access mode choice, involved in the NMT versus motorized-mode choice of how to get to and from transit stops and stations, and prime mode choice, such as the choice to walk or bicycle all the way instead of using transit or driving, was explained earlier within the lead- in to the “Pedestrian/Bicycle Linkages with Transit” subsection. The mode of access analysis identified a substantial positive effect of density on the overall choice of whether to access transit service via walking or some motorized mode, a finding bolstered by TOD research reported in Chapter 17, “Transit Oriented Development.” Choice of mode for access- ing transit service presents a situation akin to choice of mode for short distance local area travel, and is very sensitive to urban form. Higher densities place more riders within the walking radius. For rail transit service, walking predominates for up to 1/2 to 3/4 of a mile, but no farther under normal circumstances. Population density higher by 1 percent was found to be linked with 1 to 2 percentage points higher choice of walking to rail transit in Chicago and to the Bay Area Rapid Transit (BART) system in the San Francisco area. The associated auto use was lower by about 2 per- centage points for the more suburban systems, while bus use for rail access dropped by about 1 percentage point on the rail rapid transit systems, with more people walking. Higher residential area employment density—actually a measure of land use mix—was shown to also enhance walk- ing to the urban systems, by the same order of magnitude (Parsons Brinckerhoff et al., 1996b). Diversity Diversity, or land use mix, is a measure of the variety of land uses in a specified area. Entropy is a formulation—when used in land use applications—designed to quantify land use mix in a man- ner that the lowest values represent single-use development and progressively higher values indi- cate increasing land use mix at a scale determined by the analyst. Other descriptors used include jobs/housing balance (a ratio) and distance to stores.

Diversity and design are, in the meta-analysis elasticities derivations (Table 16-38, 1st entry), both more strongly related to prevalence and mode choice of walk trips than density. For walking, the relationship holds for all three measures examined: land use mix, jobs-housing balance, and dis- tance to a store, with walk elasticities in the 0.15 to 0.25 range. Basically these elasticities indicate that where there are more local opportunities to meet daily needs there will be more walking.37 Only one diversity measure was covered by enough applicable studies to allow weighted average elasticity calculation for transit use, and at 0.12 the elasticity shows a modest positive relationship. The reason for the positive relationship with transit use does not immediately stand out, but one possibility is that ready availability of local shopping and services—especially if along the walk to and from transit—makes not having one’s auto at hand for errand running during the commute and at work more feasible (Ewing and Cervero, 2010). As with density, three of nine 2002–2006 reviews (Table 16-38, 2nd entry) pointed to mixed land use as important for more walking. Again the synthesis researchers posit that this is a manifesta- tion of the demonstrated association of walking with proximity of destinations, which land use mix serves to intensify. Individual studies reviewed supported these conclusions (for the most part) for both utilitarian and recreational walking, although there were a number of inconsistent or insignificant results for recreational walking in particular. The generally positive relationship between land use mix and recreational walking was probably not directly caused by proximity of primary non-residential destinations, given the lack of consistent evidence that destination prox- imity is associated with recreational walking (Saelens and Handy, 2008). The 5th and 6th entries in Table 16-39 from Chapter 15 address diversity and are the counterpart to the 2nd and 3rd entries. The first-listed of the study pairs together found minimal direct density effect on NMT mode choice but elasticities on the order of +0.2 for land use balance and accessibil- ity (Kockelman, 1996). The second-listed of the pairs, in contrast, found broader impact for den- sity than diversity (Frank, 1994). The 7th and 8th studies entered in Table 16-39 address the importance of retail proximity to housing and mixed land use in general for engendering walk activity. Figure 16-6 graphs relationships devel- oped in the 11-city study (8th entry) for commute trips. Both density and land use mix have compa- rable importance for the work commute in this illustration (Parsons Brinckerhoff et al., 1996a). The 9th entry in Table 16-39 again highlights the particular sensitivity of choice of walk versus auto for access to transit. Not only is choice of transit access mode an aspect of travel behavior sensitive to land use characteristics in general, it is shown to be highly sensitive to mix in particular. The walk elasticity for access/egress to BART stations relative to an index of mix was, at 1.1, estimated to lie in the elastic range: very sensitive (Parsons Brinckerhoff et al., 1996b). The 10th and 11th entries provide additional examples of apparently positive influence of mix on walking, the for- mer involving Houston observations within suburban employment centers in the presence of detri- mental design features, and the latter, Seattle area observations with land use mix in the presence of supportive street layouts (Cervero, 1988, Rutherford, McCormack and Wilkinson, 1997). 16-167 37 The meta-analysis researchers conclude that the positive elasticity between walking and jobs-housing bal- ance demonstrates the importance of linking where people live and work (Ewing and Cervero, 2010). This certainly has some validity, but it must also be remembered that jobs in with housing may also be indicative of the presence of stores and services.

Additional findings concerning land use diversity (and density) impacts in selected individual stud- ies are shown at the outset of Table 16-40. The 1st and 2nd entries are of particular interest because the one research effort sought to quantify land use and physical environment effects on walking, while the other did the same for bicycling, with much commonality of data and procedures. These two public-health-oriented studies examined the incidence of walking in terms of nonwalkers, walkers not meeting 150-minutes-per-week walking exercise recommendations, and walkers meet- ing the recommendations. Utilitarian walking, specifically including walking for transit access, and recreational/exercise walking were both included. Incidence of bicycling was identified as cycling at least once a week (Moudon, et al., 2007, Moudon, et al., 2005). The pairing of these two research studies allows exploring differences and types of impacts as they pertain to these two primary NMT modes. 16-168 Figure 16-6 Probability of commuting by walking or bicycling as a function of density, land use mix, and auto ownership. Note: Based on modeling of survey results from the 11 metropolitan areas (MSAs or CMSAs) of Boston- Lawrence-Lowell, Dallas, Detroit, Los Angeles- Long Beach, Fort Worth-Arlington, Minneapolis- St. Paul, Philadelphia, Phoenix, San Francisco- Oakland, Tampa-St. Petersburg, and Washington, DC-MD-VA. Source: Parsons Brinckerhoff et al. (1996a).

16-169 Table 16-40 Selection of Additional Findings from Transportation and Physical Activity Research on Relationships between Pedestrian- and Bicycle-Friendly Neighborhood (NBH) Characteristics and Adult Walking/Bicycling Study (Date) Process (Limitations) Key Findings 1. Moudon et al. (2007) (see this section for more information) Cross-sectional analysis of walking activity, socio-demographics, attitudes, and objectively measured environmental variables covering 608 adults in King County, WA. (Extent of walking self-reported.) Among neighborhood environmental measures found to be most related to walking were closeness to grocery stores, restaurants, and retail; lack of office building dominance; and density of the respondent’s home parcel. 2. Moudon et al. (2005) (see this section for more information) Similar to Moudon et al. (2007) but focused on cycling (at least once a week versus less), with inclusion of perceived environmental variables. (Some evidence in 1/3 of cyclists of neighborhood “self-selection” for recreational facility accessibility.) Cycling appears to be an individual choice that “is only moderately associated with the neighborhood environment,” at least in the Puget Sound/U.S. context. Trail proximity and certain commercial use groupings had significant positive relationships. 3. Cervero and Duncan (2003), and SR 282 Cross-sectional S.F. Bay Area survey and discrete-choice modeling controlling for disability, race, gender, and auto ownership and examining walking and bicycling separately. (Street-scale design elements not examined.) Deterrents found were distance (walk and bike), slope (walk), rain (walk), and darkness (bike). Supportive factors included origin land use mix (walk), recreational/social purpose (both), or eating/shopping purpose (walk). Weaker factors included small blocks. 4. Krizek and Johnson (2006) Objectively measured Minneapolis- St. Paul household proximity to nearest neighborhood retail employ- ment site was related to home-based walk trip activity. (Unfactored samples from year 2000 regional survey; 205 exhibited walk trips.) Walk trip activity more than twice as likely for individuals in households less than 200 meters (1/8 mile) from nearest retail as compared to those greater than 600 meters (3/8 mile) from retail. Very close proximity (<1/8 mile) was found to be most important. 5. Handy et al. – 1998 and Handy and Clifton – 2001 as summarized in Heath et al. (2006) and per SR 282 Linear regression analysis of cross- sectional 1994 recall mail survey with responses from 1,368 residents in 6 NBHs of Austin, TX; 2 tradition- al, 2 early modern, 2 late modern. Closeness of stores measured using GIS; other environmental variables were perceived measures. (15% and 29% of variation explained.) Perceived safety, shade, and presence of people positive for strolling frequency; perceived presence of stores, walking incentive, walking comfort, plus closeness of stores positive for walk-to-store frequency; residence in Old West Austin NBH positive for both types. Living in tradi- tional NBH associated with 163% more walking to store than modern NBHs. 6. Ball et al. – 2001 as summarized per SR 282 Logistic regression relating neigh- borhood aesthetics, congeniality, and access to facilities/paths to incidence of walking in Australia. (Measures based on perceptions.) High aesthetics (5-point scale) linked with 41% more likelihood of walking than low; high convenience to facilities including paths linked with 36% more likelihood of walking than low. 7. Handy – 1996 as summarized in Heath et al. (2006) and per SR 282 Cross-sectional ANOVA analysis of 4 San Francisco Bay Area traditional NBHs with small, close-by shopping centers, vs. suburban NBHs and shopping, controlling for type of household. (Recall phone survey.) NBH type not significant for strolling (1% to 5% more in traditional NBHs). Respondents reported walking to stores during the month almost 50% more in traditional NBHs; walk to store frequency was 182% greater. (continued on next page)

16-170 Table 16-40 (Continued) Study (Date) Process (Limitations) Key Findings 9. Doyle et al. – 2006 as summarized in Saelens and Handy (2008) Related incidence of walking 1 unin- terrupted mile in previous month to a composite walkability measure (block size, percentage of blocks <0.01 square miles in area, and intersections per road mile). (Side- walk availability not considered.) Among 35 large U.S. counties, found higher walking likelihood, even after controlling for individual demograph- ics, for residents of counties having higher walkability scores, especially lifelong residents. Walkability had a stronger effect on walking than crime. 10. Hanson and Schwab – 1987 as summarized per SR 282 Analyzed 35-day travel diary survey covering 278 Swedish households. Calculated accessibility using number of establishments and Euclidean distance. (Evaluation based on correlation coefficients.) Percent of all stops by NMT modes positively related to home-based accessibility; percent of all work-based stops by NMT modes positively related to both home-based and work-based accessibility. 11. Krizek – 2003 as summarized per SR 282 Tested, with socio-demographic controls and same data set as Krizek – 2000 (see 16th entry), both NBH and regional accessibility variables. (Composite NBH accessibility measure.) Built environment accessibility variables showed no significant relationship to percent of trips by walking in this study of Puget Sound Area travel survey panel data. 12. Greenwald and Boarnet (2001), and SR 282 Probit model cross-sectional analysis of 1995 Portland, OR, regional travel survey and Pedestrian Environmental Factors (PEFs), small- and larger-area residential and retail densities, and survey respondent median walk trip characteristics reflecting trip cost. (Correlations reduced clarity.) Walk trip frequency positively and significantly related to TAZ (or block group) PEFs, NBH population and retail density, percent area within 1/4 mile in grid plan, and median walk distance and speed. Positive but lesser and not significant relationship with larger area (non-localized) densities. 13. Craig et al. – 2002 as summarized per SR 282 Evaluated 28 NBHs in Canada on a 10-point ecologic scale and related the results to walk-to-work rates. (No non-work trip data.) Controlling for socio-demographic factors, a 1-unit increase in the ecologic score was associated with a 25 percent- age point increase in walking. 14. Clifton and Dill – 2005 as summarized in Saelens and Handy (2008) Using a composite sample drawing from NHTS national, Portland, OR, and Baltimore data, and controlling for demographics, modeled number of walk trips relative to various objective and perceived measures. (Conflicting transit access and density results for different walking measures.) Higher walk trip incidence related to high housing densities and land use mix, good transit access and pedestrian environment, greater park access, and perception that lack of sidewalk is not a problem. Walk trips on survey travel day negatively related to street connec- tivity, transit access, and for men only, density and percent vacant. 15. Berke et al., (2007b) Cross-sectional analysis of fine- grained walkability scores for King County, WA, in conjunction with Adult Changes in Thought (ACT) cohort study data, including measures of activity. (Extent of walking self-reported.) Found, after controlling for various socio-economic and health status variables, a significant positive association between NBH walkability and a report of any walking session over 15 minutes long during the week. 8. Berrigan and Troiano (2002), and SR 282 Used logistic regression to relate physical activity in U.S. to year home built. (Year home built a proxy for multiple urban form factors from core area accessibility to NBH layout and transit availability.) Controlling for socio-demographic factors, persons living in a pre-1946 home walk 43% more and persons living in a 1946-1973 home walk 36% more relative to those living in a post- 1973 home. (Non-rural homes only).

The walking-focused King County, Washington, study (Table 16-40, 1st entry), using highly dis- aggregated data on trip origin and destination characteristics along with information on condi- tions en route, is notable for isolating impacts largely consistent with the larger body of research on land use effects. Higher net residential density, measured on the basis of each survey respon- dent’s home parcel of land, was strongly and significantly related to more walking.38 Living closer to a grocery-shopping or eating/drinking opportunity, along with typically associated retail and banking, was fairly consistently associated positively with walking. Directness of walking to the closest grocery store and the nearest school, measured in terms of airline distance versus network distance, was positively associated. Living closer to office-oriented development, particularly large sites, was negatively associated. Not significant to walking were recreational, institutional, or auto- oriented-retail land uses. Traveling through areas with more complete sidewalks along major streets (the only streets for which sidewalk data were available) was positively associated with additional walking (Moudon, et al., 2007). The companion cycling-focused study (Table 16-40, 2nd entry) did not find any of the objective distance-based fine-grained accessibility/diversity measures to be significant to prevalence of bicycling. Perceived presence of grocery stores and schools showed negative associations to cycling. A handful of disparate land use diversity measures such as number of convenience store parcels in the neighborhood and parcels within the nearest office/hospital complex were posi- tively related to cycling. A possible unifying factor was that land use descriptors positively related for cyclists were not for walkers, and vice versa (Moudon, et al., 2005). This disparity was not nec- essarily an illogical outcome considering the package-carrying limitations associated with bicy- cling and the substantively different trip length distributions that normally characterize bicycle trips as compared to pedestrian trips. 16-171 38 This finding of significant density effects is not necessarily inconsistent with disaggregate analysis findings (presented above under “Density”) of little density impact. Research efforts such as those by Ewing and Cervero, and Kockelman, focused on walk-only trips, whereas the walk activity research by Moudon et al. explicitly included transit-access walking in the dependent variable (Ewing and Cervero, 2010; Kockelman, 1996; Moudon, et al., 2007). As stressed in the “Density” discussion, more intensive (and thus better) transit service is normally provided where residential densities are higher. The transit-access component of walk- ing activity would therefore logically be positively associated with density. Table 16-40 (Continued) Study (Date) Process (Limitations) Key Findings 16. Krizek (2000), and SR 282 An “LADUF rating” (land use, density, and urban form score based on housing density, employment presence/mix, and block size) related to percent of trips by alternative mode (NMT and transit) as obtained in longitudinal Puget Sound panel survey, and to shifts with change in LADUF rating. (Small LADUF-change sample sizes of 19 to 84 persons moving.) Raw 1997 alternative mode shares were, for high LADUF, 29%; medium, LADUF, 14%; low LADUF, 6%. Drop in alternative mode share for 1989-1997 panel members moving from high to medium LADUF (before- and after- move time-series data) was 9.9 percentage points. Other mode shifts with LADUF change were logical in sign but not statistically significant. Note: Where substantial additional information on individual studies is provided in text and tables or figures, this is noted — and the location within the chapter is given — in the first column. Sources: As indicated in the first column. The notation “SR 282” is shorthand for Committee on Physical Activity, Health, Transportation, and Land Use (2005) together with Handy (2004).

No objectively-measured infrastructure or route-related characteristics showed significant associ- ations with cycling except for closeness to an off-road trail, which was a positive. Indeed, it was found that 33 percent of cyclists—as compared to 17 percent of non-cyclists—had considered recre- ational amenities when choosing their current residence. Also related significantly and positively to more cycling was perceived closeness of trails and bike lanes. Perceived traffic problems and pres- ence of auto-oriented facilities were negatively related to cycling at the two ends of the spectrum— major traffic issues/many facilities and also insignificant traffic issues/few facilities. The researchers conclude “that the decision to bicycle seems to rest largely on personal, and not environmental, factors” although “improving the built and transportation environment for cycling may still help promote general increases . . .” An unexpectedly high 21 percent of research survey respondents proved to be cyclists, according to the “at least once a week” definition, but most bicy- cle trips were for recreation. Limited infrastructure for cycling in the study area may have hindered ferreting out significant relationships with land use and route-related characteristics (Moudon, et al., 2005). The 3rd entry in Table 16-40 also pertains to research examining both walking and bicycling. In this San Francisco Bay Area study, trip origin-area land use mix was found positively associated with walking, but not significantly with cycling. The only other strongly significant physical envi- ronment factors, excluding rain and darkness, were slope (negatively associated with walking) and trip distance (negatively associated with both walking and bicycling). The 4th entry presents, on the basis of research from Minneapolis-St. Paul, a commonly encountered overall positive relation- ship between closeness of retail and walking. As noted, very close proximity was determined to be especially important, a finding of particular relevance to the upcoming “Design” and destina- tion accessibility discussions. In the more fully specified of three research models, the odds of walking at greater than 600 meters (3/8 mile) from the nearest retail establishment were 2/5ths the odds at less than 200 meters (1/8 mile), while the odds of walking at intermediate distances were still only 1/2 to 3/5ths the odds at less than 200 meters. While only the odds for shortest versus longest distances from closest retail were statistically significant, the odds for all four studied dis- tance categories exhibited logical interrelationships (Krizek and Johnson, 2006). Design Design, as a land use descriptor, covers small-to-intermediate-scale transportation network and streetscape characteristics. Measures may indicate sidewalk extent, streetscape features such as building setbacks and parking front or rear, and NMT network continuity. Continuity in particular is often represented by surrogates such as average block size, intersections per unit area, or preva- lence of four-way intersections. The relatively substantial 0.39 elasticity obtained in the meta-analysis derivations (Table 16-38, 1st entry) for the effect of intersection/street density on walking presumably results from the impor- tance of a fine-grained infrastructure for walk trip efficiency (Ewing and Cervero, 2010). The same importance pertains to transit use as well (see Table 16-42 below for elasticities), given the desirabil- ity of being able to walk directly to the nearest bus stop. The substantial positive relationship between transit use and prevalence of four-way intersections may relate to the efficiency of bus services pos- sible in a grid system of streets as well as to efficiency of walk access to transit. The slightly negative relationship (based on five studies) between the walk-only mode and prevalence of four-way inter- sections may suggest that other measures do a better job of representing pedestrian interconnectiv- ity or it may simply be an artifact of analysis based on small numbers of available studies within individual categories. 16-172

Six of nine reviews published in the 2002–2006 period (Table 16-38, 2nd entry) pointed to aesthetics, or attractiveness of the environment, as being associated positively with walking. Sidewalks and net- work connectivity were similarly found to be positively correlated. The researchers note that connec- tivity affects proximity, separately identified as being important, by virtue of providing more direct and thus shorter routes. Nevertheless, in both these matters of design, there was substantial variabil- ity across studies and an indication that different attributes of the built environment are important for recreational as compared to utilitarian walking. In the individual studies reviewed, little or no evi- dence was found for correlation between utilitarian walking and either aesthetics or conditions of pedestrian infrastructure and traffic. Relationships with route/network connectivity and presence of parks or open space were equivocal. Recreational walking findings were more limited, but appeared to support associations with aesthetics and quality of pedestrian infrastructure. Positive recreational walking relationships with connectivity measures were identified in just two more individual studies than the number showing insignificant or negative findings. No recreational walking associations were established in these studies with parks, open space, or traffic conditions (Saelens and Handy, 2008). The SR 282 review (Table 16-38, 3rd entry) highlighted traditional, transit-served, and walkable neighborhoods—typically characterized by grid street systems—as being positively associated with greater use of active transportation (Committee on Physical Activity, Health, Transportation, and Land Use, 2005). Caveats presented in the “Analytical Considerations” section of Chapter 15, “Land Use and Site Design,” may pertain. There it was noted that weak attention to socio-economic variables such as income and family size, or other study limitations in some earlier investigations, results in a need for careful interpretation of certain neighborhood-type study conclusions. Full transferability of traditional neighborhood travel characteristics findings to newer land-use con- structs such as neo-traditional communities cannot be taken for granted and must be viewed in appropriate socio-demographic and regional accessibility contexts. The 12th and 13th extractions from Chapter 15 in Table 16-39 provide various individual research perspectives on design impacts, as do the 5th through 9th individual studies in Table 16-40. Design aspects addressed range from aesthetics to neighborhood type to system connectivity. Three of the seven research efforts explicitly or implicitly tested measures of aesthetics and produced findings ranging from qualitative identification of positive effect to 25 and even 41 percent positive differ- entials in NMT or walk shares in prescribed circumstances and comparisons (Parsons Brinckerhoff, 1996a, Heath et al., 2006, Committee on Physical Activity, Health, Transportation, and Land Use, 2005). Three of the research efforts explore comparisons of traditional neighborhood design (TND) with more typical suburban designs.39 One of the seven design-oriented research efforts, the 9th study in Table 16-40, focused exclusively— other than taking demographics into account—on street and block layout. It used a composite walkabil- ity score, encompassing three different measures of street system connectivity, and related it to walking activity. Street interconnectivity served as a surrogate for the extent to which the pedestrian system was tightly interconnected. The study found higher incidence of walking among counties with higher scores (Saelens and Handy, 2008). 16-173 39 Further insights into the Austin, Texas, findings (5th entry in Table 16-40) are afforded by additional evaluations involving a reexamination of the research data. See the “Sidewalks and Along-Street Walking” subsection under “Sidewalk Coverage and Traffic Conditions,” starting with the 1st entry in Table 16-2 (Cao, Handy, and Mokhtarian, 2006). Also, further exploration of factors related to aesthetics is found in the “Underlying Traveler Response Factors” section (see “Environmental Factors”—“Surroundings Environment”—“Ambiance”).

Other “D’s” The two additional “D’s” covered here are destination accessibility and distance to transit. Destination accessibility is a measure of ease of access to jobs, shopping, and other non-home destinations—“attractions” in demand modeling parlance. Common regional destination accessibil- ity measures include attractions reachable within a given mode-specific travel time and attraction accessibility as calculated using a gravity-model-type of formulation. Also used is the simpler mea- sure of distance to the CBD (Ewing and Cervero, 2010). Local accessibility measures, often more apro- pos for NMT analysis, can range from jobs within a given walkable/bikeable distance to distance to the nearest store. Distance to transit, often treated as a transit service parameter rather than a land use descriptor, is typically measured as distance to the nearest bus stop or rail station. Alternatively, stop, station, or route coverage measures may be used. The elasticities meta-analysis (Table 16-38, 1st entry) finds all investigated forms of regional accessibil- ity measures, including simple closeness to downtown, to be negatively related to VMT. Data was insufficient, however, for computation of average elasticities of walking and transit use to these regional measures. Accessibility measures, most particularly closeness to downtown, probably act as a surro- gate for lack of auto dependency and presence of impediments to auto use such as congestion and park- ing costs. Walking was found positively related to jobs within 1 mile, with an elasticity of 0.15, this being the only walking-scale average accessibility elasticity derivation allowable given numbers of studies available. Distance to the nearest transit stop produced elasticities with the expected signs. The most straightforward explanation pertains to the transit use elasticity: that closeness of transit service sup- ports transit use (Ewing and Cervero, 2010). Five of nine reviews published between 2002 and 2006 (Table 16-38, 2nd entry), as previously noted, consistently found accessibility to be significantly associated with additional walking. The accessibil- ity measures used effectively described closeness to destinations. Individual studies examined also found consistent positive associations between walking and closeness to non-residential destinations, except in the case of research focused on recreation walking, where little or no identifiable relationship was demonstrable (Saelens and Handy, 2008). SR 282 (Table 16-38, 3rd entry) used a slightly different approach to trip purpose differentiation, noting that studies from the transportation literature tended to focus on utilitarian travel, while physical activity research in the preceding years was primarily concerned with walking and per- haps cycling for recreation and exercise. (The physical activity research also made more use of accessibility measures constructed on the basis of survey respondent perceptions.) In the trans- portation literature significant associations with walking and cycling tended to involve destina- tions such as stores, bus stops, and parks, particularly in the case of shopping and schoolchild trips. In the physical activity literature, the more significant destinations for active transportation were partially the same (parks, local shopping, and transit stops) but also specifically included bicycle paths/trails (Committee on Physical Activity, Health, Transportation, and Land Use, 2005). However, there were no reported overall conclusions as to whether the importance of transit stops in both categories of studies was a reflection of transit access activity or some secondary effect of transit availability on choice of walking or cycling as the primary mode. Several of the individual studies in Table 16-40 used local-accessibility measures to examine effects of land use mix, but the 10th and 11th entries explicitly investigated accessibility to multiple poten- tial destinations. The Swedish research found both home-based and work-based accessibilities to have a positive relationship with NMT mode use. The study utilizing Washington State Puget Sound Area data found no significance for regional accessibility or for a composite neighborhood accessi- bility measure (Committee on Physical Activity, Health, Transportation, and Land Use, 2005). 16-174

Overall Neighborhood Environment The meta-analysis derivation of built environment elasticities (see Table 16-38, 1st entry, and Table 16-42) found them all to be inelastic—the impact always proportionally less than the stimulus provided by changing any one particular land use or urban design characteristic. In fact, most of the elasticities are much smaller than the 0.39 value for the walking response to greater intersection or street density. Nevertheless, “the combined effect of several built environment variables on travel could be quite large” (Ewing and Cervero, 2010). It is of interest to note that the elasticities for walking and transit use are almost all larger, some substantially so (especially for walking), than corresponding elasticities for VMT reduction. This outcome may reflect complex factors such as differential shifting from carpool- ing versus driving, or short versus long vehicle trips, to active transportation modes. Alternatively there may be a degree of elevated trip-making (trip generation) where walking and transit use are easy. Transit Oriented Development (TOD) is a particular application of “smart growth” land use and design precepts that should by definition cover enhancement of most of the “D’s” within the overall neighborhood environment. Properly implemented TODs focus on provision of higher densities within walking distance of the transit station or stop, follow guidelines suggesting land use diver- sity, design for quality NMT connections to the stop, offer good regional transit accessibility and tran- sit stop accessibility, and obviously incorporate many housing locations in close proximity to transit. TOD is addressed within this chapter in the “Pedestrian/Bicycle Linkages with Transit” subsection under “Transit Oriented Development” as well as being the topic of Chapter 17. There it can be seen that researched TOD outcomes are quite consistent with those identified for pedestrian/bicycle friendly neighborhoods in general. The TOD objective of enhancing transit rid- ership as a primary travel mode has been very successfully met in many TODs, and less success- fully in others, but with only one small-scale partially negative outcome reported. Whatever the degree of transit mode share increase a particular TOD achieves, logic and substantial experience indicate that the walk mode of access share for users of the primary transit stop or station is very high even in suburban locations. Typically TOD studies report walk access shares between about 70 and 100 percent walk. Finally, comparative and prior-circumstance information indicates that trips made exclusively by walk or bicycle are more prevalent in the typical TOD than conventional development. The observed differences range from roughly double in the case of non-work trips to and from TODs outside the central area of Portland, Oregon, to no significant difference for work trips in the Pleasant Hill TOD in the East Bay Hills of the San Francisco region. The 14th entry in Table 16-39 (from Chapter 15), the paired-communities analysis of the Rockridge and Walnut Creek communities in the San Francisco East Bay, provides texture for the TOD find- ings even though neither community was formally planned as one. Rockridge, a TND neighbor- hood, grew up as a “streetcar suburb,” with retail and other development oriented toward surface transit. Walnut Creek is of conventional suburban design (CSD). Today each has a centrally located Bay Area Rapid Transit (BART) station. They are on the same BART line, albeit separated by the first range of East Bay Hills. As indicated in Table 16-39 the walk/bike-only mode choice for non- work travel was found to be 10 percent in the TND neighborhood versus 2 percent in the CSD area. The comparable NMT shares for work purpose trips were 7 percent (TND) versus 1 percent (CSD). For non-work trips under 2 miles in length, a 52 percent NMT share was encountered in the TND environment versus 17 percent for the CSD community. Also highly significant for local area walk versus driving activity is the previously introduced finding of 31 percent TND neighborhood walk share for BART station access, compared to 13 percent walk access for the station in a CSD envi- ronment. This differential is particularly noteworthy considering the similarity of the work com- mute rail transit shares at 21 percent for the Rockridge TND neighborhood and 20 percent for the Walnut Creek CSD community (Cervero and Radisch, 1995). 16-175

Three of nine reviews published in the 2002–2006 period (Table 16-38, 2nd entry) concluded that neighborhood-based composite walkability measures were positively correlated with walking (Saelens and Handy, 2008). The 15th and 16th entries in Table 16-39 from Chapter 15 present two additional perspectives on overall neighborhood environment effects. One suggests that attitudes are most important in NMT choice (Kitamura, Mokhtarian, and Laidet, 1994). The other reports on two travel demand modeling efforts that successfully used multi-faceted neighborhood environ- ment measures in estimating NMT mode shares. The environment measures involved are gener- ally known as Pedestrian Environmental Factors (PEFs) (Cambridge Systematics, Putman Associates, and Calthorpe Associates, 1992, Cambridge Systematics et al., 2002). The last five individual studies summarized in Table 16-40, the 12th through 16th entries, also show relationships between better NMT environment scores—sometimes in combination with other measures—and higher walk, or walk and bike, mode shares (or walking activity in one case). The 16th table entry is of special interest because of its use of Puget Sound panel time series travel data rather than cross-sectional data. The vast majority of land use and transportation research relies on cross- sectional studies. In such studies, although which caused what may often seem fairly obvious, causal- ity cannot be absolutely demonstrated. With panel survey data over time, however, the effects reported involve the same respondents before and after moving from one neighborhood type to another. All respondent moves from one rating category to another resulted in logical shifts in walking, bicycling, and transit shares. Even though the numbers of movers were small and shifts reached statistical signif- icance for only one category of change, the commonality of logical outcomes gives fairly strong evi- dence that change from less to more favorable neighborhood environments is linked to increases in alternative travel mode use, and vice versa (Krizek, 2000). The Built Environment and Child Walking and Bicycling Research on choice of mode for children and adolescents traveling to school demonstrates clearly that distance between home and school is a dominant factor in the choice to walk or not. The SUNY synthesis of environmental influences on children’s physical activity (Table 16-38, 4th entry) found that three out of three studies which directly or indirectly examined the role of distance found walking to be inversely related to distance or walking time (Davison and Lawson, 2006). Less than a half-decade later, the WSDOT review of SRTS-related research (Table 16-38, 5th entry) located 19 studies that examined distance between home and school. All 19 found a significant negative relationship between this distance and active transportation to school (ATS). The WSDOT reviewers concluded: “Distance from a child’s home to school is the strongest predic- tor of ATS.” They report that 47 percent of the decline in ATS between 1969 and 2001 is explain- able on the basis of greater home and school separation (Moudon, Stewart, and Lin, 2010).40 The synthesis of nine reviews and other studies (Table 16-38, 2nd entry) reported that the relationship has not been “universally found,” but agreed that the preponderance of evidence shows proxim- ity to exhibit a consistently positive relationship (Saelens and Handy, 2008). ATS is primarily com- prised of walking and cycling, but may include other NMT modes such as scooters where allowed by schools and parents. 16-176 40 Walking and bicycling to school by children up to age 18 declined from just under 41 percent in 1969 to 13 per- cent in 2001 (Moudon, Stewart, and Lin, 2010) and 10 percent in 2009 (Kuzmyak et al., 2011). (See Table 16-88 in “Extent of Bicycling” under “Related Information and Impacts”—“Extent of Walking and Bicycling.”)

Two studies examined in the SUNY synthesis examined possible links between population den- sity and ATS. One found no significant association when considering densities in the immediate area around children’s homes. The other found walking and cycling to school to be more preva- lent where densities were higher (Davison and Lawson, 2006). It is logical that there should be a positive relationship, in that higher densities put more students close to their schools (U.S. Environmental Protection Agency, 2003). In that regard, it is of note that a national study reviewed for the WSDOT synthesis (Moudon, Stewart, and Lin, 2010) found high negative sensitivity to walk time to school (a marker for distance) and also determined higher density to have a significant pos- itive relationship to more walking (McDonald, 2008). Other findings on land use and design relationships to walking and cycling by children are equiv- ocal, aside from attractiveness of a good pedestrian infrastructure and associated traffic safety, for which there is strong evidence (Saelens and Handy, 2008). Three of four studies in the SUNY syn- thesis (Table 16-38, 4th entry) did find a significant positive relationship between children’s phys- ical activity, including active transportation, and destination proximity. Destination measures ranged from retail to transit stops. Only two of four studies found a significant effect for street con- nectivity in the expected direction (Davison and Lawson, 2006). The WSDOT review (Table 16-38, 5th entry) reached the conclusion that a “majority” of neighborhood characteristics researched have shown either a relationship to ATS in the expected direction or no significant association. Characteristics tested include urbanization level, population density, land use mix, and street lay- out (Moudon, Stewart, and Lin, 2010). Obviously a number of the same few studies have been assessed more than once. Table 16-41 encapsulates six of the available studies on interrelationships between patterns of childhood travel and the extent of neighborhood pedestrian and bicycle friendliness.41 16-177 41 Additional gleanings from the limited knowledge base available on active transportation by children are pri- marily concentrated in specific child-related discussions within this chapter. One, later in this “Response by Type of NMT Strategy” section, is under “NMT Policies and Programs”—“Schoolchild-Focused Programs.” Others are in the “Underlying Traveler Response Factors” section (see “Behavioral Paradigms”—“The Travel Choice Making of and for Children,” and “Trip Factors”—“Schoolchild Trip Factors,” plus brief entries in the “User Factors” subsection under “Age” and “Ethnicity”). The final key child travel discussion is in the “Related Information and Impacts” section under “Public Health Issues and Relationships”—“Health Benefits for Children of Enhanced NMT Systems and Policies.”

16-178 Table 16-41 Selection of Additional Findings from Transportation and Physical Activity Research on Relationships between Pedestrian- and Bicycle-Friendly Neighborhood (NBH) Characteristics and Child Walking/Bicycling Study (Date) Process (Limitations) Key Findings 1. Braza et al. – 2004 as summariz- ed by Davison and Lawson (2006) Employed bivariate modeling of ob- jective measures of California school area characteristics and surveyed rates of walking/biking to school. Walking and biking rates to school were found to be associated with higher population and intersection densities, but not school size. 2. Carver et al. – 2005 as summarized by Davison and Lawson (2006) Cross-sectional analysis of parent and child perceptions of various facilities and environmental conditions. (Used self-reported physical activity measures as well as p