National Academies Press: OpenBook
« Previous: Response by Type of NMT Strategy
Page 227
Suggested Citation:"Underlying Traveler Response Factors." 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 227
Page 228
Suggested Citation:"Underlying Traveler Response Factors." 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 228
Page 229
Suggested Citation:"Underlying Traveler Response Factors." 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 229
Page 230
Suggested Citation:"Underlying Traveler Response Factors." 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 230
Page 231
Suggested Citation:"Underlying Traveler Response Factors." 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 231
Page 232
Suggested Citation:"Underlying Traveler Response Factors." 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 232
Page 233
Suggested Citation:"Underlying Traveler Response Factors." 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 233
Page 234
Suggested Citation:"Underlying Traveler Response Factors." 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 234
Page 235
Suggested Citation:"Underlying Traveler Response Factors." 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 235
Page 236
Suggested Citation:"Underlying Traveler Response Factors." 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 236
Page 237
Suggested Citation:"Underlying Traveler Response Factors." 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 237
Page 238
Suggested Citation:"Underlying Traveler Response Factors." 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 238
Page 239
Suggested Citation:"Underlying Traveler Response Factors." 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 239
Page 240
Suggested Citation:"Underlying Traveler Response Factors." 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 240
Page 241
Suggested Citation:"Underlying Traveler Response Factors." 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 241
Page 242
Suggested Citation:"Underlying Traveler Response Factors." 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 242
Page 243
Suggested Citation:"Underlying Traveler Response Factors." 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 243
Page 244
Suggested Citation:"Underlying Traveler Response Factors." 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 244
Page 245
Suggested Citation:"Underlying Traveler Response Factors." 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 245
Page 246
Suggested Citation:"Underlying Traveler Response Factors." 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 246
Page 247
Suggested Citation:"Underlying Traveler Response Factors." 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 247
Page 248
Suggested Citation:"Underlying Traveler Response Factors." 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 248
Page 249
Suggested Citation:"Underlying Traveler Response Factors." 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 249
Page 250
Suggested Citation:"Underlying Traveler Response Factors." 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 250
Page 251
Suggested Citation:"Underlying Traveler Response Factors." 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 251
Page 252
Suggested Citation:"Underlying Traveler Response Factors." 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 252
Page 253
Suggested Citation:"Underlying Traveler Response Factors." 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 253
Page 254
Suggested Citation:"Underlying Traveler Response Factors." 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 254
Page 255
Suggested Citation:"Underlying Traveler Response Factors." 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 255
Page 256
Suggested Citation:"Underlying Traveler Response Factors." 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 256
Page 257
Suggested Citation:"Underlying Traveler Response Factors." 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 257
Page 258
Suggested Citation:"Underlying Traveler Response Factors." 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 258
Page 259
Suggested Citation:"Underlying Traveler Response Factors." 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 259
Page 260
Suggested Citation:"Underlying Traveler Response Factors." 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 260
Page 261
Suggested Citation:"Underlying Traveler Response Factors." 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 261
Page 262
Suggested Citation:"Underlying Traveler Response Factors." 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 262
Page 263
Suggested Citation:"Underlying Traveler Response Factors." 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 263
Page 264
Suggested Citation:"Underlying Traveler Response Factors." 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 264
Page 265
Suggested Citation:"Underlying Traveler Response Factors." 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 265
Page 266
Suggested Citation:"Underlying Traveler Response Factors." 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 266
Page 267
Suggested Citation:"Underlying Traveler Response Factors." 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 267
Page 268
Suggested Citation:"Underlying Traveler Response Factors." 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 268
Page 269
Suggested Citation:"Underlying Traveler Response Factors." 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 269
Page 270
Suggested Citation:"Underlying Traveler Response Factors." 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 270
Page 271
Suggested Citation:"Underlying Traveler Response Factors." 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 271
Page 272
Suggested Citation:"Underlying Traveler Response Factors." 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 272
Page 273
Suggested Citation:"Underlying Traveler Response Factors." 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 273
Page 274
Suggested Citation:"Underlying Traveler Response Factors." 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 274
Page 275
Suggested Citation:"Underlying Traveler Response Factors." 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 275
Page 276
Suggested Citation:"Underlying Traveler Response Factors." 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 276
Page 277
Suggested Citation:"Underlying Traveler Response Factors." 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 277
Page 278
Suggested Citation:"Underlying Traveler Response Factors." 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 278
Page 279
Suggested Citation:"Underlying Traveler Response Factors." 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 279
Page 280
Suggested Citation:"Underlying Traveler Response Factors." 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 280
Page 281
Suggested Citation:"Underlying Traveler Response Factors." 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 281
Page 282
Suggested Citation:"Underlying Traveler Response Factors." 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 282
Page 283
Suggested Citation:"Underlying Traveler Response Factors." 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 283
Page 284
Suggested Citation:"Underlying Traveler Response Factors." 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 284
Page 285
Suggested Citation:"Underlying Traveler Response Factors." 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 285
Page 286
Suggested Citation:"Underlying Traveler Response Factors." 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 286
Page 287
Suggested Citation:"Underlying Traveler Response Factors." 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 287
Page 288
Suggested Citation:"Underlying Traveler Response Factors." 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 288
Page 289
Suggested Citation:"Underlying Traveler Response Factors." 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 289
Page 290
Suggested Citation:"Underlying Traveler Response Factors." 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 290
Page 291
Suggested Citation:"Underlying Traveler Response Factors." 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 291
Page 292
Suggested Citation:"Underlying Traveler Response Factors." 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 292
Page 293
Suggested Citation:"Underlying Traveler Response Factors." 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 293
Page 294
Suggested Citation:"Underlying Traveler Response Factors." 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 294
Page 295
Suggested Citation:"Underlying Traveler Response Factors." 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 295
Page 296
Suggested Citation:"Underlying Traveler Response Factors." 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 296

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.

The SPARC researchers concluded that people can be encouraged to walk more through interven- tions targeted at individuals or households and tailored to individual needs. Evidence was found less convincing in the case of measures taken at the institutional level, whether workplace, school, or community. The more substantial increases relative to baseline walking occurred with the most sedentary people, who also were the subjects of many of the general walking interventions. Targeting persons more motivated to change, common in programs focused on promoting envi- ronmentally friendly and/or active transportation, also led to larger increases in walking. In summary, the SPARC researchers judged that “[t]he most successful interventions could increase walking among targeted populations by up to 30–60 minutes a week on average, at least in the short term.” The successful transportation-focused individualized marketing examples they examined led to walking increases of up to 15–30 minutes a week. By comparison, getting 30 minutes of moder- ate intensity exercise on most days is the current minimum desirable activity recommendation. The researchers warn that, so far, the available intervention research presents stronger evidence of efficacy—the potential illustrated by ability to produce desired results in a controlled setting—than of effectiveness under real world conditions (Ogilvie et al., 2007). Additional evaluations of transportation-focused individualized marketing were provided in the preceding “Individualized Marketing” discussion. A major question is sustainability of intervention results over time. Only five of the 27 studies cov- ered in Table 16-62 examined this issue. One found evidence of intermediate and long-term sus- tainability of walking increases (7.3 miles/week more in a 10-year follow-up). It involved a program for post-menopausal women in Pittsburgh that was relatively intensive, starting with twice-weekly walking training for 8 weeks, and followed with encouragements and advice includ- ing some home visits. Another trial found intermediate term sustainability (12 months) for initially inactive participants but none for participants already walking 15–60 minutes/day at the begin- ning. Typical of the other three studies was an intervention with pedometers that found an increase of 1,500 steps/day at 12 weeks, dropping to under 700 steps/day at 24 weeks—no longer statisti- cally significant (Ogilvie et al., 2007). UNDERLYING TRAVELER RESPONSE FACTORS The underlying factors that motivate or deter selection of non-motorized travel (NMT) as a means of conveyance and/or exercise are examined in this section. As with other modes of travel, trip- specific factors such as purpose, travel time, and cost play an important role. However, perhaps more than with other modes, there are also a host of environmental and user factors affecting the decision to walk or bicycle. These underlying factors do not necessarily work concurrently or with equal weighting in the decisionmaking process. Although the specifics for walking and bicycling do differ, the general categories of underlying traveler response factors are similar for both. Responses to specific facilities, design treatments, programs, and promotion were addressed in the preceding section. Covered here—following a review of behavioral paradigms—are cate- gories of influences on NMT behavior, including natural environment factors such as weather and topography; built environment characteristics including systems and surroundings environ- ments; trip attributes such as trip purpose, time, distance, and cost; and user considerations like gender, age, income, and auto ownership. Also examined are interactions of factors working in combination, effects of predispositions or attitudes, and neighborhood choice (so-called “self- selection”). 16-227

Behavioral Paradigms The active transportation choice responses to various types of factors discussed in this “Underlying Traveler Response Factors” section may be thought of as occurring within one or more unifying behavioral frameworks. NMT activity choices have aspects in common with other travel decisions, but also aspects which are unique to walking and bicycling. Highlighted first is the role of direct achievement of satisfaction (characterized here as “direct-benefit demand”) as an alternative to or in combination with the derived-demand decisionmaking commonly associated with utilitarian trip choices. Next is introduced a postulated mode choice decision paradigm which effectively combines the two. After that, differing choice sensitivities among mode choice, mode of transit- access choice, and route choice are addressed. Finally, the often quite different patterns of travel choicemaking by and for children and adolescents are examined. The focus here is on mode choice (choosing to walk or bicycle versus traveling by motorized means) and route choice (such as deciding between use of a shared-use path and a sidewalk or bike lane). However, it must be remembered that other choices affect outcomes. There is the decision of whether to take a trip or exercise at all (trip generation). There is also the choice of a trip destination, such as between a store close at hand and a shopping center farther away (trip distribution). There are other choices as well, including where to live in the first place (neighborhood choice) and what time of day to travel or engage in active recreation (time of day choice). Derived Versus Direct-Benefit Demand Of special interest in the attempt to understand pedestrian and bicyclist behavior is a postulate— rooted in economic analysis—by researchers who have studied neighborhood walking relationships in Austin, Texas, for over a decade. They suggest that the derived-demand paradigm central to con- ventional travel behavior theory may not apply well to pedestrian behavior, insofar as a significant number of walk trips are taken wholly or in part for their own sake (enjoyment or exercise). The derived demand theory views urban trip-making as travel done primarily as an essential step in accomplishing some other activity of benefit, such as work or shopping (Cao, Handy, and Mokhtarian, 2006), making minimization of travel disutility a driving force in urban travel route and mode choices (Pratt, 1970). In contrast, when walking, jogging, cycling, or rollerblading for recreation or exercise, the actual activity may be the main objective. Conventional economic demand theory should apply in these circumstances where the “good” (the activity) itself is what is desired. Such cases could be characterized as “direct-benefit demand” in contrast to “derived demand.” Though developed in the Austin context on the basis of pedestrian research, the alternative perspec- tive that some walking is a benefit demanded in its own right should apply equally well to bicycling. Indeed, it has been noted that when bicycling for exercise the path followed is itself the destination objective (Broach, Gliebe, and Dill, 2009a), much as when vacationers seek out a scenic highway. Questioning of walkers and cyclists on Indiana trails about reasons for trail visits uncovered substan- tial evidence of primary-purpose exercise or recreation use being combined with secondary-purpose commuting and other utilitarian travel (see “. . . Indiana Trails Study” under “Case Studies”). Similar combinations, without prioritization, were found in surveying users of the Goodwill Bridge in Brisbane, Australia (Abrahams, 2002). When there is a utilitarian purpose combined with exercise or recreation, the non-derived-demand component of the trip choice process probably affects mainly route and mode choice, possibly along with destination choice in the instance of some shopping and restaurant trips. When the motive and purpose for the activity is purely recreation or exercise, con- ventional non-derived demand would logically drive all aspects of the trip-making decision process including whether or not to engage in the activity at all. 16-228

The Austin pedestrian studies are covered more extensively in the preceding “Response by Type of NMT Strategy” section under “Sidewalks and Along-Street Walking”—“Sidewalk Coverage and Traffic Conditions.” Also in the “Response by Type of NMT Strategy” section are several examples of active transportation behavior being fairly obviously influenced, at least in part, by conventional rather than derived demand theory. They are primarily found in the “Pedestrian/Bicycle Systems and Interconnections” subsection under “River Bridges and Other Linkages.” A Combined Mode Choice Decision Paradigm A “Theory of Routine Mode Choice Decisions” has been proposed that encompasses elements of both derived-demand and direct-benefit-demand theory, and also draws from behavior-change- encouragement practice and includes considerations of special importance to walking and bicy- cling. This theory is based on study of the literature and findings of in-depth interviews and travel data collection (from persons making utilitarian-purpose tours with a shopping activity stop), including accompanying research model results. Most of the NMT trips intercepted and studied were walk trips rather than bicycle trips. The theory highlights five steps or components sug- gested as being critical in the choice-making process when selecting among the pedestrian, bicy- cle, transit, and automobile modes (Schneider, 2011): 1. Awareness & Availability. This component gives recognition to the reality that a person must actually have the mode in question available as an option for travel to their intended activity, and also must be aware of it, before the mode can possibly be selected. 2. Basic Safety & Security. One of three situational tradeoffs, this component suggests that a per- son must perceive that a mode offers a basic level of safety and security from traffic crashes and crime before the mode will be selected. 3. Convenience & Cost. Another situational tradeoff, this component acknowledges the impor- tance of travel time, effort, and cost in the choice of a travel mode. 4. Enjoyment. The final situational tradeoff, this component introduces the concept that in NMT choice making, a person will a seek a mode that provides personal, social, and environmental ben- efits, with the personal benefits including physical (exercise), mental, and emotional considerations. 5. Habit. This component adds the concept that if a person regularly chooses a particular mode, that option is likely to be considered in the future as an option. In this five-step theory, socio-economic factors are seen as influencing all of the first four steps or components, explaining differences in how a person responds in the course of each step (Schneider, 2011). This aspect and the “Awareness & Availability” and “Convenience & Cost” steps draw heav- ily from derived-demand urban travel analysis theory and practice. The “Enjoyment” step gives recognition to the direct-benefit aspects of obtaining exercise, fresh air, and recreation in the course of walking and bicycling. A major consideration of particular importance for walking and bicycling is brought in with the “Basic Safety & Security” step, and the “Habit” component draws from behavior- modification theory. Other postulated paradigms offer additional perspectives. For example, one such proposal focused on barriers to walking and cycling. It asserted that a three-tiered decision process is undertaken, involving a process of considering: (1) initial barriers, (2) trip barriers, and (3) destination barriers. In this framework, trip barriers are only a consideration if the initial barriers to NMT are overcome. 16-229

Initial barriers might include safety or weather concerns. Trip barriers would include travel time, distance, and cost. Destination barriers, such as dress code, lack of showers, or lack of bicycle park- ing, may remain after initial and trip barriers have been overcome (Goldsmith, 1992). Some decision paradigms are keyed more to steps found to be useful in encouraging certain behav- ior, such as promoting use of active transportation. The “Other Factors and Factor Combinations” subsection below (see specifically “Attitudes and Modal Biases”) provides a five-stage “model of behavioral change” example adapted from smoking cessation programs. Differential Sensitivities Among Different Choice Categories The “Response by Type of NMT Strategy” section, in the introduction to “Pedestrian/Bicycle Linkages with Transit,” provides definitions of mode share, sub-mode share, and mode of access share. In brief, mode share (or mode choice) refers to the distribution or selection among “prime modes,” such as travel from origin to destination via private automobile, transit, walking, or bicycling. Sub-mode share (or choice) is a term normally applied to transit route choice involving alternative sub-modes such as local bus, express bus, and rail transit. The NMT equivalent would be something like the bicycle routing choice among cycling via a shared-use path, a bike-lane, or a street with no bicycle facility; however, no com- parable term other than simple “route choice” has yet been applied to NMT analysis. Mode of access share (or choice) refers to the distribution or selection among different means of accessing or egressing transit service, such as access via automobile, feeder bus, walking, or bicycling. Each of these choice categories exhibits different sensitivities. A highly sensitive choice relation- ship indicates that a modest change in parameters affecting satisfaction with a particular option or options, such as a change in the conditions outlined in the “Trip Factors” subsection below, will result in a relatively substantial shift in the travel choice involved. A choice relationship with lesser sensitivity indicates that the same change in parameters will produce a smaller shift (Pratt, 1971). Some of these relationships are only beginning to be explored in an NMT context, although mode of access choice modeling—in addition to prime mode choice modeling—has been common for some time. It is a logical assumption that relative sensitivities seen in a motorized transportation context can serve to suggest what order the NMT choice sensitivity hierarchy will take. Relative sensitivities were initially explored in a mode choice versus transit sub-mode choice context. Sub-mode choice was found to be much more sensitive to conditions than prime mode choice. For example, sub-mode choice sensitivity in the north corridor of Chicago and its inner suburbs was found to be over 5 times the sensitivity of the prime mode choice. This meant that introduction of a new rail transit line, for instance, would cause more route shifting among persons already riding transit than between use of an auto and transit riding (Schultz and Pratt, 1971). The structure and “nesting coeffi- cients” of modern nested mode choice models continue to show this greater sensitivity for sub-mode choice relative to prime mode choice, and also tend to indicate that the sensitivity of mode of access choice lies somewhere in between (Abdel-Aty and Abdelwahab, 2001). Imputing the same basic relationships to NMT analysis, sub-mode choice (essentially route choice in the NMT context) is seen to be much more sensitive to changes in facility characteristics than the prime mode choice between auto, transit, walking, and bicycling. Mode of access choice between auto, feeder bus, walk, and bicycle is also more sensitive than prime mode choice, although prob- ably with less of a relative difference than in the case of NMT route choice. The implication for NMT planning and operations is that introduction of a new NMT facility will cause more shifting among routes and facility types than between travel modes. Thus providing a 16-230

new bike lane, for example, will likely attract more bike riders from other routes than it will attract persons to the bicycling mode from other modes such as auto or transit. This circumstance is why it is difficult to assess the impact of, say, new bike lanes on the basis of counts alone, without knowing what proportion of the new riders on a street have simply made route choice shifts as compared to prime mode choice shifts. Similarly, facility improvements and land use improvements—such as hav- ing denser development close to transit stations—will normally lead to much higher shifts toward the walk mode for transit station access than toward the walk-only mode as an alternative to driving. This is why walk mode of access in Transit Oriented Development (TOD) will typically exhibit 70 to 100 percent walk shares, while transit use shares and walk-only shares will be less strongly (although significantly) affected by TOD designs. (Within the “Response by Type of NMT Strategy” section, see “Pedestrian/Bicycle Linkages with Transit”—“Transit-Oriented Development”—“Mode of Access Share Observations” for examples.) Another example of relative sensitivity phenomena is provided by the mode shifts reported in the “Related Information and Impacts” section (see “Travel Behavior Shifts”) upon opening of the Goodwill Bridge for pedestrians and bicyclists across the Brisbane River in Australia. Among four user subgroups defined by NMT mode (walk or bicycle) and trip purpose (commuter or non- commuter), only commuter pedestrians walking the bridge showed more mode changing (52 percent) than shifting of routes without a mode change.51 Among the other three groups, only 19 percent to 34 percent changed modes as compared to shifting routes while continuing to walk or bike as before (Abrahams, 2002). The Travel Choice Making of and for Children It has been headlined that the travel behavior of children reflects “a world of difference” (Zwerts and Wets, 2006). Instead of the adult pattern of individual choice moderated by travel options availability and financial, familial, and some institutional responsibilities, the behavioral paradigm governing childhood travel choices is one of parental decisionmaking imposed “from above,” with gradual diminution of parental control as the child grows toward adulthood. The literature review for a 2010 appraisal of Safe Routes to School (SRTS) programs by the Washington State Department of Transportation (WSDOT) presents a conceptual framework for children’s travel behavior originally developed by McMillan in the context of the elementary- school commute. The concept draws from both the transportation field, specifically the activity- based paradigm of travel behavior, and the public health field’s social ecological model. The conceptual framework places parental decisionmaking at the center, posited to take place informed by mediating factors and moderating factors. Mediating factors are parental opinions developed in consideration of urban form characteristics, related neighborhood parameters, and transporta- tion options. Moderating factors are exogenous to the trip to school and the immediate environ- ment. The moderating factors intensify or diminish the impact on parental decisionmaking of the mediating factors. The conceptual framework is rounded out by presumption that the parental decisionmaking determines the child’s mode of travel to and from school, which if active trans- portation to school (ATS) is selected, leads to health, environmental, and congestion relief out- comes through the mechanisms of physical activity, air pollution reduction, and traffic volume reduction (Moudon, Stewart, and Lin, 2010). 16-231 51 In this particular instance, many of the mode changes identified were not actually shifts in the primary mode of travel, but rather changes in the mode of egress/access to/from downtown Brisbane in connection with use of a motorized mode such as auto or train (Abrahams, 2002).

An example is provided by a trip from home to school that requires passing by vacant and dilapi- dated buildings. This urban form factor may cause the parent to believe that the neighborhood crime rate is high, a mediating factor. (Other examples of mediating factors might include perceptions of heavy or light traffic, condition or lack of sidewalks, or presence of crossing guards.) A mediating factor of perceived high crime levels may result in parental judgment that having a child walk to school would be unsafe. If the child is young, age being an example of a moderating factor, this fear may be intensified. (Other examples of moderating factors include cultural norms and attitudes.) In the example at hand, the mediating and moderating factors may lead to the child being driven to school, lessening the child’s physical activity and increasing risk of obesity, a health outcome. Some researchers have concluded that children’s travel behavior is not so much dictated as negoti- ated between parent and child. Clearly parental control will be dominant in the case of younger chil- dren, but will tend to become less so as the child matures (Moudon, Stewart, and Lin, 2010). A child-travel-to-school mode-choice modeling effort has been constructed in this context, utilizing the 2001 NHTS as the database, and covering the auto passenger, school-bus/transit, and walk modes. The structure of the multinomial logit model assumes that parents, together with their children, choose the child’s travel mode as a family unit to maximize household utility (McDonald, 2008). A broad definition of utility is implied here—a definition that could encompass such factors as child safety. Variables tested that clearly would fall in the categorization of moderating variables include age, gender, and number of siblings. The modeled auto share elasticity for auto travel time was small (−0.08), and negligible for walk share (+0.01). On the other hand, the modeled walk share elasticity for walk time, though still in the inelas- tic range, was quite pronounced (−0.75). This elasticity suggests that a 10 percent longer walk time to school is associated with a 7.5 percent lower walk mode share. The walk share cross elasticity for auto travel time was small by comparison (+0.10). Population density exhibited a very small nega- tive auto share elasticity (−0.02) and a modest walk elasticity (+0.12). Among child characteristics variables, age was estimated to have an auto mode elasticity of −0.58 and a walk mode elasticity of +0.82. Each additional year in age was associated with a 1.4 percent lower auto mode share and an 0.4 percent higher walk share. Gender differences were insignificant in this formulation, which could not include bicycling for lack of sufficient data. Number of siblings had an auto share elasticity of −0.10 and a walk share elasticity of +0.15, perhaps reflecting both difficulties of chauffeuring more children to school and opportunities for family members to walk together. Race was an insignificant factor when other parameters such as walk time to school were accounted for. Higher income had a positive effect on being driven to school and a negative effect on walking to school, with elasticities of +0.21 and −0.26, respectively. Oddly, the vehicles per driver ratio exhib- ited a negative auto mode elasticity (−0.02). The corresponding walk elasticity was logical and some- what more substantial (−0.15). The pseudo-R2 for the model was 0.27 (McDonald, 2008). The estimate of a higher walk share for each additional year of age presumably relates, at least indi- rectly, to growing rates of parental permission to walk or bike as children mature. SRTS parent sur- vey data collected from 2007 through 2009 from over 1,200 schools in 47 states bears on this relationship. The cumulative percentage of parents reporting they would allow their child to walk or bike to school “without an adult” was, in order by grade, 1 (kindergarten), 4 (1st), 9 (2nd), 22 (3rd), 39 (4th), 58 (5th), 78 (6th), 90 (7th), and 100 percent (8th) (Marchetti, 2010). Another set of figures relevant to parent versus child decisionmaking on school travel mode comes from parent and student surveys at elementary and middle schools in Hillsborough County, Florida (the greater Tampa area). A total of 489 classroom tally sheets and 3,213 parent survey forms were returned, representing response rates of 84 and 26 percent, respectively. The tallies cov- 16-232

ered five consecutive days of travel to and from kindergarten through 8th grade. It was found that 79 percent of students had asked permission to walk or bike to school, while 33 percent of parents reported allowing or planning to allow their child to do so at an elementary or middle school grade level they deemed appropriate. (It will be noted that this result does not fit well with the SRTS sur- vey data reported above, which adds to 100 percent permission to walk or bike without an adult by 8th grade, likely because of differences in the survey question or its administration.) On the average Hillsborough County trip tally day, 8.3 percent of boys and 13.5 percent of girls actually walked, and 4.6 percent of boys bicycled (no girls did so), for an average walking and bicy- cling to school rate of 13.2 percent (Zhou et al., 2009). This compares with U.S. SRTS student and parent survey results, for 2007–2009, that averaged about 15 to 16 percent walking or biking (Marchetti, 2010). The Hillsborough County school travel data allows examination of differences in mode coming and going. Of 419 children reporting walking to and/or from school, 16.2 percent walked only to school, 7.4 percent walked only from school, and 76.4 percent walked in both directions. All 73 chil- dren who bicycled did so in both ways (Zhou et al., 2009). It may be assumed that mode choice dif- ferences by direction are commonplace in many situations, and for adults as well as children, but simplified reportings of mode share (as contrasted to reportings based on travel diary surveys, for example) often fail to take directional differentials into account. The unique balance of concerns affecting child travel choices lead to a somewhat different set of factors being taken into account than seen with adult travel choices, and with a different ordering of priorities as well. Results of investigating child travel factors in the context of the trip to and from school are presented below in the “Trip Factors” subsection (see “Schoolchild Trip Factors”). Environmental Factors Two broad categories of environmental factors play a role in influencing the amount of walking and bicycling. In this chapter, these categories are organized for convenience of discussion rather than in any hierarchy of importance. First, is the natural environment, including weather, season, climate, topography, and daylight and darkness. This natural environment can to some extent be mitigated through facilities. For example, walkways can be heated and/or covered, ravines bridged, and path- ways lighted. Many aspects remain, however, beyond practical means of human control. Second, is the built environment, as expressed in land use configuration and transportation infrastruc- ture. The density of development, mix of uses, orientation of streets, and presence and nature of facil- ities for non-motorized travel are key built environment attributes. Facility- and neighborhood-specific built environment factors and their effects were covered earlier in the “Response by Type of NMT Strategy” section. Within that section, see especially “Pedestrian/Bicycle Friendly Neighborhoods.” In this “Environmental Factors” subsection a more over-arching view of the built environment is taken. This overview is expressed under “Systems Environment” in terms of accessibility and con- nectivity, and under “Surroundings Environment” in terms of facility compatibility measures and ambiance. Natural environment factors are covered first. Natural Environment Precipitation and temperature are, as would be expected, important factors in the day-to-day choice of biking or walking as travel modes. Seasonal effects are also observed. There may not be 16-233

that much year-round climate impact, however, on overall annual non-motorized transportation (NMT) usage rates. Rain or snow leads to fewer walking and bicycling trips during the weather event except for pedes- trians having covered walkway systems available. Very hot or cold weather also reduces the attrac- tiveness of NMT activity open to the elements. Nevertheless, although weather is regularly mentioned in surveys as a consideration in choosing to bicycle or walk, it is best viewed as a day- to-day factor. Weather appears to affect the incidence of walkable or bikeable days rather than the overall choice to walk or bike in general (Goldsmith, 1992, Heglund, 1980). This finding appears to hold true whether one looks at data for the United States or elsewhere. For ease of presentation, weather, season, and climate effects on walking are treated separately from effects on bicycling. Those discussions are followed by all-NMT-modes data on seasonal variations. Effects of topography and daylight and darkness conclude the natural environment discussion. Weather, Season, and Climate Effects on Walking. Studies in the Province of Ontario, Canada, and New York City have each illustrated that temperature extremes and precipitation are deter- rents in the day-to-day decision to walk. Precipitation proved a greater impediment than temper- ature in the Ontario research, which was based on a stated preference survey. Extremely cold temperatures (less than 20° C (−4° F)) and hot temperatures (greater than 30° C (86° F)) each were estimated to deter more than one-third of Ontario respondents from walking. It was further esti- mated that about 70 percent would not walk if there was heavy snow. In direct observations in midtown Manhattan, heavy rain reduced the number of sidewalk pedestri- ans on 42nd Street by 24 to 55 percent, depending on the intensity of the rain. Researchers found that most affected pedestrians either used the subway or changed or cancelled their itineraries. Finally, in Seattle, a locale known for its moderate temperatures and persistent but usually gentle precipitation, a survey found that 9 percent of respondents identified weather as a reason for not walking more often. The researchers cautioned that this figure may be inconsistent with other research, because the question focused on reasons for not walking more often, rather than about the specific effect of weather on walking overall (Goldsmith, 1992, University of North Carolina, 1994). NCHRP Project 08-78, “Estimating Bicycling and Walking for Planning and Project Development,” has reviewed the state of environmental-factors research as part of its work effort. It located direct weather-effect observations for Montpelier, Vermont, published by Aultman-Hall et al. in 2009. Pedestrian flows were obtained for an entire year with infrared sensor monitoring of a single downtown intersection. Temperature, humidity, precipitation, and wind speed data from a weather station 3 miles away were linked to the pedestrian count data hour by hour. Winter pre- cipitation was found to reduce hourly volumes by 16 percent. Precipitation in the rest of the year was associated with approximately a 13 percent reduction. Weather effects such as cold and pre- cipitation together consistently reduced overall levels of walking, but by an amount less than 20 percent. The estimated maximum combined effect of weather variables, barring extreme events, was a 30 percent reduction in pedestrian flows. The impact of winter overall (January through April) was significant with a 30 percent negative effect relative to the rest of the year, but it was not possible to know what proportion of that effect related to visitor and tourist traffic variations (Kuzmyak et al., 2011). Spring/summer infrared counter monitoring, for 3-1/3 months, of sidewalks adjacent to 11 inter- sections in the San Francisco East Bay Area’s Alameda County provides observations from a more temperate climate. Here, instead of a 13 percent reduction, measurable rainfall (≤0.01 inches) was associated with a 7.1 percent pedestrian traffic reduction relative to average flows. (There were 16-234

only 8 hours of measurable rainfall in the study period.) Cloudy conditions, defined as less than 60 percent of the solar radiation average for the hour and place, were associated with 5.3 percent less walking activity. Temperatures at or below 50° F saw volumes lower by 2.3 percent while tem- peratures at or over 80° F, between noon and 6:00 PM, were accompanied by a 3.6 percent volume reduction. On the other hand, temperatures at or over 80° F during other hours were associated with an 0.4 percent pedestrian volume increase (Schneider, Arnold, and Ragland, 2009). Though the researchers did not report on the possibility, one might speculate that hotter weather induced some shifting of walking out of the heat of the afternoon into other hours. The argument has been made that the perception of adverse weather may be a stronger deterrent to walking than the weather itself. This possibility has not been much researched, and neither has the effect of humidity (Kuzmyak et al., 2011). The low levels of walking in the southeastern and south-central United States have been well documented, as well as the parallels with higher tem- peratures (Alliance for Biking & Walking, 2010), but descriptive or bivariate analysis alone cannot separate heat and humidity effects from confounding factors such as prevalence of sprawling cities in much of the south. Most regular pedestrians readily adapt to normal weather. An umbrella or rain coat make walk- ing in the rain less unpleasant, and clothing can be adjusted to the temperature if dress codes allow or do not pertain. Nonetheless, weather extremes are significant deterrents, and experience from cities with skywalks clearly suggests a preference during inclement or uncomfortable weather for climate-controlled pedestrian environments. Data on skywalk versus sidewalk route choice in response to weather are found under “Response by Type of NMT Strategy”—“Pedestrian Zones, Malls, and Skywalks”—“Pedestrian Skywalks.” Weather, Season, and Climate Effects on Bicycling. Weather may have a greater impact on bicy- clists than on pedestrians because it is not as simple to mitigate, especially in the case of precipita- tion. In addition to comfort considerations, precipitation introduces concrete safety concerns. Spray from passing motorists and the bicycle itself can prove messy and potentially hazardous. Traction and visibility can each be adversely impacted during rain or snow. Several researchers have observed that precipitation is probably the most important climactic factor for bicyclists. Respondents to surveys conducted in several cities (including Boston, Gainesville, Portland, and Vancouver) report adverse weather to be influential in the decision not to cycle (mentioned by 86, 90, 52, and 51 percent, respectively, with multiple responses allowed) (Goldsmith, 1992, Pinsof, 1982). There are significant numbers of all-weather cyclists, however, who dress warmly in the winter, coolly in the summer, and have appropriate equipment for precipitation days. The existence of seasonal effects of heat, cold, and rain on bicycling is supported by empirical observations/counts made around North America and internationally. Chicago studies of bicycle paths found lower usage from December to February as compared to April through October, when at least half the days have low temperatures no colder than 40° F (Welzenbach, 1996, Pinsof, 1982). In Santa Barbara, 49 percent of users of a bicycle-transit trailer service said they would not use their bike in rainy weather, and the actual usage did decrease during the December-to-March rainy sea- son (Newman and Bebendorf, 1983). In the Netherlands, known for extensive bicycling, there is nonetheless a reduction in winter. Weather there has been observed to have the strongest impact on recreational cyclists. On utilitarian-trip routings and in built-up areas the fluctuations per month are generally a maximum of 30 to 40 percent relative to the yearly average (C.R.O.W., 1993). A Washington State study performed extensive field data collection and analysis of effects on bicycling of both weather changes and general seasonal trends. Weekday bicycle counts were gathered at five locations over a variety of conditions. In all locations observed, rider volumes 16-235

were lower on cloudy days than on sunny days and lowest by far on rainy days. In the AM peak period, volumes of all cyclists recorded on rainy days were 45 to 60 percent lower than on sunny days, and volumes recorded on cloudy days were 10 to 20 percent lower than on sunny days. In the PM peak period, rainy day cyclist volumes were 55 to 68 percent lower than on sunny days, while cloudy day volumes were 25 to 41 percent lower. As in the Netherlands, the weather seemed to have the strongest impact on recreational cyclists. The researchers suggest that utili- tarian riders are about one-half to one-third as sensitive to adverse weather conditions as recre- ational riders (Niemeier, Rutherford, and Ishimaru, 1995a). It has further been noted that the heightened effect in the PM peak hours likely reflects the higher proportion of non-commuter cyclists in those hours as compared to the morning peak. Studies in Austria and Australia con- firm the higher sensitivity of recreational riders to weather extremes than bicycle commuters (Kuzmyak et al., 2011). The Washington State researchers used selected counts and National Weather Service data to model seasonal impact on volumes, concluding that winter bicycle counts could be counted on to average about 50 percent of summer counts (Niemeier, Rutherford, and Ishimaru, 1995a). A num- ber of additional studies find winter months to exhibit about one-half the cycling activity of sum- mer months. This effect may, of course, relate in part to factors other than weather, such as vacation schedules. Examples of finding approximately a 2:1 cycling activity relationship between summer and winter include the National Omnibus Household Survey for November 2001 to October 2002, (Bureau of Transportation Statistics, 2002) and the 2001 NHTS. Seasonal adult bicycling daily rates established in the 2001 NHTS were 1.1 percent of the U.S. population in summer, 0.88 percent in spring/fall, and 0.56 percent in winter, relative to the annual average of 0.9 percent (Krizek et al., 2007). A Boston study found that cycling activity decreased when temperatures fell below 40° F, and that only 10 percent of the student population continued to cycle for a full 10–12 months of the year. That compared with 22 percent reported bicycling for 6 to 9 months (University of North Carolina, 1994). Where the weather is more severe, wintertime differences may be greater. In a Toronto survey, 88 per- cent of utilitarian bicyclists reported cycling in the spring, 98 percent in the summer, 89 percent in the fall, but only 23 percent in the winter (City of Toronto, 2001). The Nonmotorized Transportation Pilot Program Evaluation Study avoided summer vacation effects by surveying from September 2006 into January 2007. The percentage of respondents reporting having bicycled on their survey day declined from 3.2 percent at the beginning of the survey down to 1.1 percent overall in the five urban areas sur- veyed, including a drop from 4.4 percent to 0.9 percent in Minneapolis (Krizek et al., 2007). In contrast to seasonal-effect findings such as these, a 1990 Boulder (Colorado) Diary Survey revealed that “season of the year had little effect on mode choice” and that rainy weather tended to reduce bicycle and pedestrian travel by only 2 to 3 percent (University of North Carolina, 1994). However, a newer study of bicycle volumes on four trails in Boulder concluded that summer counts were 2.3 to 4 times higher than in winter across the four trails. Modeling of bicycle counts and weather data showed bicycling activity to increase with temperature up through 90° F. As daily highs exceeded 90° F, a decline set in. Modeled effects of snow and rain indicated bicycling reductions in response, but the results were not statistically significant. Fairly similar results have been reported by a study in the United Kingdom, where cycle count and weather modeling detected a 3 percent increase in cycling volumes per 1° C (1.8° F) increase, with no maximum estab- lished. The U.K. research suggested that whether it rained or not had more influence than the amount of rain (Kuzmyak et al., 2011). Daily and seasonal effects notwithstanding, in descriptive and bivariate analyses of cycling levels and annual climate data for several cities (20 cities in the more recent such analysis), neither tem- 16-236

perature nor precipitation showed any obvious correlation with a city’s average overall bicycle-to- work mode share. These studies worked with annual measures of daily temperatures and the num- ber of days per year with measurable precipitation. Essentially the same conclusion was reached in an analysis of bicycle commute shares in each of the 50 states, done using average winter and average summer temperatures as weather variables. In these studies the mode share data was derived in terms of usual travel mode and did not capture short-term, day-to-day impacts (Goldsmith, 1992, Alliance for Biking & Walking, 2010). Three additional nationwide aggregate analyses, designed to investigate the impact of bike lane or lane-and-path system extent on commuter bicycling rates, took the further step of utilizing multi- variate analysis. They again used U.S. Census or American Community Survey (ACS) journey-to- work data, and have been detailed in the “Response by Type of NMT Strategy” section (see “Bicycle Lane System Coverage”—“Bicycle Lanes and Routes”). The first utilized a cross-sectional model describing likelihood of bicycling for the work commute under average conditions in 16 cities. Mean daily temperature had insufficient significance for retention in the final model. Rain days during the year proved, on the other hand, to be a significant and negative explanatory variable. Nevertheless, each rain day had only fractional importance in comparison to either having one more mile of bicycle facilities per 100,000 residents or one percentage point more of college stu- dents among the population (Nelson and Allen, 1997). The second such study covered 42 large cities. It likewise found rain days to be the one statistically significant weather variable, when esti- mating overall commuter cycling share, though it was not a major contributor to explanatory value of the research models (Dill and Carr, 2003). The third and largest such study also modeled aggregate city-level commuter cycling, facility- extent, and weather data, this one on the basis of 90 of the 100 largest U.S. cities. Number of days below 32° F and annual precipitation levels proved to have small and insignificant associations with bicycle commute share in preliminary bivariate statistical tests and were omitted from the final models. Annual number of days above 90° F did prove statistically significant in bivariate analysis and was included in the multivariate models. In these final formulations the measure had a negative relationship to cycling levels but failed to show statistical significance (Buehler and Pucher, 2011). Given these various mixed results showing weak or no significant weather effects on annual (or equivalent) levels of commuter bicycling, it is likely that while temperature and precipitation clearly affect day-to-day cycling decisions, they do not overly constrain the annual market for bicy- cling to work in any particular area. There is insufficient information to draw parallel conclusions for either utilitarian or recreational non-work-purpose cycling or for walking. Combined Walk and Bike Seasonal Effects. Seasonal usage rates have been obtained for a num- ber of off-road shared use paths, covering all facility-user traffic in combination. Though the paths do serve some utilitarian NMT trips, almost all of those included are predominantly taken advan- tage of for recreation and exercise. Thus it would be inappropriate to assume, for example, that the seasonality-of-use extremes tabulated would apply to the likes of a downtown sidewalk or a tran- sit station access route. Table 16-63 gives percentages of observed or estimated path traffic in each season. The paths or path groupings covered are listed in increasing proportion of wintertime usage. Month by month detail for the Indianapolis and San Diego observations is provided within the “Related Information and Impacts” section (see Table 16-103 in “Facility Usage and User Characteristics”—“Off-Road Shared Use Paths”—“Path Volume and Usage Patterns”). 16-237

Despite derivation of seasonal percentage distributions utilizing disparate methodological approaches, ranging from counts to self-reported survey responses to questions in a variety of formats, the eight paths and path groupings in Table 16-63 form a logical progression when listed from lowest to highest relative wintertime usage. North Central and Northeast U.S. paths have the least winter- time usage, with the tightest mid-year concentration of use in the North Central states. The Mid- Atlantic W&OD Trail is clearly intermediate despite the questionable four-season total of 106 percent, and the least season-by-season variation is found in mild California climates. Topography. Pedestrians and bicyclists both exhibit a resistance to change in grade. Climbing hills is more strenuous than traversing flat terrain and requires the individual to be more physically fit. Moreover, the exertion associated with difficult terrain may cause sweating to be more of a con- cern and reduce the number of willing participants where condition upon arrival is a concern. A study of bicycle commuters in England revealed a strong negative correlation between the hilli- ness of an area and the level of bicycle commuting. The resistance of pedestrians to climbing is among the factors that help explain reluctance to use many of the overpasses or underpasses that have been provided for crossing roadways (Goldsmith, 1992, AASHTO, 2001). Quantitative confirmation of these observations on effects of topology is provided by the City and County of San Francisco travel demand modeling effort introduced earlier in the “Point-of- Destination Facilities” subsection (see “Other Destination Amenities”) and more fully covered in Chapter 15, “Land Use and Site Design.” Topology had the highest impact on work and work- based tours of five trip destination pedestrian environment characteristics scaled for model use by a Delphi panel. (Although presence of steep grades was the major component of the topology mea- sure, other natural barriers were also included.) It was second to vitality in effect on “other” pur- pose travel and individual work-related trips within tours. Milder topology (flatter grades and fewer natural barriers) was found to be an indicator of higher mode shares for walking, walk- access transit use, and bicycling. The strongest effect was on choice of the bicycle mode for school trips (Cambridge Systematics et al., 2002). 16-238 Table 16-63 Path Use Seasonal Distribution Percentages, Walking and Cycling Combined Trail(s) / Location Type Winter Spring Summer Fall 30 Indianapolis locations, urban/suburban Various 7.6% 25.4% 43.6% 23.4% 3 Hennepin Co. (Minneapolis) trails, urb./sub. Rail trails 9.0 24.7 39.3 27.0 4 Rhode Island trails, suburban/towns/rural Rail trails 9.2 29.4 31.5 29.8 Monon Trail, Indianapolis, urban/suburban Rail trail 10.8 29.0 38.2 22.0 W&OD Trail, Northern Virginia, sub./exurban Rail trail 12 28 39 27 Gilman Bike Path, San Diego, suburban I-5 corridor 17.4 23.2 36.2 23.2 Iron Horse Trail, S. F. East Bay, exurban Rail trail 18.2 28.0 28.0 25.9 Strand Bike Path, San Diego, urban/scenic Beachfront 18.6 27.0 31.6 22.8 Notes: Indianapolis and San Diego percentages based on path counting, other values based on survey responses to questions on path use by season. W&OD Trail percentages as reported. Count-based percentages treat December through February as winter, March through May as spring, and so on. Other values based on survey respondent perceptions of season. Sources: Jones (2009), Hennepin County (2005), Gonzales et al. (2004), Bowker et al. (2004), East Bay Regional Park District (1998), with seasonal percentages calculated (except W&OD Trail) by the Handbook authors.

Comparative sensitivities for slope, published in 2003 by Cervero and Duncan, were reviewed in NCHRP Project 08-78. Modeling of walk and bicycle mode choice using the year 2000 [San Francisco] Bay Area Travel Survey (BATS 2000) found both NMT modes to be negatively affected by steeper gradients but with almost twice the adverse effect on bicycling as on walking. Weather was not included in the bicycle-share model, but slope in the walk-share model—measured as rise divided by distance—was a shade more important (about 14 percent more) than rain on the travel day (Kuzmyak et al., 2011). Explanatory modeling based on GPS-and-network-analysis of non-recreational bicycle route choice in Portland, Oregon, provides estimated elasticities that quantify the negative route choice effects of upgrades. Average upgrade slope (feet or meters of gain in elevation per 100 feet/meters), ignoring downgrades, was used as the analytical measure. An “elastic” response was estimated, with percent- age decrease in cyclists choosing a route moderately exceeding the percentage increase in upgrade incurred. The specific elasticity estimated was about −1.3 (Broach, Gliebe, and Dill, 2009a). It was fur- ther estimated that for the typical utilitarian bicycle trip, a cyclist would be willing to go 27 percent more distance to avoid each 1 percent additional average upslope (Broach, Gliebe, and Dill, 2009b). This effect was found to be stronger for women than for men, and more pronounced for infrequent cyclists than for frequent cyclists (Dill and Gliebe, 2008). These San Francisco and Portland research findings concerning topography pertain primarily or exclusively to utilitarian NMT trips. Cross-sectional modeling drawing on Seattle GIS data, down to the parcel level, and Walkable and Bikeable Communities project survey results, confirms the negative association between utilitarian walking and even moderate slopes. Recreational walking, however, had a positive association with slope. Both associations were statistically significant in most model formulations, but the strongest significance was found for the positive relationship between recreational walking and slope. Recreational walking was about 15 percent more likely to occur in the presence of grades averaging 8 percent (8 feet or meters elevation change per 100) within a 1 km. buffer. The researchers speculate, “Recreational walkers may like the views and greater exercise opportunities that come with a hilly landform.” Effects on cycling were not exam- ined (Lee and Moudon, 2006a). Daylight and Darkness. A limited amount of research has been done on the discrete impact of day- light versus darkness on the choice of pedestrian or bicycle trip making. It is clear that visibility would be a concern of any pedestrian or bicyclist traveling after dark. In stated preference surveys, “adequate lighting” is often given as a consideration for such travelers. Visibility is important from the standpoint of being able to see where one is going, but also so that one may be seen by motorists. Perceived safety from crime is also related to daylight and good lighting as compared to darkness. Safety concerns are addressed further in the “User Factors” subsection, and in the “Safety Information and Comparisons” subsection under “Related Information and Impacts.” A Florida survey of NMT found that barely over two percent of trips were made at night. The remain- der were nearly evenly split among the morning (29 percent), afternoon (30 percent), and evening (39 percent). Almost all of these trips were made under daylight conditions (95 percent). The remain- der were made in the dark (2 percent) or at dawn or dusk (3 percent) (NuStats International, 1998). The only research reviewed in NCHRP Project 08-78 that attempted to explicitly control for dark- ness was the previously mentioned walk and bike share model derivation done on the basis of San Francisco region BATS 2000 travel data. The mode choice model coefficients in that study indicated that cyclists are around 5 times as sensitive to traveling in the dark as pedestrians. The walk model darkness coefficient suggested “a minor but significant [. . .] deterrent effect” for walking in the dark (Kuzmyak et al., 2011). The deterrent effect of darkness was about 1/12 of the effect of having 16-239

to walk an extra mile and between 1/4 and 1/5 as disadvantageous as precipitation. Choice of walking was 1/25 as sensitive to darkness as to slope expressed as rise/run for the entire trip. For bicyclists, darkness held roughly 1/10 the importance of slope, but was more than twice the deter- rence of having to cycle an extra mile. Both models covered only non-work trips, under 5 miles in length, made for purposes/durations not likely to entail carrying large packages (Cervero and Duncan, 2003). Other indications have been seen of the effects of darkness. When three poorly lit streets and a foot- path in London received street lighting improvements, pedestrian volumes (stratified by gender and presumably after dark) increased by 34 to 101 percent; 51 percent on average (Cao, Mokhtarian, and Handy, 2007, Heath et al., 2006). The strong afternoon/evening walking and bicycling peak seen on six Indiana trails in year 2000 September counts was observed to move forward in time of day, and became compressed, with the onset of shorter days in October (Indiana University, 2001). These two studies are covered, respectively, in the “Response by Type of NMT Strategy” section (see “Sidewalks and Along-Street Walking”—“Individual Sidewalk Provision Examples” including Table 16-1) and in the “Case Studies” section under “Six Urban, Suburban, and Semi-Rural Trails— Indiana Trails Study” (see “Results” discussion including Table 16-136). Systems Environment The built environment has a role in influencing the prevalence of walking and bicycling that is every bit as important as the natural environment, if not more so. It has been rightly proposed that improved evaluation, planning, and design of bicycle facilities requires recognition that two com- ponents are necessary for such analyses, one for evaluating the overall system, and one for evalu- ating the links that make up the bicycle system’s network (McCahill and Garrick, 2008). The same certainly applies with respect to pedestrian facilities, albeit at an even finer geographic scale. The component addressing the system and network aspects of the built environment is covered here under the label “Systems Environment,” and the component focusing on link evaluation is dealt with subsequently under the label “Surroundings Environment.” With respect to “Systems Environment,” accessibility is judged the most fundamental influence and discussed first, followed by connectivity, which together with land use and associated activity is what produces accessibility. Accessibility. Accessibility as an analytical concept was originally developed within the trans- portation and land use planning community as a tool for forecasting land development, and valu- ing land, on the basis of existing and projected transportation facilities paired with defined land use patterns. It was later found useful as a mode choice forecasting parameter, wherein good acces- sibility to jobs, goods, and services via a particular travel mode indicates likely higher use of that mode than one with poorer accessibility. NMT accessibility measures are exceptionally useful for describing built environments amenable to the meeting of many daily needs by walking and cycling activity. They serve both as tools for pedestrian/bicycle-friendly development planning guidance and as walking and cycling activity estimation variables. In his landmark exploratory paper on accessibility, Hansen defined it as “the potential of opportu- nities for interaction.” Accessibility is thus an opportunity measure. It measures “the intensity of the possibility of interaction rather than just [. . .] ease of interaction.” (Hansen, 1959). It is more than simply a measure of mobility. Accessibility has also been defined, relying on more concrete terms, as “the ability to reach desired goods, services, activities, and destinations (together called opportunities).” Jobs are explicitly con- 16-240

sidered as opportunities. Such perspectives are as viewed from residences. When accessibility is viewed from the perspective of employers, merchants, or institutions, it becomes ability to be read- ily reached from the urban population. Four component factors make up personal and public accessibility (Victoria Transport Policy Institute, 2010): • Personal Mobility, the ability to move about without incurring excessive travel time and cost. NMT mobility is provided by walking and bicycling, while motorized mobility is obtained through use of private vehicles, ridesharing, taxis, and public transportation. • Mobility Substitutes, such as telecommunication allowing transfer of information or web- based sale of goods, and delivery services providing goods transfer that would otherwise require personal travel. • Transportation System Connectivity, reflecting both the density of connection between the transportation network’s links and directness of the individual links themselves, together arranged to provide direct and fluid passage through the overall network. • Land Use, specifically the geographic arrangement of housing, activities, and destinations in general. If the geographic arrangement is compact and cohesive, then accessibility—and espe- cially accessibility via walking and bicycling—will tend to be enhanced. If the arrangement is dispersed, as in urban sprawl, more mobility—motorized vehicle mobility in particular—will be required to maintain even a basic level of accessibility. Accessibility can be complex to measure, particularly if all possible travel modes are covered and all mobility impedance factors are considered, including time, money, convenience, and risk. Ideally all impedances to mobility would be addressed through use of generalized cost measures (Victoria Transport Policy Institute, 2010). Gravity model formulations developed for trip distri- bution estimation are often drawn upon for sophisticated accessibility calculations. Quite simple accessibility measures are nevertheless very effective in analyzing NMT accessibil- ity. Since variations in everyday walking and bicycling speeds among facilities are much less than encountered with motorized traffic, and a significant factor in walking or cycling is simply the physical effort of locomotion, plain along-the-road (or path) distance measures can form the basis for robust accessibility calculations and comparisons. Number of activities within 1/4 mile, 1 mile, or 5 miles can be a very useful measure, with the distance selection being a function of the analy- sis objectives. “Activities” may be expressed in terms of jobs, retail jobs, or whatever type of des- tinations are of interest. Conversely, accessibility to employment, schools, institutions, or transit stations can be measured as number of households within the selected fraction or number of miles. An example of applying this type of basic NMT accessibility measure is afforded by analyses made by the San Francisco area’s Metropolitan Transportation Commission (MTC). MTC analyzed the population of, and travel generated in, all areas within 1/2 mile of commuter rail and Bay Area Rapid Transit (BART) stations, commuter ferry terminals, light rail transit (LRT) stops, and street- car and cable car lines. (San Francisco Bay Area land use and transportation system layout is such that a large majority of the population relatively close to urban office and traditional urban on- street commercial areas was likely thereby included.) It was found that in these high accessibility areas, residents made 1/2 of their short trips (trips of 1 mile or less) by walking, compared to 1/4 for residents of other areas. It was also determined that for trips of any length, residents within 1/2 mile of rail/ferry stops were twice as likely to choose the walk mode of travel, three times as likely to choose the bicycle mode, and four times as likely to choose the transit mode (with its high likeli- hood of walk access and egress). Persons with both their residence and workplace within these 16-241

highly transit-accessible areas made 42 percent of their commute trips by transit, as compared to 4 percent for those with neither home nor workplace in such areas (Gossen, 2006). Recent research has also suggested that accessibility to typically interchangeable routine daily des- tinations such as grocery stores, banks, or libraries can be equally well or better analyzed in terms of distance to the nearest such facility instead of number of activities within a given distance (Lee and Moudon, 2006b, Moudon et al., 2005, Moudon et al., 2007). This type of accessibility measure- ment lends itself to a “directness” approach for computing an Accessibility Index. This is done by dividing direct (“airline”) travel distance into the actual minimum travel distance to destinations. The lower the value, the better. The ideal is an index of 1.0, indicating that a truly direct walk or bike ride is possible. A value of 1.5 has been proffered as an acceptable average (Victoria Transport Policy Institute, 2011b, Litman, 2011b). Directness research applications/outcomes are described within the “Pedestrian/Bicycle Friendly Neighborhoods” subsection of the “Response by Type of NMT Strategy” section of this chapter. The “Pedestrian/Bicycle Friendly Neighborhoods” subsection focuses heavily on the effects on walking and bicycling activity of enhancing land development density, diversity, and design, all contributors to enhanced accessibility at the fine-grained scale important to NMT trip attractiveness and practicality. Most of the various topics covered in the “Response by Type of NMT Strategy” section address trav- eler response to some type of mobility enhancement involving new, improved, expanded, or better deployed NMT facilities and systems. Because many of these strategies are focused more on NMT mobility than accessibility, it is important to keep in mind that success rates may be lower in built envi- ronments with lesser underlying NMT accessibility. Accessibility environments for best transporta- tion results are provided by compact, mixed-use development and fine-grained, high-connectivity transportation networks such as traditional grid street systems. In low-density, low-accessibility areas, residences and activity destinations are likely to be too far apart for most trip makers to contemplate utilitarian walking or bicycling even with improved facilities (Schneider, 2010). Connectivity. Connectivity is a primary contributor to favorable walking and bicycling physical environments, largely because of its role as a key element of accessibility, but also because “direct- ness” per se is favored by walkers and cyclists making utilitarian trips. Three measures taken together describe connectivity that is useful to NMT tripmakers: 1. Density of connections in the road and NMT facility network, providing more travel options and network resiliency (Victoria Transport Policy Institute, 2010 and 2011b). 2. Directness of links (Victoria Transport Policy Institute, 2011b), and interconnection into direct routes/paths. In studies of utilitarian trip making, turns—especially left turns—have been shown to render bicycle routings less inviting (Broach, Gliebe, and Dill, 2009a and b), and indi- cations have been found that directness offers walk mode attractiveness above and beyond the benefit of walking distance saved (Moudon et al., 2007). 3. Alignment of interconnected links, in logical and direct routings, with travel needs (Alta Planning + Design, 2009a). It is in this manner that connectivity directly contributes to accessibility. Various connectivity measures have been offered as a basis for Connectivity Indices. Examples include (Victoria Transport Policy Institute, 2011b): • Number of roadway links divided by number of nodes, with the count of nodes including both intersections and cul-de-sac/dead-end-street termini. A grid nine square blocks in extent (not 16-242

counting any exterior connections) receives a score of 1.5. It is suggested that “[a] score of 1.4 is the minimum needed for a walkable community.” • The ratio of street intersections divided by the sum of street intersections and dead ends. A score of over 0.75 is suggested as desirable. Unfortunately, these and similar measures take into direct account neither the directness produced by the interconnections nor the association or lack thereof with desired destinations. The “direct- ness” Accessibility Index introduced in the preceding “Accessibility” discussion encompasses all elements of connectivity more fully: direct (“airline”) travel distance divided into actual minimum travel distance to destinations, with smaller ratios approaching 1.0 the more desirable. A related measure, essentially catchment area coverage, is the proportion of the circular area described by a given radius that can be reached from the center within an actual walking or bicy- cling distance equal to the radius. Not only is the coverage always less than 1.0 because of the need to often angle through even a grid system, it also may be less because of cul-de-sacs, other system elements with poor connectivity, missing sidewalk links, and natural or man-made barriers. LRT station access examples in Chapter 17, “Transit Oriented Development,” provide such a catchment area analysis (see the first-listed results in the Chapter 17 case study “Travel Findings for Individual Portland, Oregon, Area TODs”). The examples were constructed for walk access assum- ing the sidewalk and walkway system to be adequately represented, for computational purposes, by the street network. On this non-conservative basis it was estimated that only 21 to 57 percent of the areas within a 1/4-mile airline-distance radius around the four studied LRT stations was actu- ally within a 1/4-mile along-the-road walking distance (Schlossberg et al., 2004). Catchment area analysis can be elevated from a connectivity measure to an accessibility measure by bringing land use into the calculation. Such an approach was utilized to map and understand impediments to walking and bicycling to school in Hillsborough County, Florida. The county defines a “walk zone” of 2-mile radius around each school within which no school bus transporta- tion is provided. (It has been suggested that this could better be described as a “parent responsi- bility zone” than “walk zone.”) Land use was introduced into the analysis by making coverage calculations not on the basis of area per se, but on the basis of residential parcels. When this was done, at least one elementary school was shown to have dramatically inferior access from residen- tial parcels than from undifferentiated land uses. The thrust of the remaining steps is best illus- trated by example, for which the suburban Walden Lakes Elementary school is used: Walden Lakes Elementary’s school attendance area does not mesh perfectly with its so-called “walk zone”; in fact, only 81 percent of residential parcels in the attendance area are located within the “walk zone.” The remainder are beyond a 2-mile airline-distance radius from the school. (Some portions of the “walk zone” are in the attendance areas of other schools.) When 2-mile walk distances are measured along the roadway system to delineate a “connected network zone” (similar to the Portland catchment area analysis, but not yet adjusted for highway barriers), then only 58 percent of the attendance area res- idential parcels are included. When the barrier effect of major roads is taken into account, only 49 per- cent of attendance area residential parcels are within a 2-mile walk without major impediments. The Walden Lakes Elementary situation is somewhat worse than that of the average Hillsborough County suburban elementary school—it serves only as an example of the analytical approach. The approach is designed to help involved parties better understand the impact of school siting and attendance area insti- tutional decisions on school accessibility and walk-to-school possibilities (Steiner et al., 2008). Motorized and NMT connectivity are not necessarily the same in any given situation, and may be separately calculated, and compared. Limited access highway facilities may prohibit walking and 16-243

bicycling; they may also (along with major arterials) create barriers to NMT, and lack of sidewalks on busy streets may further inhibit walking. All such negative factors need to be taken into account in NMT accessibility calculation. At the other end of the scale, hilly neighborhoods may have stair- ways and pedestrian/bicycle ramps in lieu of steep street segments. There are subdivision and new-town designs (not typically found in the vast auto-dominated suburban housing tracts of the latter half of the 20th Century) that use pathway connections and path connectivity through small- and-medium-sized parks to facilitate pedestrian and bicycle flow similar to grid street/sidewalk systems while, at the same time, inhibiting through traffic. Such “Fused Grid” layouts have been estimated to increase the odds of walking by almost 10 percent (Victoria Transport Policy Institute, 2011b, Stover and Koepke, 2002). Connectivity in application is the focus of the subsection “Pedestrian/Bicycle Systems and Inter- connections” within the “Response by Type of NMT Strategy” section. For example, further infor- mation is found there on the mode choice effects of differential motorized and NMT connectivity, and on travel impacts of “Fused Grid” subdivisions (see “River Bridges and Other Linkages”— “Interconnections of Modest Scale,” and the 14th entry of Table 16-21). Connectivity’s full partner in producing accessibility—land use—is likewise addressed, along with design, in the “Pedestrian/Bicycle Friendly Neighborhoods” subsection within the same “Response by Type of NMT Strategy” section. Surroundings Environment The most obvious and direct manifestation of the link-level built environment for the pedestrian or bicyclist is the set of travel conditions encountered on a pedestrian/bicycle facility segment itself. This aspect is covered here in terms of “Facility Compatibility Measures.” Next most imme- diate is the setting through which the facility segments pass. That is discussed under “Ambiance.” Facility Compatibility Measures. Similar to the familiar case of roadways designed for motor vehicle drivers, it is recognized that pedestrian and bicycle facilities should be designed, improved, maintained, and operated to meet the primary needs and preferences of non-motorized traffic and travelers. Preferences for certain types and/or designs of facilities may vary somewhat by trip pur- pose, length, and user characteristics. However, elements of safety, comfort, convenience, and min- imal space conflicts are nearly universal. A number of studies to define and quantify these preferences have been conducted, with most such studies surveying representative pedestrians and bicyclists about their satisfaction (or lack of sat- isfaction) with various physical elements of their walking or bicycling trip. Width of sidewalk or bike lane and separation from motor vehicle traffic are examples of physical elements addressed. When quantified into a single measure, these preferences have been given many names such as suitability criteria, compatibility criteria, level of service measure, and stress level. Measures of pedestrian and bicycle facility compatibility with user needs and preferences not only serve as design tools and improvement prioritization criteria; they also help describe facility attractiveness to present and prospective walkers and cyclists. Two comprehensive studies, one relevant to pedestrian preferences and the other to bicyclist preferences, are described here to illustrate the makeup of compatibility measures. Multi-modal research utilizing field survey evaluations was the basis for development of a pedes- trian level of service model for the Florida Department of Transportation. The equation for the pedes- trian level of service is shown below and considers many of the physical elements that one intuitively uses to “grade” a particular walking experience. In particular, the presence of a sidewalk, and lateral separation from motor vehicle traffic, were significant determinants of pedestrian level of service. 16-244

Motor vehicle volumes and speeds in adjacent traffic lanes were also determined to be significant variables (Landis et al., 2001). Where: Wol = Width of outside lane (feet) Wl = Width of shoulder or bike lane (feet) fp = On-street parking effect coefficient (=0.20) %OSP = Percent of segment with on-street parking fb = Buffer area barrier coefficient (=5.37 for trees spaced 20 feet on center) Wb = Buffer width (distance between edge of pavement and sidewalk, feet) Ws = Width of sidewalk, feet fsw = Sidewalk presence coefficient = 6-0.3Ws Vol15 = Average traffic during a fifteen (15) minute period L = Total number of (through) lanes (for road or street) SPD = Average running speed of motor vehicle traffic (mph) The pedestrian level of service equation was developed using a stepwise multi-variable regression of 1,250 observations from an experiment using 75 walkers on a Pensacola, Florida, roadway course. The 75 walkers proceeded through a 21-segment (42-directional-segment) roadway course, with many starting at different segments and walking in different directions. The walkers were instructed to grade the segments immediately after they were walked, with the opportunity to re-grade previ- ous segments based upon accumulated experience on the walking course. Walkers graded the road- way segments on a numerical scale of 1 to 6, corresponding to level of service A to F. The pedestrian level of service model developed through this research was for use in the Florida Department of Transportation’s multimodal corridor evaluation set of techniques as mandated by the state legislature. A similar field survey experiment was conducted for bicyclists in the mid-1990s and resulted in a similar bicycle level of service model (Landis, Vattikuti, and Brannick, 1997). A team of researchers developed a Federal Highway Administration (FHWA) bicycle compatibil- ity index in the late 1990s to quantify the “bicycle friendliness” of roadways. It was developed to allow practitioners to evaluate existing facilities to determine what improvements may be required, as well as determine the geometric and operational requirements for new bicycle facili- ties. The index is calculated as shown in Table 16-64. The significant variables include: (a) the pres- ence and width of a paved shoulder or bicycle lane, (b) motor vehicle traffic volume and speed in adjacent lanes, (c) presence of motor vehicle parking, and (d) the type of roadside development (Harkey et al., 1998b and a). Ped LOS W W f OSP f W f Wol p b b sw= − + + × + + + +1 2021 1. ln % s 15 2Vol L SPD ( ) + ( )+ +0 253 0 0005 5 3876. ln . . 16-245

In developing the bicycle compatibility index, the research team used the perspectives of more than 200 persons in three cities (Olympia, Washington, Austin, Texas, and Chapel Hill, North Carolina) to subjectively evaluate the perceived bicycling “comfort level” in different roadway environ- ments. The approach used in this study relied on participants viewing roadway segments on videotape. Validation of the videotape method was accomplished with an on-street pilot study using 24 participants and 13 different roadway segments. After viewing the videotape for a par- ticular roadway segment, each of the 200-plus participants was asked to “grade” the segment on a numerical scale of 1 to 6, corresponding to level of service A to F (Harkey et al., 1998b). 16-246 Table 16-64 Bicycle Compatibility Index (BCI) Model, Variable Definitions and Adjustment Factors BCI = 3.67 – 0.966BL – 0.410BLW – 0.498CLW + 0.002CLV + 0.0004OLV + 0.022SPD + 0.506PKG - 0.264AREA + AF Where: BL = presence of a bicycle lane or paved shoulder > 0.9 m No = 0 Yes = 1 PKG = presence of a parking lane with more than 30 percent occupancy No = 0 Yes = 1 BLW = bicycle lane (or paved shoulder) width m (to the nearest tenth) AREA = type of roadside development Residential = 1 Other type = 0 CLW = curb lane width m (to the nearest tenth) AF = ft+ fp + frt CLV = curb lane volume vph in one direction Where: OLV = other lane(s) volume – same direction vph ft = adjustment factor for truck volumes (see below) SPD = 85th percentile speed of traffic, km/h fp = adjustment factor for parking turnover (see below) frt = adjustment factor for right-turn volumes (see below) Adjustment Factors Hourly Curb Lane Large Truck Volume a ft Parking Time Limit (min.) fp > 120 60 – 119 30 – 59 20 – 29 10 – 19 < 10 0.5 0.4 0.3 0.2 0.1 0.0 < 15 16 – 30 31 – 60 61 – 120 121 – 240 241 – 480 > 480 0.6 0.5 0.4 0.3 0.2 0.1 0.0 Hourly Right- Turn Volume b frt > 270 < 270 0.1 0.0 Notes: a Large trucks are defined as all vehicles with six or more tires. b Includes total number of right turns into driveways or minor intersections along a roadway segment. Source: Harkey et al. (1998b).

Various other compatibility or level of service measures have been developed in recent decades for both pedestrians and bicycles. Elements frequently used include type of facility provided, such as mixed traffic lane vs. bicycle lane vs. shared use path; sidewalk, path, curb lane, bicycle lane, or paved shoulder widths; some form of vehicular traffic volume measure (typically curb lane vol- ume); motor vehicle speeds (speed limit is often used as a surrogate); presence and type of sepa- ration from motor vehicle traffic; roadway and driveway crossing conditions; and type of adjacent land use (Victoria Transport Policy Institute, 2011a). In addition to use as design and sufficiency study tools, bicycle compatibility criteria also can be employed to identify streets or highways par- ticularly amenable to bicycle travel. Additional factors or variables can be used to supplement those already listed to determine those facilities most compatible for walking and bicycling and, therefore, most likely to elicit positive traveler response. As useful as they may be as measures of suitability, compatibility measures such as link-based level of service share one substantial limitation. They can be used to identify network segments with substandard and unattractive characteristics, but they cannot be used to rank the importance of such links to the completeness or connectivity of the NMT network. A facility segment built to high standards may or may not be a crucial contributor to NMT network functionality. Conversely, a link may be identified as deficient, but some other segment—existing or yet unbuilt—may be more important to network completeness and connectivity (McCahill & Garrick, 2008). Broad level of service measures of continuity and connectivity have been proposed (Victoria Transport Policy Institute, 2011a), but inclusion of accessibility in compatibility measures apparently awaits further developments. Ambiance. Various studies have, with somewhat mixed results, attempted to quantify the effect on choice to walk or bicycle of such features as tree shade, streetscape variety, opportunity to see peo- ple, building setbacks, and intrusion of automobile parking. On balance, there appears to be a mod- est positive effect of pleasant environment on walking (Cao, Handy, and Mokhtarian, 2006, Saelens and Handy, 2008, Ewing and Cervero, 2010). The relationships have proved difficult to ferret out, and are quite likely direct and robust only for “other” purpose trips such as shopping and strolling. An effort to “comprehensively and objectively measure subjective qualities of the urban street environment” and then test them as walkability descriptors reached the point in early 2009 of set- ting forth operational definitions of five selected urban design qualities. Definitions development was facilitated by a panel of ten experts, each bringing different perspectives from diverse fields related to urban design and planning. Prior research, the panel’s expertise, and a factorial design were utilized to organize a street-scene-aided process of rating over 50 perceptual qualities and winnowing them down into a set of urban design qualities capable of linkage with significant physical features. The rating process involved viewing by the panel of streetscape video clips cov- ering 48 commercial district streets selected from dozens of cities across the United States. Statistical models were estimated with physical characteristics as independent variables and the ratings as dependent variables. These models indicated which physical characteristics are signifi- cantly related with each perceptual quality, along with the strength and direction of the associa- tion. Of eight urban design qualities carried through the entire process, three could not be defined operationally. The following five were those retained for further study and validation (Ewing and Handy, 2009): • Imageability—Distinctive and recognizable quality of place that captures attention and creates a lasting impression. • Enclosure—Visual definition of streets and other public spaces by walls, trees, and other ver- tical elements of proportions that create a room-like quality. 16-247

• Human Scale—Size, details, and articulation of physical elements matching the size and pro- portions of people and their walking speed. • Transparency—Ability to see or perceive what is going on or lies beyond the street edge, specif- ically including human activity. • Complexity—Visual richness of a place, including number and diversity of buildings, orna- mentation, landscape and street furniture elements, and human activity. Testing of these measures has been facilitated by a project to carry out urban design measurement of 588 block faces in New York City, representing a stratified sample from all five boroughs rang- ing in development intensity from Manhattan to partially rural Staten Island (Ewing and Handy, 2009). Initial unpublished results of development of a blockface pedestrian volume model incor- porating as variables the urban design quality measures suggest that some but not all have statis- tical significance in describing pedestrian activity, at least in this particular application. The only one fully reaching statistical significance is transparency. Imageability “comes close.” Both are pos- itively related to higher pedestrian volumes. Enclosure is also marginally significant, but negative, suggesting that it is perhaps the least promising descriptor of walkability (Ewing, Connors, and Neckerman, 2011). More conventional quantitative urban design variables were also incorporated. Significant and positive are floor area ratio (FAR), a measure of density; and walkscore, a measure of destination accessibility. An entropy measure of land use mix is positively related, but does not reach statisti- cal significance. Significant and negative are distance to the nearest subway station and intersec- tion density. The negative subway distance relationship relates to the added walking activity generally found around (not distant from) major transit stops. The negative intersection density relationship, which must be viewed in context with the generally high intersection density of New York City, is thought to reflect the associated greater use of land area for streets and the corre- sponding reduction of trip-generating acreage (Ewing, Connors, and Neckerman, 2011). In con- templating this it must be kept in mind that this is a pedestrian-volume direct-demand model, not a mode split model calibrated to estimate choice probabilities. The most significant components of the statistically significant transparency variable are proportion of the first floor with windows and proportion of frontage with active uses. Less important is proportion of blockface with building frontage. The most significant components retained for validation of the other contributing urban design quality measure (imageability) are proportion of historic buildings; number of courtyards, plazas, and parks; and presence of outdoor dining. With these components, the two variables in question encompass all of the “physical features” describing activity and ability to per- ceive what lies beyond the street edge. Among the four non-contributing urban design quality mea- sures only “complexity” includes such features, none of them as strongly related components (Ewing and Handy, 2009, Ewing, Connors, and Neckerman, 2011). “Activity” and “ability to perceive what lies beyond the street edge” could well be characterized as “Jane Jacobs” variables—important to the life of cities. It is interesting to contemplate that these initial New York City validation outcomes may be identifying activity (or human presence or vitality) and eyes-on-the-street (or vice versa) as being particularly strong perceptual indicators of walkability. Some of the more definitive already-published results with respect to ambiance come from an advanced travel demand modeling effort in San Francisco proper. Aside from topography, which in San Francisco tends toward the dramatic, the most influential destination pedestrian environ- ment factor (PEF)—among those scaled for model use by a Delphi panel—proved to be urban vital- 16-248

ity. 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 (for “other” trip pur- poses only) bicycling. Only destination (non-home) PEFs proved useful in estimating mode choice in the San Francisco modeling context (Cambridge Systematics et al., 2002). There has also been one study that found high workplace and vicinity aesthetic appeal to be a marker for walk and bike work trip shares higher by an average of 0.7 percentage points (without Travel Demand Management financial incentives) to 1.3 percentage points (with incentives) as compared to sites with low aesthetic appeal. At least part of the underlying cause is presumably the secondary influence of having an employment area pleasant to get around in when making midday trips without an automobile (Cambridge Systematics with Deakin, Harvey, Skabardonis, 1994). These two Cambridge Systematics studies are more fully described in Chapter 15, “Land Use and Site Design,” in that chapter’s “Response by Type of Strategy” section (see “Site Design”— “Transit Supportive Design and Travel Behavior” including Tables 15-41, 15-42, and 15-44). A University of California doctoral dissertation has closely examined factors pertaining to the choice between walking and auto use in the context of routine trip tours involving at least one shopping stop. For convenience of surveying and detailed interviews, the shopping activity selected was an intercepted visit to one of various pharmacies in commercial areas distributed throughout much of the San Francisco Bay Area. Complete trip and socioeconomic data were obtained in 959 out of 1,003 customer surveys for the tour intercepted, and these data were linked with travel and neighborhood characteristics data. Mode choice models were developed for three travel categories. For the 397 tours that visited solely a single shopping district, mode share to and from the district was modeled. For all 959 surveyed tours, tour mode choice was addressed. Lastly, a mode choice model was calibrated for only those trips within tours that took place entirely within one of the various shopping districts. One ambiance-related environmental variable was included among the many socioeconomic, travel, shopping district, and attitude/perception factors. This shopping district variable was defined as tree canopy coverage within the public right-of-way of all multi-lane streets within 1/2 mile. The sample mean was 6.5 percent and the maximum was 18.1 percent. (These seemingly low percentages quite likely resulted from minimal tree canopy coverage over the central portions of the broad roadways involved.) The variable was significant in the “to and from” model of mode share for single shopping district tours, but not in the all-respondent tour-mode model or the shop- ping-district internal-trip model (Schneider, 2011). Using the calibrated “to and from” mode choice model (R2 = 0.52) it was estimated that 1 percent more tree coverage was worth, for the average respondent, taking 2.1 more minutes to walk. Sensitivity testing suggests that doubling tree coverage would have over 2-1/2 times the positive effect on walk mode share as doubling population and employment density, and 3 times the pos- itive effect on walk share as halving the parking supply at the store. It is perhaps telling that one of the respondents in the 26 follow-up interviews stated: “ ‘Generally streets that also have trees are nicer streets . . .’ ” The dissertation finds that “. . . improving the quality of the street environ- ment may extend walking distances and increase the pedestrian catchment area . . .” (Schneider, 2011). This measured conclusion seems appropriate. The sensitivity tests on one of three models provide a tenuous basis for any stronger judgment given that the other two models find no tree canopy coverage significance and, as the quoted interviewee implies, tree coverage may be stand- ing in as a measure of overall street environment quality (a subject of interest in itself). The shade from street trees has also been found to be important in Austin, Texas, research, but only for “strolling” recreational/exercise walking and not for shopping trips (Shriver, 1997, Cao, 16-249

Handy, and Mokhtarian, 2006). Research focused on the city of Seattle concluded that architectural variety was related to frequent recreational walking but not utilitarian walking (Lee and Moudon, 2006a). The pertinent studies are summarized in the “Response by Type of NMT Strategy section” (see “Sidewalks and Along-Street Walking”—“Sidewalk Coverage and Traffic Conditions”). Additional relevant findings from the San Francisco trip tour modeling are also provided in that same discussion. Detailed study of travel routes selected for walking to rail transit, combined with interview results, indicate that taking the most direct route dominates over considerations of ambiance for purposes of route choice when the walker has a crucial practical objective such as catching a train or getting to work on time (Weinstein et al., 2007). At the other end of the scale, walking and cycling done for recreation and exercise clearly follow a different decision paradigm than purely utilitarian NMT travel, as already discussed under “Behavioral Paradigms”—“Derived Versus Direct-Benefit Demand.” For non-utilitarian trips, there tends not to be a precise destination, although for survey purposes the farthest point reached may be selected as an arbitrary trip end point. Global Positioning System (GPS) studies in Portland, Oregon, found—in the context of that region’s mul- tiplicity of bicycle facilities—that bicycle trips for exercise were typically structured as loops. The route followed is itself the “destination” for recreation and exercise trips (Weinstein and Schimek, 2005, Broach, Gliebe, and Dill, 2009a), and the ambiance—as expressed in factors such as views, scenery, and the strolling inducements identified in the Austin studies—likely has heightened influence. Trip Factors Trip factors include attributes of a specific journey including the origin-destination pairing, route, travel cost, trip purpose, and time of day. Trip distance is dependent on the origin-destination pair- ing and the route selected for getting from one place to the other. Travel time and cost will vary in accordance with both the trip distance and travel mode and route selection. Most travel models assume that individuals are aware of the distance, time, and cost associated with their potential choices and thus make a decision about how to travel on the basis of this knowledge. Of course, not all individuals think alike, have the same level of travel options knowledge, or have the same actual options available. In that context, this subsection is concerned with the general influence of trip factors on travel decisions. The next subsection, “User Factors,” addresses how travel deci- sions are influenced by the characteristics of individuals. Pedestrians and bicyclists are particularly sensitive to trip distance and are keen to seek out the most direct routes possible, particularly for utilitarian trips. This predilection is mainly attribut- able to the slower speeds at which persons walking or cycling cover the ground as compared to motorists and transit riders. Pedestrians and cyclists are also more exposed to their environment than are people inside vehicles. That circumstance can contribute to trip distance sensitivity, as people seek to minimize time spent in unpleasant or unsafe surroundings (Ewing, 1997). For clarity of presentation, the following discussion of trip distance, time, and route characteris- tics is organized into separate parts for walking and bicycling as pertains primarily to adult travel. Within each is a discussion of average trip lengths, access to transit trips, and route choice. Factors as they pertain to the travel of children to and from school are introduced at the close of this “Trip Factors” subsection, following discussions of cost and trip purpose effects. The “Pedestrian/ Bicycle Linkages with Transit” subsection within the “Response by Type of NMT Strategy” sec- tion provides additional information on impacts of distance on choice of walking or bicycling to access transit service. 16-250

Walk Trip Distance, Time, and Route Characteristics Utilitarian walking is done to accomplish activities requiring travel to another location, while recreational walking is done for exercise or enjoyment. The two objectives may be combined (Cao, Handy, and Mokhtarian, 2006). Sometimes people will tolerate a longer walk because they enjoy walking or because they recognize the exercise benefits. Trip purpose is discussed more fully below, as a separate traveler response factor. However, it is important to bear in mind the potential blending of trip purposes and objectives as findings about walk trip lengths are reviewed. Walk Trip Speeds and Lengths. The average pedestrian can walk between 3 and 4 miles per hour (mph). The most thoroughly studied walking condition, in terms of speed, is that of pedestrians crossing streets. Joint TCRP/NCHRP research, based on both original data collection and prior studies, provides a recommendation that a street crossing speed of 3.5 ft./sec. (2.4 mph) be assumed for the general population and that 3.0 ft./sec. (2.0 mph) be used for older or less able persons. These recommendations relate to intersection design and signal timing, however, and thus represent more conservative (lower) speeds than seen with typical walking. An Australian study of signalized intersections, for example, found 15th, 50th, and 85th percentile street crossing speeds for those pedestrians walking without difficulty or encumbrances of 4.27 ft./sec. (2.9 mph), 5.25 ft./sec. (3.6 mph), and 6.69 ft./sec. (4.6 mph), respectively. (Examples of encumbrances included large packages and small children in tow.) The corresponding values for pedestrians walking with difficulty or encumbrances (6 percent of the observations) were 3.74 ft./sec. (2.6 mph), 4.23 ft./sec. (2.9 mph), and 5.34 ft./sec. (3.6 mph). The mean speeds observed at midblock cross- ings with pedestrian-actuated signals were, for able-bodied, unencumbered pedestrians, 10 per- cent lower (Fitzpatrick et. al, 2006). Walking speeds at signalized crossings may be affected by the pressure of the need to cross safely. They are thus are not necessarily representative of the speeds walked over longer distances, which are of interest to transportation planners and public health practitioners. Average unimpeded speeds suggested by NCHRP research for benefit analysis of pedestrian grade separations are 4.92 ft./sec. (3.4 mph) for normal conditions, 4.45 ft./sec. (3.0 mph) for commuters in busy down- town areas, and 5.33 (3.6 mph) for students. These were based on observed ranges of 4.50 to 5.00 ft./sec. (3.1 to 3.4 mph) in downtown Ottawa and 4.07 to 4.30 ft./sec. (2.8 to 2.9 mph) in more- crowded downtown Brooklyn. Note that these speeds exclude delays at street crossings, which are intended to be added in as a separate analytical step (Roddin, 1981). Studies in Brisbane, Australia, took the further step of determining average pedestrian speeds along routes extending 1 to 9 blocks through intersections that were almost all signalized. Walking was thus subject to signal delays. Measurements were made using the pedestrian equivalent of the “floating car” highway travel time measurement technique. Speeds along 13 individual routes ranged from 40.6 to 84.1 meters/minute (2.22 to 4.60 ft./sec., or 1.5 to 3.1 mph), and averaged 66.9 meters/minute (3.66 ft./sec., or 2.5 mph). Speeds were observed to be affected by signal timing and coordination. This same Brisbane study also obtained free-flow measurements, observing 345 pedestrians away from the influence of traffic signals, and obtained a bell-shaped distribution with a mean of 90.0 meters per minute (4.92 ft./sec., or 3.4 mph) (Virkler, 1998)—identical to the NCHRP recommendation for “nor- mal” conditions reported above. A 10-minute walk at 3 to 4 mph can take a pedestrian 1/2 to 2/3 miles. Indeed, the 1990 and 1995 National Personal Transportation Surveys (NPTS) found average walk trip lengths (transit access walking excluded) of 0.6 and 0.5 miles, respectively. The 2001 National Household Transportation Survey (NHTS), with its enhanced walk trip surveying protocol, again found the average walk-only 16-251

trip length to be 0.6 miles. However, the reported travel time averaged not 10 minutes, but 16.4 minutes.52 Walking is the mode of choice for nearly all trips of 1/10 of a mile or less. In contrast, for trips over 1/2 mile, walking is chosen only 10 percent of the time. Median distances are generally less than one-half the means for the various trip purposes (Ewing, 1997, Agrawal and Schimek, 2007). The 2009 NHTS obtained a mean walk trip distance of 0.70 miles, a mean reported travel time of 14.9 minutes, a median walk distance of 0.44 miles, and a median walk time of 10 minutes. The mean calculated speed was thus 2.8 mph, while the median calculated speed was 2.6 mph (Kuzmyak et al., 2011). Despite the inaccuracies inherent in self-reported travel times obtained in surveys such as the NHTS, these speeds are within the range of 2.5 mph average impeded walk speed and 3.4 mph average unimpeded walk speed as determined in Brisbane. The median walk time finding lends further substance to the rule of thumb that 10 minutes is the typical amount of time devoted to a walk trip in the United States. Note, however, that the NPTS and NHTS results reported here are for walk-only trips and exclude transit access, parking facility access, and other walking for access to motorized transportation. The shortest walk trips are most common in central cities where potential origins and destinations are located close together. The 1995 NPTS also found that people would walk longest for commut- ing purposes, and longer for recreational trips than for non-work utilitarian trips (Morris, 2001). A Florida survey found an average home-to-work walk trip length of about 0.7 miles (NuStats International, 1998). The 2001 NHTS likewise found non-work utilitarian trips to be the shortest on average. However, it found recreation/exercise walk trips to average 1.2 miles each way, more than the 0.8 miles average for walk trips to or from work (Agrawal and Schimek, 2007). The 2009 NHTS obtained longer work-related walk trip lengths, ranging from 1.0 miles for the work com- mute to 1.1 miles for work-related trips. Reminiscent of the 1995 NPTS, it found social/recreational walk trips to be longer than most non-work trips, but shorter than work trips. However, the social/recreational trip category used for the distance calculation of 0.8 miles included a broad range of activities from exercise to “get/eat meal” (Kuzmyak et al., 2011). The relatively long distances covered by recreational walk trips provide one possible explanation for results from other surveys and studies that have derived longer average walking distances. For example, the 2002 National Survey of Pedestrians and Bicyclist Attitudes and Behaviors reported an average length of 1.2 miles for summer walking trips (NHTSA and BTS, 2002). The survey methodology, which focused on the most recent day a respondent walked rather than on a fixed survey day, would have over-weighted walk trips that occur less than daily—such as, perhaps, recreational trips. Another likely factor is the use in some surveys of round-trip mileage for “loop” or “out-and-back” recreational trips, a protocol employed by the BTS 2002 summer survey. More on national trip length statistics is found in the initial subsections of the “Related Information and Impacts” section, most particularly under “Characteristics of Walking and Cycling Overall”— “Trip Distance and Duration.” A University of Minnesota “Access to Destinations” research effort has looked at whether the old “one quarter mile assumption” used in community planning as a measure of walk access viability is truly valid. Data from the Twin Cities regional travel survey and complementary transit and trail 16-252 52 This suggests an average speed of only 2.2 mph; however, travel surveys are an imperfect source of travel times, since respondents tend to round off start, finish, and elapsed times.

travel surveys were employed to plot “decay functions” of walking and cycling activity relative to increasing distance. A summary illustration of the walking prevalence data points and fitted expo- nential decay function curves for work, shopping, restaurant, and recreation trips is provided in Figure 16-8. One notable finding for walking was that there is not much difference among the decay functions for the work, shopping, and restaurant utilitarian trip purposes.53 The researchers also con- cluded that “a surprising number of [walk] trips are made at distances up to and even exceeding 1 km. (0.6 mile)” (Iacono, Krizek, and El-Geneidy, 2008). 16-253 53 Entertainment, recreation, and fitness trips tended to cover longer distances, magnified in this instance by the recreation/fitness trips having been predominantly recorded in terms of round-trip distances (Iacono, 2011, Filipi, 2011). Figure 16-8 Walk trip distance decay plots with exponential curves. Note: “Entertainment” includes recreation/fitness trips, predominantly recorded in terms of round-trip travel distances (Iacono, 2011, Filipi, 2011). Source: Iacono, Krizek, and El-Geneidy (2008). Others, as well, have seen neighborhood design implications in trip length consistencies among util- itarian walk trip categories. Litman, for example—noting that the mean “across many demographic groups and in different neighborhood densities” does not deviate far from 1/2 mile—suggests it may be reasonable for planning purposes to use this distance as the maximum many Americans are ordi- narily willing to walk in satisfaction of travel needs (Victoria Transport Policy Institute, 2007).

Examination of the individual data plots from which Figure 16-8 is summarized shows a tendency for walk trip percentages to plateau at the shortest distance intervals rather than following the expo- nential decay function, a phenomenon familiar to modelers practiced in the conventional calibration of “gravity” trip distribution models. In other words, there is little empirical indication of walk trip travel choice differences at, say, 0.2 km. compared to 0.4 km. The sharp drop-off for work and shop trips, in particular, starts at distances higher than 0.5 to 0.6 km. (1/3 mile). Within the 1/3 mile walk trip threshold, approximately 55 to 65 percent of work, shopping, and restaurant trips occur. Beyond 1/3 mile, the decline in walking is steep, especially considering that each additional equal increment of distance from the central point encompasses a greater land area. Using this threshold as a design- guidance indicator might suggest that the preferred outer limit for advantageous accommodation of utilitarian walk trips may be 1/3 mile as measured along the walkway system. That equates to a pedestrian-oriented airline-distance radius of roughly 1/4 mile around the land use activity central point of interest, assuming a grid system of walkways, but with the space thus defined “bumped out” to 1/3 mile at streets and walkways radial to the center. There is still another instructive way to look at the meaning of the Twin Cities utilitarian walk trip decay data, this one from the perspective of facility design. The decay plots and functions suggest that, in the interval between 1/3 and 3/4 miles of trip length, degree of walking is highly sensitive to walk distance. Walk network directness will, therefore, have a major effect on the choice of walk- ing from origins in the critical band around the destination of interest. This band, if treated as ring or “donut,” is approximately 900 acres (1.4 square miles) in area. The formulae for the exponential curves in Figure 16-8 can be applied to produce an approximate estimate that 0.25 km. (820 feet) of walkway indirectness, affecting trips from within this critical band to the central point of inter- est, will result in a reduction in utilitarian walking of some 30 to 40 percent or more.54 Adding additional complexity is the phenomenon of tours, series or chains of work-related or non- work trips made starting at and ultimately returning to the same location. A short trip within a tour may not be a candidate for walking or bicycling because other travel within the tour requires use of an automobile (Schneider, 2010). It has been shown, using data on tours intercepted at phar- macies at various locations in the San Francisco Bay Area, that tour length is a good predictor of whether the walk or bike mode will be selected as the primary tour mode. Approximate walk mode shares for tours by distance were: 0.0 to 0.5 miles, 80 percent; 0.5 to 1.0 miles, 76 percent; 1.0 to 1.5 miles, 62 percent; 1.5 to 2.0 miles, 48 percent; 2.0 to 2.5 miles, 28 percent; 2.5 to 3.0 miles, 20 per- cent; and greater than 3.0 miles, 4 percent. Whereas the median distance for all intercepted tours was 5.2 miles, the median walk tour distance was only 1.2 miles. Median tour distances for other primary tour modes were: bicycle, 3.1 miles; transit, 8.2 miles; and auto, 7.8 miles (Schneider, 2011). 16-254 54 To place 820 feet of indirectness in context, consider that a superblock 800 feet on a side, straddling a direct pedestrian route, will introduce that amount of indirectness unless the superblock is pierced by a midblock public walkway with suitable street crossings at each end. Or consider that elimination of a pedestrian cutoff saving about 800 feet was the subject of public debate, outcome uncertain, during an actual transit- adjacent development approval process affecting land and access at a Washington Metrorail station. Or that 1/2-mile spacing of pedestrian crossings of a suburban arterial introduces up to 2,640 feet (average 1,320 feet) of indirectness between intermediate local streets, bus stops, and/or building entrances on oppo- site sides of the arterial. Note that implicit in the decay functions are aerial coverage effects that render this particular application of the formulae an approximation likely to produce conservative impact estimates. The analytical assessments in this and the preceding paragraph, and in this Footnote, are solely by the Handbook authors.

The distances pedestrians are willing to walk are influenced by the built environment. A number of studies have concluded that interesting walks seem shorter than boring walks. The underlying hypothesis is that pedestrians latch onto intermediate goals or destinations as points of orientation along the way, such that the sense of distance and time is psychologically shortened. The researchers assert that for this reason pedestrians tend to walk further in areas with short block lengths or mixed land uses. More frequent intersections within a grid pattern can also mean shorter and easier trips when the straight-line path would be a diagonal. Other benefits of short blocks include the potential for greater dispersion of automobiles, thereby resulting in lower traffic vol- umes on adjacent streets and easier street crossings (Zehnpfenning et al., 1993, Ewing, 1996). Topics related to block size, land use mix, and intersection frequency are covered conceptually in the preceding “Environmental Factors” subsection (see both “Systems Environment” and “Surroundings Environment”). They are examined in terms of specific neighborhood land use mix and design features in the “Pedestrian/Bicycle Friendly Neighborhoods” subsection of the “Response by Type of NMT Strategy” section. Another line of inquiry suggests that a higher density of landmarks results in perceptions of space that exaggerate the actual distance involved. The underlying “feature accumulation hypothesis” states that when there are more intersections, turns, and other information to remember about an environment, distances are perceived as longer. Apparently unknown is whether the distance exaggeration reported to be perceived, in these circumstances, actually diminishes walking or not. One clear finding is that peo- ple estimate walking distance poorly. In research involving 910 usable responses from a 3,000-house- hold survey in Minneapolis and two of its suburbs, only 38 percent of the distance estimates obtained fell into the correct 5-minute (up to 10 minutes), 10-minute (11 to 30 minutes), or over-30-minutes cate- gory. The Minneapolis researchers posit that to encourage walking, in addition to providing as many businesses close at hand as possible, it may be important to provide “consumer education” about oppor- tunities to meet utilitarian travel needs by walking (Horning, El-Geneidy, and Krizek, 2008). (See the “Walking/Bicycling Promotion and Information” subsection of the “Response by Type of NMT Strategy” section for an examination of such approaches.) Much of the available distance perception research concludes that people tend to overestimate the distances that they might walk. Distance overestimation may help explain some decisions not to walk even when distances are within typical walking norms (Goldsmith, 1992, Loutzenheiser, 1997). Other research, specifically the experiments in Minneapolis and its suburbs comparing per- ceived distances to network and also airline actual distances, have found distance overestimation among residents of closer-in areas where built-environment features are more concentrated and distance underestimation among people living further out in less dense environments. Various fac- tors were associated with more accurate travel time estimation, but only having the closest desti- nation within 5 minutes was consistently significant among destination characteristics as a predictor of accuracy. Closeness was a positive variable, and occurs more frequently, of course, where land uses are mixed and concentrated (Horning, El-Geneidy, and Krizek, 2008). Mostly-newer research findings indicate that responses to the immediate built environment differ sub- stantially by trip purpose, with recreational, exercise, and discretionary utilitarian trips (such as shop- ping) much more influenced by ambiance than non-discretionary utilitarian trips such as commuting (Cao, Handy, and Mokhtarian, 2006, Weinstein et al., 2007). These and related issues are among the findings covered in the “Ambiance” discussion at the end of the preceding “Environmental Factors” subsection. Walk Access to Transit. Travel survey conventions are such that much of the data on average walk trip lengths in the preceding discussion pertain only to walk trips which use no other travel mode. 16-255

Thus different data, or at least different data compilations, are required to examine factors pertain- ing to choice and use of the walk mode for access to transit service. Most transit patrons will walk about 1/4 mile to bus service and farther to rail service (Replogle and Parcells, 1992). In general terms, the greater the distance from a transit stop, the less likely a potential transit rider is to walk or even to use the transit service at all. A number of study and research examples demonstrating and further quantifying this phenomenon are provided in the “Pedestrian/Bicycle Linkages with Transit” subsection within the “Response by Type of NMT Strategy” section (see “Non-Motorized Access to Transit”—“Pedestrian Access and Egress”). Often overlooked in interpreting observed walk distances to bus transit is that the more intensive urban bus services are typically designed so that no more than a 1/4-mile walk is necessary. If there is no cause to walk over a 1/4 mile, then obviously there will be few observations of any- one doing so.55 Longer walk distances to rail services are encountered both because rail transit typically provides better service than the average bus line and because rail transit station and line spacings are further apart than those for urban bus services. (Bus rapid transit stations, such as those on the busways of Ottawa, Pittsburgh, and the San Fernando Valley Orange Line in Los Angeles, tend to be “lost” in overall bus-rider survey averages and may well actually attract walk access trips more like those to rail stations.) Actual distances traced on maps by West Coast survey respondents showed the walk distance to one Bay Area Rapid Transit (BART) station (El Cerrito, California), one LRT station in San Jose, California, and three LRT stations in Portland, Oregon, to average 0.52 miles overall. The 25th percentile was 0.27 miles, the 50th percentile (median) was 0.47 miles, and the 75th percentile was 0.68 miles (Weinstein et al., 2007). Graphs showing the percentage of BART heavy rail transit patrons choosing to walk to their sta- tion, as a function of distance, were presented in Figures 16-4 and 16-5 of the “Pedestrian/Bicycle Linkages with Transit” subsection. As noted there, somewhat more than half of BART riders living 1/2 mile from urban stations elect to walk to their station. At suburban stations, however, where sidewalk systems are less likely to be complete and direct, the proportion walking is less than half even at only 3/8 miles distance. Walk Route Choice. As already noted, walking is a slower travel mode; thus, having a direct route matters more. Pedestrians are very distance sensitive, tending to take the shortest convenient routes possible. Urban transportation modelers have had success in using intersection density as an indicator of greater propensity to walk (Lawrence Frank & Co., SACOG, and Mark Bradley Associates, 2008, Reiff and Kim, 2003), clearly because it acts as a surrogate for connectivity and corresponding ability to walk more directly to destinations. Sometimes minimum-distance paths are taken despite efforts to discourage or prohibit them. Mid-block jaywalking, diagonal crossings, walking in traffic, and unpaved short-cuts may be utilized by pedestrians seeking a direct path. At the extremes, pedestrians may scale or breech fences and directly cross high-speed facilities to avoid circuitous routings. Most pedestrians with motorized choices will, however, simply elect not to walk at all if a reasonable, safe, and secure route is not available (Zehnpfenning et al., 1993, 16-256 55 An analysis of travel activity by transportation-disadvantaged persons in an area of sparse bus route cover- age, the Hampton Roads region of Virginia, found bus use to be “fairly consistent up to a one mile distance” from the nearest bus stop for non-drivers not hampered by poor health or walker/cane use. A plot of per- cent using the bus on the survey travel day for this particular population suggests only a very slight decline with increasing distance up to the 1-mile threshold, with a sharp drop-off thereafter (Case, 2007).

AASHTO, 2001). The “Special Mini-Studies in Montgomery County, Maryland” case study pro- vides quantified examples, under “More—Sidewalk Indirectness,” of circumstances under which pedestrians have forged shortcuts. The importance of time and distance to route choice for utilitarian walking is underscored by find- ings of the surveys in the San Francisco Bay Area and Portland, previously described, that focused on transit riders walking to rail transit stations. Both open-ended and structured survey question responses showed that such pedestrians believe minimizing time and distance is their primary con- sideration. Respondents ranked 11 attributes as being very important, somewhat important, or not important in their route choice. “Very important” ranking was attached to “shortest route” by 82 per- cent of respondents, to “traffic devices are present” by 55 percent, to “traffic drives at safe speeds” by 46 percent, and to “sidewalks in good condition” by 43 percent. Assigned importance (very and somewhat important) dropped off more sharply after these four attributes. The remaining attributes were all concerned with ambiance, amenities, and people-activity measures, except for one address- ing traffic signals “where it takes a long time to cross” (Weinstein et al., 2007). These rankings would not necessarily, and probably don’t fully, apply to walking for recreation or exercise. Bicycle Trip Distance, Time, and Route Characteristics As with walking, bicycling may be chosen as a travel/exercise mode for either utilitarian or recre- ational purposes. Some motivations, notably recreation and exercise, will lead to a greater toler- ance of longer trips. Bicycling is relatively more dependent on facility improvements than is walking, especially parking provisions. Without adequate facilities, the market for bicycling may be curtailed. Bicycle Trip Speeds and Lengths. As a rule of thumb, 10 to 12 mph has been used for average bicycle speed. Global Positioning System (GPS) data for 164 Portland, Oregon, adults—primarily but not exclu- sively regular cyclists—provide a refinement. Speeds for work, work-related, and school-purpose trips were found to be 12.0 mph (mean and median), speeds for exercise and organized rides were 11.3 (mean) and 11.7 (median), speeds for social/recreation trips were 10.1 (mean) and 10.3 (median), and speeds for shopping, dining, personal business, and miscellaneous trips were 9.6 mph (mean and median). Speeds for all adult bicycle trips overall were 10.8 mph (mean and median), but with women averaging only 9.8 mph as compared to 11.6 for men (Dill and Gliebe, 2008). At 11 mph a 4-mile trip takes 22 minutes. Trip length is cited as the largest deterrent to cycling in most surveys. It is but one cycling-choice factor, but it seems to be the most recognized. Just how far is too far is a matter of debate and has yielded a range of answers. The 2009 NHTS found an average length of 2.3 miles for bicycle trips overall, but an average length of 3.8 miles for bicycle to work trips. Averages or means, in contrast to median values, may be boosted by a relatively few long trips. The median bicycle trip distance as derived from the 2009 NHTS is 1 mile. A 1981 study found that 90 percent of work trips and 84 percent of other utilitarian trips taken by bicycle were 2 miles or less in length. This is fairly consistent with a 2009 NHTS-based finding that 74 percent of all bicycle trips, including those taken for recreation and exercise, are 2 miles or less (Goldsmith, 1992, Kuzmyak et al., 2011). Other researchers have, however, derived average one-way bicycle commute trip distances of up to 6 miles. The 2002 summer survey performed by NHTSA and BTS obtained, for all bicycling trip purposes together, a 3.9 mile average trip with 57 percent less than 2 miles. Survey differences rel- ative to the NHTS included a focus on the day when bicycling most recently occurred, rather than on a fixed survey day, and use of round-trip distance for trips starting and originating at the home 16-257

without an intermediate stop (Goldsmith, 1992, NHTSA and BTS, 2002). An intercept survey in Washington State found regular bicycle commuters willing to travel slightly longer distances than occasional bicycle commuters, with neither willing to cycle for more than 1 hour each way (Niemeier, Rutherford, and Ishimaru, 1995b). Table 16-65 lists the median and mean trip lengths obtained for each reported travel purpose for bicy- cle trips tracked using GPS technology in the Portland, Oregon, research. As in the NHTSA and BTS reporting, round trip distances are given for out-and-back and “loop” exercise trips, with one-way distances for all other trips. Only adults participated, although a handful were accompanied by chil- dren (Dill and Gliebe, 2008). The high standard deviations, close to or exceeding the means, are indicative of high trip length variability. The presence of means consistently higher than the corre- sponding medians reflects distance distributions that are not normal distributions, but instead are skewed toward longer distances, inflating the means. 16-258 Table 16-65 GPS-Tracked Cycle Trip Distance by Destination Purpose in Portland, Oregon Trip Purpose Median (miles) Mean (miles) Std. Deviation Observations Exercise 8.5 12.7 13.2 94 Work 3.8 5.2 5.2 445 All Trips 2.8 4.3 5.6 1,777 Home 2.8 3.7 3.5 586 Social/Recreation 2.1 3.6 4.9 218 School (Adults) 1.8 2.8 3.1 20 Work-related 1.7 2.6 2.8 58 Shopping 1.3 2.4 4.4 117 Personal Business 1.3 2.4 2.6 142 Dining 1.0 2.0 2.3 54 Note: Round-trip miles used for out-and-back and “loop” exercise trips; one-way miles used for all other trips. Source: Dill and Gliebe (2008). The University of Minnesota “Access to Destinations” research effort introduced in the “Walk Trip Distance, Time, and Route Characteristics” discussion (see Figure 16-8) developed Twin Cities regional bicycle trip decay functions in parallel with the walk trip analysis. Illustrated in Figure 16-9, the fitted exponential decay functions relate proportion of cycling activity to distance. The figure shows cycling prevalence data points and decay function curves for work, shopping, school, and enter- tainment trips (Iacono, Krizek, and El-Geneidy, 2008). As with walking, prevalence of long bicy- cle trips is highest for entertainment, recreation, and fitness purposes. (Again, recreational/fitness trip distances were predominantly recorded in terms of round trips.) However, whereas walking trips for recreation (“entertainment”) tail off at about 6 km. (3.7 miles), some long cycling trips for recreation occur in the range of 30 to 40 km. (18.6 to 24.8 miles). Bicycle trips for work purposes, while much shorter than recreational trips overall, are similarly observed to extend 5 to 7 times further in length relative to walk trips for work purposes. Unlike the case for walk trips, there is substantial difference between the bicycle trip decay func- tions for the work and shopping utilitarian trip purposes, with the shopping trips being much shorter. School trips lie in-between. Only the shopping bicycle trips exhibit the close-in plateauing

(actually peaking) seen for both work and shopping walk trips, with the highest prevalence at roughly a 1-1/2 km. (about 1 mile) trip length. Caution must be applied in interpretations, as the cycle trip sam- ple sizes were less than 70 observations total for each purpose displayed in Figure 16-9. Nevertheless, the broader reach of bicycle trips as compared to walk trips is clearly evident (Iacono, Krizek, and El-Geneidy, 2008). 16-259 Figure 16-9 Bicycle trip distance decay plots with exponential curves. Note: ”Entertainment” includes recreation/fitness trips, predominantly recorded in terms of round-trip travel distances (Iacono, 2011, Filipi, 2011). Source: Iacono, Krizek, and El-Geneidy (2008), with curve-labeling errata resolutions per El-Geneidy (2011). The Portland, Oregon, bicycling route choice studies compared bicycle trip travel time with esti- mated auto travel times for each GPS-tracked trip. All but a handful of the bike trips took longer than driving would have, 13.4 minutes longer on average, with a median difference of 9.5 minutes. The time difference for trips of under 3 miles in length was, however, less than 5 minutes (Dill and Gliebe, 2008). Bicycle Access to Transit. The sparse data available on bicycle trips for purposes of accessing and egressing public transit service limits the certainty with which broad conclusions can be made. The one comprehensive data source encountered requires extrapolation from bike-on-bus access and egress rather than bicycle-park-and-ride activity, and is also limited to Florida locations. For dis- tance comparisons with bike-only trips, it seems reasonable that access and egress distances be summed, as has been done in several walk-transit-walk trip investigations. Such comparisons remain inconclusive, however, given lack of bike-on-bus information for travel purposes other

than the work commute, and other data issues. Bike-on-bus access trips are longer than walk access to transit trips, as demonstrated in Table 16-35 within the “Response by Type of NMT Strategy” section, under “Pedestrian/Bicycle Linkages with Transit”—“Bicycles on Transit Vehicles”— “Bike-on-Bus Programs.” Table 16-66 provides bicycle access and egress trip length distributions for three Florida bike-on-bus operations. The trip length distributions are for work-purpose trips only, thereby encompassing 72 percent of the surveyed bike-on-bus activity. “Access” pertains to the bicycle trip from home to the boarding bus stop, and “Egress” pertains to the bicycle trip from the bus stop of alighting to the workplace. The median work-purpose bike-on-bus access distance for these three Florida systems is 1 mile, and the median egress distance is 1/4 mile (Hagelin, 2005). 16-260 Table 16-66 Work-Purpose Trip Bicycle Access and Egress Distance Distributions for Three Florida Bike-on-Bus Operations Distance (miles) Hillsborough Area Reg. Tran. (N=55) Miami-Dade Transit (N=60) Pinellas Suncoast Tran. Auth. (N=47) Three-System Total (N=162) Access Egress Access Egress Access a Egress a Access Egress < 1/4 5.5% 14.5% 6.7% 18.3% 3.6% 20.0% 5.6% 18.5% 1/4 7.3 30.9 10.0 30.0 5.5 27.3 8.0 30.9 1/2, 3/4 b 16.3 34.5 28.4 33.4 20.0 18.2 22.9 30.2 1 38.2 10.9 16.7 6.7 41.8 10.9 33.3 9.9 2 23.6 7.3 18.3 5.0 12.7 7.3 19.1 6.8 3 5.5 1.8 5.0 0.0 1.8 1.8 4.3 1.2 4 1.8 0.0 3.3 0.0 0.0 0.0 1.9 0.0 5 1.8 0.0 1.7 0.0 0.0 0.0 1.2 0.0 >5 0.0 0.0 3.3 0.0 0.0 0.0 1.2 0.0 Unreported 0.0% 0.0% 6.7% 6.7% 0.0% 0.0% 2.5% 2.5% Note: a The published bicycle access/egress distance distributions for Pinellas Suncoast Transit Authority total ±85.5%, not ±100%. b Lumpiness in the self-reported Florida bike-on-bus access and egress distances resulted in less than 2 percent of respondents reporting a 3/4 mile access or egress distance. Source: Hagelin (2005). Bicycle Route Choice. Route choice for bicyclists may not be quite as distance sensitive as in the case of pedestrians, but distance and travel time are still the most important considerations when choosing a bicycle route. Greater perceived safety or even better pavement surfaces can attract cyclists to a par- ticular route, but most cyclists are found to divert very little from minimum paths. Research in the 1990s on bicyclist route choice found one-half of cyclists to use a route less than 6 percent longer than the shortest distance possible and less than 5 percent more time consuming than the quickest time route identified in the network. Journey time seemed slightly more important than distance as a choice fac- tor, but imperfect or insufficient information on the part of the cyclist may have been responsible rather than a conscious choice of time over distance. Over 70 percent of cyclists studied had selected routes that were within 10 percent of the minimum time network path (Aultman-Hall, Hall, and Baetz, 1997, C.R.O.W., 1993).

Newer GPS-and-network-based research in Portland, Oregon, shows slightly more willingness to divert from minimum-distance routings, possibly because the sample was designed to give roughly equal representation to women and to include as many less-frequent cyclists as possible.56 In this study only one-half of the GPS-tracked utilitarian trips were less than 10 percent longer than the shortest possible routing. Almost 5 percent were over 50 percent longer (Broach, Gliebe, and Dill, 2009b). Table 16-67 compares bicycle miles of travel on the shortest paths, derived from a computer network, with bicycle miles of travel on the observed paths actually used. Male cyclists (864 observed trips) and female cyclists (713 trips) are separately shown, as are frequent cyclists (1,337 trips) and infre- quent cyclists (204 trips). When the actual bicycle trip miles (i.e., bicycle miles of travel) on a facility type are proportionally less than the minimum-path bicycle trip miles allocated to that facility type, as is consistently the case with arterials lacking a bike lane, then cyclists overall are shown to be delib- erately avoiding that type of routing. When the reverse is the case, as consistently seen with low traf- fic streets, bicycle boulevards, and off-road trails, cyclists overall are shown to prefer that type of facility (Dill and Gliebe, 2008). The methodology compensates for facility orientation relative to the trips being made, providing preference indications that are relatively independent of convenience of facility location. 16-261 56 It may also be relevant that the primary 1990s North American research discussed above, done in Guelph, Ontario, Canada, was conducted on the basis of a trail system with minimal extent of hard-surfaced facilities. A majority of the trails involved were surfaced with limestone screenings, and most of the remainder were unsurfaced (Aultman-Hall, Hall, and Baetz, 1997). Portland’s primary trails are hard-surfaced and thus may be more attractive for utilitarian travel.

From Table 16-67 it may be concluded that aversion to bicycling on arterials with moderate to heavy vehicular traffic and no bicycle lanes applies regardless of gender or bicycling frequency (used here as a surrogate for skill level). Similarly universal, albeit showing some difference in strength of preference, is the propensity to use low-traffic streets and bicycle boulevards. (The neg- ligible preference for bicycle boulevards by infrequent cyclists is hard to explain given the prefer- ence for low-traffic streets, and may well be an artifact of low sample size in the applicable classifications.57) Preference for off-road shared use trails is also universal, but with a near- doubling of apparent preference for such facilities among infrequent cyclists. There is a more mod- 16-262 Table 16-67 Percentage of Utilitarian Bicycle Trip Miles by Facility Type in Portland, Oregon—Minimum-Distance Routings Versus Actual Routings Type of Bicyclist Type of Path or Statistic Arterials, No Bike Lane Low Traffic Streets Streets with Bike Lanes Bicycle Boulevards Off-Road Shared Use Trails Male Minimum 38% 31% 25% 4% 6% Actual 20% 36% 30% 8% 15% Difference -18% +5% +4% +5% +8% Female Minimum 32% 42% 22% 5% 5% Actual 15% 51% 24% 13% 12% Difference -16% +9% +2% +8% +7% Frequent Minimum 36% 35% 24% 4% 5% Actual 19% 41% 29% 11% 13% Difference -17% +6% +5% +6% +7% Infrequent Minimum 34% 33% 25% 4% 7% Actual 16% 40% 24% 6% 20% Difference -17% +7% -1% +1% +13% All Minimum 36% 36% 24% 4% 6% Actual 19% 42% 28% 10% 14% Difference -17% +6% +4% +6% +8% Notes: “Low Traffic Streets” category includes streets with bike lanes and bicycle boulevards (Dill, 2010). “Minimum” paths are the least-distance routings determined with network analysis. “Actual” paths are the routings observed with GPS tracking. “Difference” statistics are in percentage points and may not match exactly due to rounding. Percentages sum to more than 100% because bicycling on “Low Traffic Streets” with bike lanes, and also with bicycle boulevards, is included both under “Low Traffic Streets” and under the applicable bicycle-preference treatment category. Source: Dill and Gliebe, (2008). 57 Only 204 bicycle trips by infrequent cyclists were tracked, and only 4 percent of Portland’s bicycle facility mileage was composed of bicycle boulevards (Dill and Gliebe, 2008).

erate preference for bicycle lanes among most categories, with the preference among women being weak, and with a very slight aversion to bicycle lanes indicated for infrequent cyclists. Additional details on this study, along with alternative analytical perspectives on the route choice findings, are found in the “Response by Type of NMT Strategy” section (see “Bicycle Lanes and Routes”— “Popularity, Preferences, and Route Choice”—“GPS- and Network-Based Research”). The Portland GPS-tracking participants were also asked in structured questions about factors important to them in each of the route selection choices they made. They reported placing “high- est importance on minimizing distance and avoiding streets with lots of vehicle traffic.” In third and fourth place were presence of bicycle lanes and avoidance of lost time at traffic signals and signs. Comparing women to men, on the basis of both stated preference and revealed preference analysis results, the women were less likely to prefer riding on bike lanes on busy streets and more likely to prefer low-traffic streets. Responses pertaining to the small sample of trips made accom- panied by a child (87 trips by 11 participants) indicate that, with a child, avoiding “lots of” traffic had significant additional importance and that added importance was also assigned to minimiz- ing distance, riding on a path or trail, and avoiding hills (Dill and Gliebe, 2008). Some of the Portland findings concerning different bicyclist valuations of alternative facility types, depending on bicyclist characteristics, were hinted at in earlier work. Stated-preference- experiment modeling done on the basis of mid-1990s conditions and responses in Edmonton, Alberta, Canada, stratified results by cyclist comfort levels and degree of experience. Relative unattractiveness of bicycling in mixed traffic decreased with both increasing levels of comfort and experience with cycling under such conditions. Persons with the highest levels of comfort in mixed traffic were relatively indifferent to bicycle facility type. Time on bike lanes was found more attractive than time on bike paths for respondents self-reporting higher comfort and expe- rience levels. The opposite, preference for bike paths, was true for those reporting lower levels of comfort in traffic. Experience alone did not seem to much affect bike path preference (Hunt and Abraham, 2007).58 Travel Cost The user costs of walking or bicycling are relatively little. Cycling does require ownership and maintenance of a bicycle, or rental, but the costs involved are small relative to owning and oper- ating an auto. The most secure and convenient bicycle parking may entail a fee, but free bicycle parking is the norm. The primary influence of costs on the choice to walk or bike is thus the other- mode cost avoidance these NMT modes afford. Active transportation researchers and demand modelers focus primarily on the avoidable costs inherent in using competitive motorized modes, if costs are considered at all. 16-263 58 Overall values for all survey participants were reported but must be used with caution because the conve- nience sample used for the questionnaire and stated preference survey reflected a bias toward inclusion of more frequent cyclists (the survey was attached to parked bicycles or handed to cyclists passing by). Also, Edmonton at the time had 102 kilometers of shared use paths and trails but only 3 kilometers of bike lanes. Be that as it may, cycling in mixed traffic by survey participants overall was estimated to be 4.1 times as oner- ous per minute as cycling on bike lanes, while cycling with pedestrians on paths and trails was estimated to be 1.4 times as onerous per minute as cycling on bike lanes. The latter comparison was “not highly signifi- cant in a statistical sense” (Hunt and Abraham, 2007).

In addition to travel time, most multimodal travel models assign major importance to motorized trip direct user costs, especially tolls, parking charges, and transit fares. For short suburban trips, however, there are generally no tolls or parking charges that bear on auto use. In urban core areas, vehicular parking charges are more common and therefore the cost benefits of walking and bicy- cling may become a more significant part of the decision process. Studies in Portland and Eugene, Oregon, each found that large numbers of survey respondents (37 percent and nearly one-half, respectively) cited the inexpensive nature of bicycle transportation as a mode choice decision fac- tor (Goldsmith, 1992, David Evans and Associates, 1992, Lipton, 1979). Historically, researchers have tended to find that regular ownership and operating costs of auto- mobiles, even including fuel, are given less weight than other user costs in day-to-day mode choice decisions. Recent anecdotal and circumstantial evidence suggests, however, some NMT choice response to sharp gasoline price increases. U.S. retail gasoline prices rose over a 3-year period from $2.00/gal. in early 2005 to $3.00/gal. in February, 2008, and then climbed in only 4 months to $4.00/gal. in June (Energy Information Administration, 2008). Increased bicycle and accessories sales, particularly for commuting purposes, were being widely reported by early-to-mid-2008 (Emond, Tang, and Handy, 2009, Relyea, 2008). The Florida commuter rail system, Tri-Rail (Dade, Broward, and Palm Beach Counties) noted bike-on-rail increases from two to three bikes per rail car to six to seven bikes. Broward County Transit buses carried 68,000 bikes in May, nearly 6,000 more than in March, 2008 (Campbell, 2008), in a climate where approach of summer is not an expla- nation. It will be some time, though, before hard data allow quantification of the net effect on NMT mode choice of higher fuel prices counterbalanced by shifts to more fuel-efficient autos. Pricing effects on transit riding affect walking as well. Walking in connection with accessing transit presently makes up 16 percent of all U.S. walk trips (see “Related Information and Impacts”—“Extent of Walking and Bicycling”—“Extent of Walking”). Walking is also a significant alternative mode for short transit trips, such that with reduced local transit fares or improved local transit service, there is not only the increase in walking that comes with more transit access activity but also a loss of walk- ing with shifts from the walk mode to the transit mode. Transit mode shifts are discussed in Chapter 9, “Transit Scheduling and Frequency,” Chapter 10, “Bus Routing and Coverage,” and Chapter 12, “Transit Pricing and Fares.” On the other hand, if transit riding goes up because alternative motorized modes become more expensive or less attractive, then the added transit access activity will definitely effect a net increase in walking. U.S. transit riding overall increased 12 percent from 2004 to 2008. Highway VMT growth slowed starting in 2004, and then VMT dropped in 2008 to below 2004 levels (American Public Transportation Association, 2010). It seems reasonable to attribute a large mea- sure of these effects to the doubling of gasoline prices noted above.59 Trip Purpose The decision to walk or bicycle and the relative importance of the factors influencing that decision vary by trip purpose. Recreational trip makers want to get exercise and/or enjoyment from their trip and therefore may be more concerned about the environment in which they travel. Utilitarian trip makers are more interested in efficiency and other practical factors. For example, shoppers may consider ability to carry purchased goods and the variety of stores available within a reason- 16-264 59 The 2008 highway VMT reduction probably reflected the approaching financial recession as well, even though transit riding continued its strong upward climb.

able distance. Commuters are particularly concerned with distance and need to minimize travel time. Bicycle commuters have added considerations of bicycle parking, road safety, and their cloth- ing and clean-up (Zehnpfenning et al., 1993, Epperson, Hendricks, and York, 1995, Goldsmith, 1992, Pucher and Renne, 2003). Purpose-Related Effects on Route Choice. The Portland GPS-tracking participant responses con- cerning route choice factors important to them, introduced in the “Bicycle Route Choice” discus- sion, were amplified by tabulating responses separately by trip purpose. Table 16-68 presents the results in terms of average route choice factor scores. The higher the average score, based on a 1 to 5 scale, the more important the factor. The major differences in priorities among trip purposes that stand out in the tabulation include the lesser concern with minimizing distance in the case of exer- cise cycling, the greater interest in using bike lanes when making a work or (adult) school purpose trip, the lesser importance of using paths/trails for shopping and other miscellaneous business trips (paths are probably not closely aligned with commerce), the lesser concern with avoiding hills when exercising, and the elevated interest in avoiding traffic control delays when making work and school trips (Dill and Gliebe, 2008). 16-265 Table 16-68 Importance of Factors Influencing Bicycle Trip Route Choice in Portland, Oregon, Quantified as Average Scores and Arrayed by Trip Purpose Trip Purpose Work, Work- Related, School Shopping, Dining, Personal Business, Other Social/ Recrea- tional Exercise and Organized Rides Home Factor Description Minimize distance 3.8 3.6 3.2 1.6 3.6 Ride in bike lane 3.2 2.6 2.8 2.7 3.0 Ride on path/trail 2.4 1.9 2.1 2.5 2.3 Ride on signed bike route 2.8 2.3 2.6 2.3 2.7 Avoid streets with lots of traffic 3.6 3.4 3.4 3.6 3.7 Avoid hills 2.1 2.0 2.2 1.7 2.1 Reduce wait time at signs/lights 2.9 2.4 2.5 2.2 2.7 Notes: Average scores are means of individual scores on a 1 to 5 scale: 1 = Not at all important. 5 = Very important. Results cover adult cyclists only. Source: Dill and Gliebe (2008). Purpose-Influenced Relationships with Neighborhood Environment. Another proffered struc- turing of purpose-influenced relationships, this one in the context of neighborhood trip factors, employs a “proposed ecological model of neighborhood environment influence on walking and cycling.” The suggested model hypothesizes the relative effects of various neighborhood physical environment factors on utilitarian-purpose NMT trip making on the one hand and recreation/ exercise walking and bicycling activity on the other (Saelens, Sallis, and Frank, 2002): • Density, land use mix, and connectivity—strong influence on utilitarian trips, but no effect of consequence on recreation/exercise NMT activity.

• Safety from traffic dangers and crime—weak factor for utilitarian trips, but a strong influence on choice to walk or bike for exercise. • Sidewalks, paths, and bike lanes—weak influence on both utilitarian and recreation/exercise active travel decisions. • Parks and other physical activity facilities—no effect on utilitarian trips, and a weak effect on recreation/exercise NMT activity. • Aesthetics and topography—weak role in utilitarian NMT choice, but a strong role in recreation/ exercise walking and bicycling decisions. In this listing, the term “weak” is a relative one and is not intended to necessarily imply lack of sig- nificance. “Utilitarian trips” encompass both non-discretionary “transportation” such as trips to work or medical appointments, and discretionary “transportation” such as shopping or eating out. Discretionary travel may engender choice responses intermediate between those seen for non- discretionary utilitarian trips and recreational/exercise NMT activity. In any case, this represents but one set of hypotheses, and reference should be made to the “Response by Type of NMT Strategy” section and other discussions for prior and subsequent study and research findings. The “ecological model” also encompasses individual user factors. Car ownership is posited to affect only utilitarian NMT choice, but strongly so. Income, age, and gender are hypothesized to have a weak role in utilitarian NMT choice, but a strong one in choice to walk or bike for recreation and exercise. The user factors are seen to be “mediated” by the neighborhood environment. All these NMT choices, whether with respect to utilitarian trips or recreation and exercise, feed into activity level and corresponding health maintenance and disease prevention outcomes (Saelens, Sallis, and Frank, 2002). Purpose-Related Mode Choice Effects. As with automobile trips, non-work travel accounts for most pedestrian and bicycle trips. Indeed, in the case of non-motorized travel, an even greater pro- portion of trips are for non-work purposes. This is partly because the distance from home to work is fixed, and may be too long for a reasonable pedestrian or bicycle commute, while the distances to acceptable destinations for other trip purposes may be more reasonable. The higher proportion of non-work trips in the NMT travel mix also occurs because NMT non-work trips can be either utilitarian or for recreation and exercise (or both). One cannot achieve exercise through driving, but can obtain health and recreational benefits from walking or bicycling (Goldsmith, 1992). (See “Public Health Issues and Relationships” in the “Related Information and Impacts” section for more information on health benefits.) These various considerations influence the travel mode shares seen for trips of different trip pur- poses. Table 16-69 illustrates the differing NMT shares exhibited by travel when grouped into four trip purpose categories. This is U.S. national urban data derived from the 2001 NHTS by exclud- ing non-urban-area trips and trips over 75 miles in length (Pucher and Renne, 2003). Transit mode shares are tabulated along with the walk and bicycle mode shares because of the substantive walk- ing that occurs in connection with most transit travel. It is reasonable that the higher walk and bike shares for the “social and recreation” trip purpose category may reflect individual interest in the exercise benefit. 16-266

The lowest mode share in Table 16-69 is the 0.3 percent bicycle share for shopping and services trips. The impediment to bicycling of the need to carry goods pertains most directly in this case. An attitude survey in Portland, Oregon, found very high acknowledgement of the possibility that bicycles could be used to accomplish most travel needs. Nearly 88 percent of respondents indicated that use of a bicycle for a work trip would be appropriate. Similarly high responses were obtained for purposes of recreation (nearly 100 percent), school (96 percent), and most other utilitarian trip- making (83 percent). Shopping trips, however, scored much lower at 50 percent, an outcome attrib- uted to the difficulty in carrying packages (Goldsmith, 1992). In the 2009 NHTS, work trips were found to compose just 4.5 percent of all walk-only trips, or 6.2 percent if work-related business trips are included. In contrast, 29.8 percent of all walk-transit trips nationwide are work trips, or 33.4 percent including work-related business trips.60 For bicy- cling, 10.9 percent are work trips, or 12.7 percent including the related trips (Kuzmyak et al., 2011). These and other purpose distributions are provided in the “Related Information and Impacts” sec- tion (see Table 16-95 in “Characteristics of Walking and Cycling Overall”—“Trip Purposes”). The purpose aggregations inherent in some Table 16-95 purpose categories do not lend themselves to precise allocation into the non-discretionary-utilitarian, discretionary-utilitarian, and recreational/ exercise categories discussed above, but approximation is feasible: • Of surveyed 2009 walk-only trips, 1-in-3 (or less) are non-discretionary utilitarian, 1-in-3 (or more) are discretionary utilitarian, and about 1-in-3 are recreation/exercise. • Of walk-transit trips, 6-in-10 (or less, but over half) are non-discretionary utilitarian, roughly 3-in-10 are discretionary utilitarian, and 1-in-10 (or somewhat more) are recreation/exercise. 16-267 Table 16-69 Surveyed U.S. Urban Walk, Bike, and Transit Mode Shares, by Trip Purpose, 2001 NHTS Mode of Transportation Trip Purpose Work and Work Related Shopping and Services Social and Recreation School and Church All Travel Purposes a Walk-only 3.4% 6.5% 12.7% 10.5% 9.5% Bicycle 0.5 0.3 1.3 0.7 0.9 Transit b 3.7 1.4 1.0 2.2 1.7 Note: Includes only urban area trips 75 miles or less in length. a These mode shares differ from 2001 NHTS results presented elsewhere because of the restriction of travel data to urban area trips. b “Transit” excludes school buses. Transit mode shares are included as an approximate indicator for the substantive walking that occurs in connection with most transit travel. Source: Derived from 2001 NHTS by Pucher and Renne (2003). 60 Reflecting the constraints of published compilations, some mode-specific data involving transit use is pre- sented in terms of total transit use (“transit”) and some is presented in terms of “walk-transit,” i.e., walking to/from transit service for purposes of transit access and/or egress. The primary difference between these categories is that “transit,” while primarily walk-transit, includes park-and-ride and passenger-drop-off auto access trips and a small amount of bike-transit trips.

• Of bicycle trips, 1-in-4 (or less) are non-discretionary utilitarian, 1-in-4 (or more) are discre- tionary utilitarian, and almost 2-in-4 (almost half) are recreation/exercise. Purpose Characteristics of Most Recent Trip. Many national surveys that have collected trip pur- pose information, aside from the NHTS and predecessor NPTS, were shaped by a decision to ask about the last trip taken as opposed to gaining a perspective on the universe of NMT trips taken. What these surveys do reveal is that while many who walk and bicycle do so for a variety of rea- sons, there are also many who walk or bicycle infrequently and primarily for recreational pur- poses. In surveys of “most recent walk/bike trip taken,” recreation, health, and exercise appear as primary motivators for a very large portion of bicycling and walking trips, larger than would be seen in observed travel on any given individual day (Bureau of Transportation Statistics, 2003b). The 2002 national survey on pedestrian and bicyclist attitudes and behaviors was such a survey, with information based on the most recent trip in 30 days. As seen in Table 16-70, recreational and exercise trips predominate from this viewpoint, even though this particular survey counted only once each round trip from home having no stop as a destination (NHTSA and BTS, 2002). 16-268 Table 16-70 Attitudinal Survey 2002 Trip Purpose for Most Recent Walk/Bike Trip Purpose of Most Recent Trip a Percent for Walk Trips Percent for Bike Trips Commuting to school or work 5% 5% Personal errands 17 14 Visit a friend or relative 9 10 Recreation b 15 26 Exercise or health reasons b 27 24 Walk the dog/Bicycle ride 4 2 Other 12 5 To go home c 10 14 Total (all purposes) 100% 100% Note: a A focus on the “most recent trip” puts more emphasis on less frequent travel, such as for recreation, than surveys that focus on the universe of walk or bike trips on any given day. b The survey methodology counted only once each round trip from home that had no finite non-home destination, deflating the percentage of recreation and exercise trips relative to other “most recent” trips. c Many survey evaluations identify trips reported as having a “to go home” purpose with the reason for being away from home (e.g., school or work) but this one, at least in some cases, did not. Source: NHTSA and BTS (2002). Purpose-Related Trip Distance Effects. As would be expected, not only mode share and prevalence, but also average trip length varies by trip purpose for pedestrian and bicycle trips. (Table 16-95 of the “Related Information and Impacts” section, in “Characteristics of Walking and Cycling Overall”—“Trip Purposes”, provides U.S. national trip length and travel time means derived from the 2009 NHTS for various trip categories. Both walking and bicycling trips are covered.) For walk trips, major-category averages are 1.0 miles and 16.2 minutes for work purpose trips, 0.6 miles and 14.5 minutes for school and house-of-worship trips, 0.6 miles and 12.7 minutes for shopping trips,

0.5 miles and 11.2 minutes for various types of personal business trips, and 0.8 miles and approximately 20 minutes for a social, recreational, and exercise trip category (Kuzmyak et al., 2011). A breakout of recreation and exercise trips based on the 2001 NHTS serves to identify their unique nature when examined separately from “social” purpose trips. Recreational/exercise walk trips, mea- sured between the starting point and the furthest point reached (to or from), averaged 1.16 miles and 25.3 minutes in 2001 (Weinstein and Schimek, 2005). To the extent that these trips were components of “loop” or out-and-back trips, the mean recreation/exercise round trip walk mileage would measure 2.3 miles. In the case of 2009 NHTS bicycle trips, major-category averages are 3.8 miles and 21.2 minutes for work purpose trips, 1.6 miles and 15.2 minutes for school and house-of-worship trips, 1.3 miles and 14.0 minutes for shopping trips, 1.4 miles and 15.5 minutes for personal business trips, and 2.5 miles and approximately 22 minutes for the social, recreational, and exercise category (Kuzmyak et al., 2011). Miles, and also minutes, devoted to trip making for these major trip cate- gories are all higher for bicycle trips than for walk trips. The travel time differentials all lie in the range between 5 percent higher (school trips) and 38 percent higher (personal business trips). Work purpose bicycle trips are the longest in terms of one-way distance, but this might not be the case if recreation/exercise trips were separated from “social” trips and examined separately. Trip Purpose Overlap. With regard to identification of trip purpose, it is important to remain aware that most of the above discussion (and research in the field) treats utilitarian NMT travel and walk/bike activity for recreation/exercise as separate and discrete trip purposes, when in fact there is some—and perhaps much—overlap. Significant proportions of active transportation may fall into an area of purpose and motivation overlap, where the pedestrians and bicyclists involved are deliberately choosing NMT travel modes so as to obtain exercise and enjoyment in the course of accomplishing utilitarian travel. A full discussion of this circumstance and the paucity of relevant research, data, and even issue recognition is found in the “Analytical Considerations” sub- section of the “Overview and Summary” section (see “Trip Purpose Versus Motivation”). Cross- referencing to other related information within this chapter is provided there, including an exam- ple with quantification in the “Response by Type of NMT Strategy” section (see “Pedestrian/Bicycle Systems and Interconnections”—“River Bridges and Other Linkages”—“Goodwill Bridge, Brisbane, Australia”). Schoolchild Trip Factors Most of the school-purpose trip characterizations presented in the preceding “Trip Purpose” dis- cussion pertain to both child and adult school trips, including those to and from technical schools, colleges, and universities. Moreover, much school-trip-specific information is lost in aggregation of school trips with work trips. It is thus important to separately examine trip factors as they per- tain to the travel of children to and from school. The previously introduced survey of parents of children attending schools in Hillsborough County, Florida (in this section see “Behavioral Paradigms”—“The Travel Choice Making of and for Children”), offers insights on what parents find important in choosing whether or not and how their children may walk or bicycle to school. Parents were asked about trip factors and related con- ditions from two perspectives: One way the question was explored was to ask which of a list of fac- tors or conditions affected their decisions on allowing or not allowing their child to walk or bicycle to school. The other way was to ask if they would let their child walk or bike to school if the situ- ation were improved. Table 16-71 provides the results for each mode of questioning, listing first 16-269

those factors of highest priority according to the number of times they were selected. The two modes of questioning produced different prioritizations, but distance from home to school topped the list for both approaches (Zhou et al., 2009). 16-270 Table 16-71 Factors and Conditions Affecting Parents’ Decisionmaking on Allowing Their Children to Walk or Bicycle to School, Ranked by Frequency of Selection Factor/Condition Percentage of Parents Reporting Factor as an Influence on Choice Factor/Condition Percentage of Parents Allowing Walk/Bike If Condition Improved Home to school distance 67% Home to school distance 26% Traffic speed en route 54 Intersection safety 22 Traffic amount en route 51 Weather or climate 22 Violence or crime 42 Adults to chaperone 18 Intersection safety 38 Convenience of driving 15 Weather or climate 35 Sidewalks or pathways 12 Travel time 30 Extracurricular activity 12 Sidewalks or pathways 29 Crossing guards 12 Adults to chaperone 16 Travel time 10 Crossing guards 15 Traffic amount en route 10 Convenience of driving 12 Violence or crime 5 Extracurricular activity 6 Traffic speed en route 4 Source: Zhou et al. (2009). It is not clear which, if either, mode of questioning should be given the most weight. The researchers posit that the percentage of parents who would allow their child to walk or bike to school with improvement in a factor/condition produces the ranking most relevant to SRTS improvement effec- tiveness (Zhou et al., 2009). In terms of combined average ranking, distance stands much higher than any other factor. Three traffic-safety-related factors come next in importance—traffic amount en route, intersection safety, and traffic speed en route. These are followed by weather and climate, and then violence or crime, status of sidewalks or pathways, travel time, and availability of adults for chaper- oning the child. Last are crossing guards, convenience of driving, and extracurricular activity. Those three are either tied for next to last or are last, respectively, in the combined average ranking. Distance, as mentioned previously, has been found consistently (in all of the 19 studies reviewed) to have a significant negative relationship with active transportation to school, and to be the strongest predictor of the amount. One study determined that a 1-mile increase in distance between home and school decreases the likelihood of walking by 71 percent (Moudon, Stewart, and Lin, 2010). Another estimated that for each 1 percent increase in walking time there is an 0.2 percent decrease in the like- lihood of an elementary or middle school student walking. That study also undertook a descriptive analysis of 2001 NHTS data that showed 48 percent of elementary and middle school students living within 1 mile of school to be walking, as compared to 3 percent for schoolchildren beyond 1 mile. Only 20 percent of the children, however, lived within 1 mile of their school (McDonald, 2008). User Factors User-specific factors, largely related to user characteristics, help explain why different individuals facing the same environmental and trip considerations make different travel decisions. In this sub-

section, the differences among pedestrians and bicyclists across several demographic characteris- tics are presented from a global perspective and discussed in terms of effects on NMT choice. Characteristics examined include gender, age, income, automobile ownership, education, and eth- nicity. In addition, selected descriptive information on facility-specific user characteristics are pre- sented in the “Facility Usage and User Characteristics” subsection within the “Related Information and Impacts” section. Multidimensional User Characteristics Several researchers have attempted to define different multidimensional bicyclist types as a way to explain observed behavior. They argue that there is no such thing as a “typical” bicyclist. To some extent, comparable distinctions might be made about pedestrian types. One author, in a manner similar to the transit industry concept of “choice riders” and “captive rid- ers,” identifies “voluntary” and “involuntary” bicyclists as two distinct types of bicycle users. Voluntary cyclists are identified as primarily cycling for recreational purposes and as being mod- erately to extremely proficient. Involuntary cyclists are typed as not having access to a car or pub- lic transit because of age, location, or circumstance. This group is identified as being less proficient and having to ride in more varied and hazardous environments as a matter of necessity. The “bicy- cle commuter” may possess attributes from both types of riders and have a variety of experience levels and needs (Epperson, Hendricks, and York, 1995). One aspect not covered in this particular characterization is the presence of cyclists-in-training amongst recreational “choice riders.” Another research team identifies four types of bicyclists; child, youth, casual, and experienced; and defines five “stress levels” that are determined by a combination of physical infrastructure attributes and user type. The researchers found that the different types of bicyclists vary in the way they per- ceive stress levels for attributes such as curb lane volume, curb lane width, and adjacent vehicle speeds (Sorton and Walsh, 1994). Still another camp divides the cycling world into those comfortable cycling with vehicular traffic and those who much prefer to avoid it. Another team argues that cyclists are bet- ter placed on a continuum of comfort level with traffic and that propensity to use the bicycle is related to the location of the cyclist on that continuum (Aultman-Hall, Hall, and Baetz, 1997). Indeed, attitude surveys seem to confirm the preference for separated facilities among young and inexperienced riders and a preference for bike lanes and wide curb lanes by more mature and expe- rienced bicyclists (Antonakos, 1994). Bicyclist route choice studies stratified by bicyclist character- istics have obtained results supporting this pattern of preferences and add cyclists uncomfortable in traffic and/or of the female gender to those preferring off-road facilities and quiet streets (Hunt and Abraham, 2007, Dill and Gliebe, 2008). Presented within the case study “Special Mini-Studies in Montgomery County, Maryland” are “. . . Off-Street Versus On-Street NMT User Mix” observa- tions providing additional evidence that cyclist characteristics are reflected in actual route choice and also highlight the near-total selection of the off-road parallel route by walkers and joggers. The differential travel activity choices of various population groupings to factors of personal security and traffic risks are examined in the “Other Factors and Factor Combinations” subsection under “Security and Safety.” The possibility of making “voluntary” versus “involuntary” distinctions for pedestrians, to use the approach of the first-listed bicyclist characterizations above, is suggested by survey and analysis find- ings from Seattle-area pedestrian activity study observations already introduced. Six “urban” and six “suburban” similarly sized neighborhoods, each with its own commercial area, were compared. The “suburban” neighborhoods, with poor sidewalk systems, had one-third the measured pedestrian 16-271

activity per resident of the “urban” neighborhoods. Walkers likely to represent “involuntary” pedes- trians were markedly over-represented in the suburban neighborhoods and were forced to contend with the deficient infrastructure. In the “urban” neighborhoods, with better walking environments, the greater walking activity along with lesser observed over-representation of walkers likely to be autoless suggests larger numbers of “voluntary” pedestrians (Hess et al., 1998, Moudon et al., 1997). The various typologies introduced here should be kept in mind as findings along the single- attribute dimensions of gender, age, income, and automobile ownership are reviewed. The Seattle- area comparative pedestrian research is summarized more comprehensively in the case study, “Pedestrian Activity Effects of Neighborhood Site Design—Seattle.” Gender Men and women walk at similar, though not exactly, the same rates. A summer 2002 national tele- phone survey found 78 percent of men and 79 percent of women reporting having walked, run, or jogged outdoors for 5 minutes or more in the past 30 days (NHTSA and BTS, 2002). The 2007 ACS indi- cates that men tend to walk to work more than women, but not by a large margin. Walkers to work were 54 percent men and 46 percent women (Alliance for Biking & Walking, 2010). Results from three pairs of multivariate research model derivations based on 2001 NHTS daily trip diary data suggest that men and women have similar propensities to walk to work, but that men are 13 percent less likely than women to walk for recreation or exercise, other considerations being equal. The survey data exhibited no difference in average trip distance between the sexes (Agrawal and Schimek, 2007). For bicycling, on the other hand, there tends to be a substantial difference between the trip mak- ing of males and females. In a large majority of surveys, male riders outnumber female riders, and for work trips the difference is particularly substantial (Goldsmith, 1992). In the summer 2002 national telephone survey, 34 percent of males rode a bicycle in the previous 30 days versus 21 per- cent of females (NHTSA and BTS, 2002). There is no evident leveling out over time. The 1990 com- muter disparity was reported as 75 percent male versus 25 percent female (David Evans and Associates, 1992), while the 2007 ACS indicates the commuter comparison to be 77 percent male versus 23 percent female (Alliance for Biking & Walking, 2010). Some research has suggested that travel times tend to be similar among men and women (Shafizadeh and Niemeier, 1997), but Portland, Oregon, GPS-tracked cycling distance was significantly lower for female participants (5.0 miles/day average) than for men (7.2 miles). The number of daily bicycle trips was similar for both sexes at 1.6 for females and 1.5 for males, but as previously noted, average bicycling speeds were 9.8 mph for women versus 11.6 mph for men (Dill and Gliebe, 2008). Examination of mode shares for men versus women tells a similar story. The 2001 NHTS data, including all purposes of travel, produced walk-only-plus-bike mode share totals of 10.6 percent for men and 10.5 percent for women. The modal breakdown differed, however. Bicycle shares were 1.2 percent for males and 0.5 percent for females, with corresponding walk-only shares of 9.3 per- cent for men and 9.9 percent for women (Pucher and Renne, 2003). One explanation offered for the markedly lesser amount of cycling by women in the United States is their greater need to under- take travel for household and family support activities, which in turn requires more transporting of goods and passengers. Another explanation is that women on average have a different percep- tion of safety than men (Emond, Tang, and Handy, 2009). The safety issue is examined later on under “Other Factors and Factor Combinations”—“Security and Safety.” Public transit shares were 1.7 percent for males and 1.8 percent for females in 2001 (Pucher and Renne, 2003). Only a slightly greater tendency for women to walk in connection with transit use 16-272

can be inferred from these mode shares. A Centers for Disease Control and Prevention (CDC) analysis of the 2001 NHTS to learn about walking for public transit access did not calculate the odds of making transit access walk trips per se. It did, however, examine the likelihood that peo- ple are meeting the Surgeon General’s recommendation for 30 minutes of physical activity per day simply by walking to and from transit. This research found a 21 to 23 percent greater propensity for women to obtain 30 minutes or more a day of walking activity by riding transit than for men to do so (Besser and Dannenberg, 2005). Cross-classification analysis of the 2009 NHTS by gender and age, made in terms of mode share percentages, produces results in general conformity with the gender tendencies identified in over- all statistics. The analysis is displayed in Table 16-72. The mode share for women choosing to walk is slightly more than for men in the age categories above age 24, although not at younger ages. Adult males choosing to bicycle outnumber females doing the same by 3-to-1 or more (5-to-1 for seniors). The share of adult females walking to/from transit is larger than the share of males doing so in most age groupings. 16-273 Table 16-72 Surveyed U.S. National Walk, Bike, and Walk-Transit Mode Shares for All Travel Purposes Combined, by Gender and Age Cross-Classified, 2009 NHTS Mode of Transportation Age Category Gender 5-15 16-24 25-34 35-44 45-54 55-64 >65 Walk-only Male 14.2% 10.9% 11.6% 9.2% 9.8% 9.7% 8.5% Female 12.4 8.8 13.3 10.1 9.8 10.0 9.1 All 13.3 9.8 12.5 9.6 9.8 9.9 8.8 Bicycle Male 4.3 1.4 1.1 1.3 1.2 1.0 1.0 Female 1.8 0.4 0.3 0.3 0.4 0.3 0.2 All 3.1 0.9 0.7 0.8 0.8 0.6 0.6 Walk to/from Male 1.1 2.1 1.9 1.4 1.3 1.1 1.2 Transit Female 0.9 3.0 2.4 1.3 1.6 1.3 1.1 All 1.0 2.5 2.1 1.4 1.5 1.2 1.1 Source: Derived from 2009 NHTS by Kuzmyak et al. (2011), with clarifications per communications of December 15, 2011. Children, discussed further under “Age,” exhibit a different pattern. Male children have higher mode shares than their female counterparts in each of the three primary NMT categories: walk- only, bike, and walk to/from transit (see Table 16-72 for derivation citation). While males clearly bicycle more than females, it appears that women more or less make up the difference by walking. The percentage differences between sexes are larger for bicycling, but the smaller counterbalancing percentage differences for walking apply to a much larger segment of NMT trips overall. Confirmation is provided by the 2001 NHTS analysis of walk-only-and-bicycle trips reported previously (Pucher and Renne, 2003) and the 2009 NHTS finding (derivable from Table 16-72) that in five out of seven age categories, daily total walk and bike shares for women exceed the total for men (Kuzmyak et al., 2011).

Age Adult walking and especially bicycling become less prevalent with age in most but not all countries. This trend is especially pronounced in the United States. The summer 2002 national telephone survey, as noted above under “Gender,” asked respondents over age 16 whether they had walked or bicycled out- doors for 5 minutes or more in the previous 30 days. The 2009 NHTS asked if the trip maker (age 5 or above) had walked or bicycled in the past week. The results, stratified by age group, are summarized in Table 16-73, below. While the decline with age for walking is modest, the decline for bicycling is far more substantial and—among adults—starts earlier (NHTSA and BTS, 2002, Kuzmyak et al., 2011). 16-274 Table 16-73 Percentage Having Walked or Bicycled in the Past 30 Days and in the Past Week by Age Group Walked in the Past… Bicycled in the Past… Age Group 30 Days Week 30 Days Week 5-15 n/a 76% n/a 40% 16-24 82% 68 39% 10 25-34 82 70 33 9 35-44 82 68 34 10 45-54 80 68 26 7 55-64 76 66 18 6 65+ 66 55 9 3 Note: “Ran or jogged” included with “walked” for 30-day data; not separately identified in “past week” data. Sources: Past 30 days – NHTSA and BTS (2002). Past week – derived from 2009 NHTS by Kuzmyak et al. (2011). These results parallel the findings for bicycling of a 1991 Harris Poll and other surveys. The Harris Poll distinguished between general bicycling and commuter bicycling. It found that the drop off with age in commuter cycling was the sharpest of all. Very little cycling to and from work was seen after age 40 (Goldsmith, 1992). The drop-off in overall bicycling becomes even more dramatic when the 5 to 15 age group is included in the trend analysis, as in the mode share data of Table 16-72 and the “past week” data in Table 16-73. Looked at from the opposite perspective, the incidence of child bicycling is extra- ordinary. NMT-only travel by children aged 5 through 15 is disproportionately high for both walk- ing and bicycling, as would be expected for an age group with no auto drivers. The prevalence of daily walking in this age group is almost 1/3 higher than the average for older age groups, and the prevalence of daily bicycling is over 4 times as high. Walking and bicycling differentials for chil- dren versus adults derived on the basis of numbers of trips rather than indirect comparisons are provided toward the end of this “Age” discussion. Adult bicycling shares are highest during the age 16 through 24 transition into adulthood, as seen in Table 16-72, although they are nearly down to typical adult levels. Walking to access transit ser- vice actually peaks during the ages of 16 through 24 years, with higher shares than at younger or older ages, likely reflecting greater use of transit by young adults just entering the work force. Walk-only mode shares for adults peak later, at ages 25 through 34, and then stabilize at percent- ages established in young adulthood (but with some gender differences). In this 2009 dataset they decline slightly after age 65 (see Table 16-72 for derivation citation).

NMT mode share data and walking and bicycling activity data exhibit trends that are similar but not identical to each other. Table 16-74 presents 2001 NMT-related mode share data for five age groups. Walk shares over age 65 increase relative to ages 40 through 64 in this 2001 data, return- ing almost to younger-adult levels. An all-travel-purpose decline in bicycle mode shares at age 40 is apparent in the NHTS data. The bicycle share decline is not as steep, however, as seen for activ- ity percentages such as those in Table 16-73. A major part of the differences seen, although survey methodology variance undoubtedly plays a role, is almost certainly the drop-off in absolute num- bers of trips—irrespective of travel mode—as age increases beyond age 64. Trip making by any means of transportation has been shown by the 2001 NHTS to decline fairly steadily from 4.4 trips per day per person at ages 25 through 64 down to 1.9 trips per day per person at ages 85 and above (Pucher and Renne, 2003). Some of the declines with older age seen in Table 16-73 thus may be more related to reductions in trip-making than mode shifts. 16-275 Table 16-74 Surveyed U.S. Urban Walk, Bike, and Transit Mode Shares for All Travel Purposes Combined, by Age Group, 2001 NHTS Mode of Transportation Age 5 to 15 16 to 24 25 to 39 40 to 64 65 & over All Ages a Walk-only 15.2% 9.3% 9.2% 7.8% 8.9% 9.5% Bicycle 3.2 0.6 0.6 0.4 0.4 0.9 Transit b 1.1 2.9 2.1 1.5 1.3 1.7 Note: Includes only urban area trips 75 miles or less in length. a These mode shares differ from 2001 NHTS results presented elsewhere because of the restriction of travel data to urban area trips. b “Transit” excludes school buses. Transit mode shares are included as an approximate indicator for the substantive walking that occurs in connection with most transit travel. Source: Derived from 2001 NHTS by Pucher and Renne (2003). There are striking differences between these walking and cycling trends, as people age in the United States, compared to the more age-resilient walking and cycling experience of certain European coun- tries such as Germany and the Netherlands. These differences are highlighted and interpreted in the “Response by Type of NMT Strategy” section under “NMT Policies and Programs”—“European Programs and Comparisons.” Descriptive statistical analysis of 2001 NHTS daily trip diary data found that the primary variation in walking among age groups was in the proportion of survey respondents in each group who reported making walk trips at all. Among those who walked to any degree, there was much less variation in mean number per day, mean duration, or mean distance of such trips. Two out of three pairs of multivariate model derivations from the data, the two that exclude a confounding auto ownership variable, indicated that persons under age 18 are 45 to 47 percent more likely than non- senior adults to do utilitarian walking and 12 percent more likely to do recreational/exercise walk- ing. The same pairs of model derivations indicated that persons over age 64 are about 25 percent less likely to do utilitarian walking but 39 percent more likely to do recreation/exercise walking. These comparisons are based on odds-ratio estimates that take into account other factors bearing on travel choices, such as family income and housing density. Comparable odds were not deter- mined for public transit access trips, but descriptive statistics indicate that seniors made substan- tially fewer walk trips to and from transit than non-senior adults (Agrawal and Schimek, 2007).

As this information and Tables 16-72, 16-73, and 16-74 suggest, there are many child pedestrians and bicyclists. The NHTS, with its enhanced methodology for drawing out information on walk trips, found children from ages 5 through 15 to be taking 28 percent of all walk trips and 58 percent of all bicycle trips in 2001. This age range accounted for just 24 percent of the population. Conversely, with 15 percent of the population, adults over age 65 were taking only 9 percent of all walk trips and 4 percent of all bike trips (Alliance for Biking & Walking, 2010). These aggregate childhood travel data obscure progressive shifts that take place with each addi- tional year of age, from ages 5 through 15. Mode choice modeling of child travel to school, accom- plished with 2001 NHTS data, suggests that each additional year in age—through elementary and middle school—is associated with a 1.4 percent lower auto passenger mode share and an 0.4 per- cent higher walk share (McDonald, 2008). This effect is presumably a reflection of increasing matu- rity and thus independence from parental chauffeuring. (For more on this research, refer back to “Behavioral Paradigms”—“The Travel Choice Making of and for Children.”) Income The strongest relationship at the aggregate level between walking and income is encountered when examining the work commute. Among the 50 states, from over 30 to over 60 percent of people who reported walking to and from work in the 2005 ACS earned less than $15,000 per year (Alliance for Biking & Walking, 2010). This aggregate finding may, however, have as much to do with the location of neighborhoods where lower-income households dominate as with the predilections of lower or higher income commuters. The NHTS provides the most detailed information on propensity to walk or bicycle for all travel purposes at differing household income levels. Table 16-75 provides all-travel-purpose NMT- related mode shares from the 2009 NHTS at five different income levels. A notable decline in mode shares for both walk-only and walk to/from transit trips may be observed between the lowest income level and the other income levels. No consistent income effect is discernible for bicycle travel, however, at this level of cross-classification (Kuzmyak et al., 2011). 16-276 Table 16-75 Surveyed U.S. National Walk, Bike, and Walk-Transit Mode Shares for All Travel Purposes Combined by Household Income Class, 2009 NHTS Mode of Transportation Household Income Less than $20,000 $20,000 to $39,999 $40,000 to $74,999 $75,000 to $99,999 $100,000 and over Walk-only 16.9% 10.3% 8.9% 8.9% 10.1% Bicycle 1.1 1.3 1.1 0.9 1.1 Walk-Transit a 4.8 2.1 1.1 0.7 0.7 Note: All NMT trips, both urban and rural and of any length, are included in this tabulation. a “Walk-Transit” includes only those transit trips involving walking to/from the transit stop or station. Source: Derived from 2009 NHTS by Kuzmyak et al. (2011).

The Table 16-75 data cover all 2009 U.S. NMT travel. Investigation of 2001 data excluding rural trips showed similar relationships but with mode shares in the $20,000 to $39,999 category more like those in higher income categories (Pucher and Renne, 2003, Kuzmyak et al., 2011). While the data are not quite comparable, one may speculate that gasoline price increases and other economic pressures dur- ing the decade most affected the travel decisions of persons living close to but not in poverty, giving more impetus than before to choice of walking and walk-transit use as compared to driving. Income, of course, is a major determinant of auto ownership. Ownership of a car “dramatically trans- forms travel behavior,” including mode share, for both walking and bicycling (Pucher and Renne, 2003). (This effect can be clearly seen in Table 16-78, discussed under “Automobile Ownership.”) The combination of mode shares in each mode/income category and overall number of trips made by persons in each income stratum gives the number of persons making trips in each mode/income category. The lower overall per-person trip generation rates of lower-income persons affect the outcome, dampening the number of trips for each mode in the lower income categories (Pucher and Renne, 2003). Table 16-76 shows the resulting percentage distribution of trips among income groups for each NMT-related mode from the 2009 NHTS. 16-277 Table 16-76 Surveyed U.S. National Household Income Distribution of Walkers, Bicyclists, and Transit Users, 2009 NHTS Mode of Transportation Household Income Less than $20,000 $20,000 to $39,999 $40,000 to $74,999 $75,000 to $99,999 $100,000 and over Walk-only 21.8% 19.2% 23.0% 13.6% 22.6% Bicycle 13.8 21.5 26.8 13.4 24.3 Walk-Transit a 39.8 24.3 17.7 7.4 10.8 Note: This table gives the percentage composition of each mode’s users by household income class, thus, each row totals to 100%. Income was not reported by 7% of households. Percentages were normalized accordingly. All NMT trips, both urban and rural and of any length, are included. Of persons and households, 16% to 20% are in the less than $20,000 category, about 20% are in the $20,000 to $39,999 category, about 25% are in the $40,000 to $74,999 category, 12% to 14% are in the $75,000 to $99,999 category, and 15% to 20% are in the $100,000 and over category. a “Walk-Transit” includes only those transit trips involving walking to/from the transit stop or station. Source: Derived from 2009 NHTS by Kuzmyak et al. (2011). A variety of studies relying on aggregate data have, overall, come to mixed conclusions about the relationship between income and bicycling. The U.S. Bureau of Transportation Statistics found in 2002 that 58 percent of persons who bicycle earn $50,000 or more in income. The BTS, in making this determination, did not stratify by trip purpose and allowed all levels of usage to qualify. Thus, the findings are overweighted by occasional, likely recreational, cyclists. Although the same over- all finding of cyclists having higher income was noted for persons reporting 6 or more days per month of bicycling, the difference was smaller (51 percent earning more than $50,000 versus 49 per- cent earning less) (Bureau of Transportation Statistics, 2002). Similar information from the 2009 NHTS shows a steady but small increase with higher income in percentage of survey respondents bicycling within the previous week, ranging from 10 percent in

the less than $20,000 category to 16 percent in the $100,000 and over category. Walking results are similar, though with higher percentages, starting with 67 percent in the less than $20,000 category and then ranging from 65 percent in the $20,000 to $39,999 category up to 72 percent in the $100,000 and over category (Kuzmyak et al., 2011). Tabulations from both the 2001 NHTS and 2009 NHTS, the latter as seen in Table 16-75, suggest that there is little variation—or at least no consistent variation—in bicycle mode share among income groups when all travel purposes are considered in the aggregate (Pucher and Renne, 2003, Kuzmyak et al., 2011). It has been suggested that further analysis might show more bicycling for utilitarian purposes among lower income groups and more cycling for recreation among higher income groups (Alliance for Biking & Walking, 2010). Evidence presented following Table 16-77 indicates that such relationships do pertain for walking. Studies focused exclusively on bicycle commuting have also tended to produce mixed or inconclusive results vis-à-vis income. An analysis of Orange County, California, survey data found no correlation between income and bicycle commuting. Other research found that higher income survey respondents tended to report longer bicycle commuting travel times than other respondents. That analysis also con- cluded that commuting travel time decreased as income increased for suburb-to-suburb commuters but identified the opposite relationship for suburb-to-CBD commuters (Goldsmith, 1992, NuStats International, 1998, Shafizadeh and Niemeier, 1997). Commuting cyclists are more heavily represented in the employment categories of sales, clerical, service, and laborer than in professional or technical positions (MacLachlan and Badgett, 1995). Table 16-77 presents results from a 1991 Harris Poll with respect to prevalence of cycling-to-work activity within each of six income strata (Goldsmith, 1992). The income classifications are dated, as a result of inflation, but the distribution of bicycle shares from low to high income clearly sup- ports a finding that cycling to and from work in 1991 was more prevalent in the lowest income cat- egories. This finding, presuming it is still valid, is consistent with the suggestion that lower income groups undertake more bicycling for utilitarian purposes. It does not address income relationships with incidence of bicycling for enjoyment and exercise. 16-278 Table 16-77 Percentage Commuting by Bicycle in Previous Month by Income, 1991 Income Percentage $7,500 or less 23% $7,501 - $15,000 14 $15,001 - $25,000 6 $25,001 - $35,000 7 $35,001 - $50,000 1 $50,001 and Over 7 Note: Income as of 1991, in 1991 dollars. Source: Harris Poll as reported in Goldsmith (1992). Model derivations from 2001 NHTS daily trip diary data (the three previously mentioned pairs of models) add some clarity in the case of walking. The two model pairs without a confounding auto ownership variable indicate a steady decline in the likelihood of utilitarian walking as household

incomes increase from under $15,000 to $30,000 dollars. From that threshold on, persons with more income are roughly 40 percent less likely to make utilitarian walk trips than persons with less than $15,000 in family income. In contrast, the likelihood of recreation/exercise walking increases steadily with income once the $30,000 dollar threshold is reached, according to all three models, until at over $80,000 family income such trips are 22 to 25 percent more likely (Agrawal and Schimek, 2007). The I-PLACE3S travel demand and health modeling effort in King County, Washington, found, rel- ative to middle-income households, that Seattle area households with incomes under $50,000 had slightly fewer walk-only/bike-only trips and miles. Households with incomes over $100,000 had slightly more. Given that the I-PLACE3S model estimated absolute numbers of trips/miles rather than mode shares, this outcome likely reflects—at least in part—the higher trip activity that comes with higher income (Lawrence Frank & Co., SACOG, and Mark Bradley Associates, 2008) and pos- sibly also a greater propensity to maintain a recreational/exercise walking or cycling regimen. On the other hand, the amount of walking seen in lower-income households would certainly increase if the model set were structured to identify the walking that occurs when accessing most transit trips. CDC analysis of the 2001 NHTS indeed found likelihood of meeting recommended physical activ- ity guidelines solely by walking to and from transit to be inversely related to income. A person with family income in excess of $70,000 was determined to be less than half as likely to walk 30 minutes a day for transit access than persons with family incomes below $15,000. Mean daily walk times in connection with transit use progressed upward from 20 minutes for the highest income category to 29 minutes for the lowest (Besser and Dannenberg, 2005). This distance varia- tion appears to provide roughly half the explanation for the income-related transit access walking activity difference, with the remainder presumably being accounted for by higher incidence of transit use by lower income persons. Automobile Ownership Lower automobile ownership (fewer operating vehicles per household, or fewer vehicles than adults or licensed drivers) is correlated with additional pedestrian and bicycle trip making. The phenomenon pertains whether NMT activity is expressed in mode share or absolute numbers of trips. Walk and bicycle mode shares for persons in zero-car households that are triple the equiva- lent shares in one-car households suggest that bare-bones necessity plays a major NMT choice role when there is no vehicle. Lesser automobile availability in the household may also diminish the “initial barrier” to contemplating a pedestrian or bicycle trip. Alternatively, it may be that some such households have chosen to forgo purchasing an automobile because they live in a dense area, want to save on the expense of automobile ownership, or favor pedestrian or bicycle travel, and are in a good position to meet daily needs by walking or cycling. For example, Portland, Oregon planners have found automobile ownership to be significantly related to the pedestrian environ- ment (Goldsmith, 1992, Pucher and Renne, 2003, Lawrence Frank & Co., SACOG, and Mark Bradley Associates, 2008). The investigators for the 2010 Benchmarking Project covering U.S. bicycling and walking stress that “the causation might run in both directions.” They find, in any case, that lower levels of auto ownership are strongly related to higher levels of walking and cycling to and from work (r = 0.81) based on the 2007 ACS (Alliance for Biking & Walking, 2010). Table 16-78 provides NMT-related mode shares associated with different auto ownership levels, derived for urban trips from the 2001 NHTS. The precipitous drop in these mode shares between 16-279

zero and one car households, compared to lesser differences among vehicle ownership levels once household auto availability is established, appears to serve as a measure of necessity of walking, cycling, or walking to access transit (Pucher and Renne, 2003). Equity implications are discussed in the “Economic and Equity Impacts” subsection under “Related Information and Impacts.” Results from the 2009 NHTS (including rural trips, and trips of any length, but excluding non- walk-access transit trips) are virtually the same in round numbers, except that walk-only shares in the two and three-or-more vehicles categories are higher by about 1 percentage point (Kuzmyak et al., 2011). 16-280 Table 16-78 Surveyed U.S. Urban Walk, Bike, and Transit Mode Shares for All Travel Purposes Combined by Number of Vehicles in the Household, 2001 NHTS Mode of Transportation Total Number of Vehicles in the Household 0 1 2 3 or more Walk-only 41.1% 12.5% 7.8% 6.3% Bicycle 2.4 0.7 0.9 0.8 Transit a 19.1 2.7 0.6 0.5 Note: Includes only urban area trips 75 miles or less in length. a “Transit” excludes school buses. Transit mode shares are included as an approximate indicator for the substantive walking that occurs in connection with most transit travel. Source: Derived from 2001 NHTS by Pucher and Renne (2003). One pair of the previously discussed pairs of research model derivations, from 2001 NHTS data, contained an auto ownership variable. The relevant derivations indicated that utilitarian walking is over three times as likely for a person from a zero-car household, as compared to households with one car per driver. (This ratio is similar to relationships obtained with descriptive analyses.) Other differences in auto ownership levels did not produce clear cut model outcomes. The propen- sity for recreation/exercise walking increased slightly but not significantly with higher auto own- ership, supporting a hypothesis “that driving is not the major substitute for recreational walking” (Agrawal and Schimek, 2007). Owning no cars is a stronger influence on walking and bicycling than simply having fewer cars than drivers in a household. Data from the 2009 NHTS show that active transportation (walk, bike, walk to/from transit) accounts for 8 percent of all trips by persons in households with either three or more vehicles or more vehicles than drivers. However, active transportation is used for 63 percent of all trips by persons in zero-car households, versus 17 percent of all trips by persons in households with fewer vehicles than drivers (Kuzmyak et al., 2011). Looking at such relationships from a nearly oppo- site perspective, a circa 1980 study in Santa Barbara, California, found that while over 90 percent of the general population owned at least one car, along with 80 to 85 percent of general transit users, the corresponding rate for bicycle users was 70 to 75 percent owning at least one car (Newman and Bebendorf, 1983). Descriptive analysis of walk-transit trips, with statistics drawn from the 2001 NHTS, showed that the average person in a zero-car household is 14 times as likely to make a walk-transit trip on any given day as a person in an auto-owning household. Moreover, each zero-car household walk- transit trip-maker takes such trips at a 17 percent higher frequency during the day and spends

19 percent more time walking per walk-transit trip (Agrawal and Schimek, 2007). Persons from zero-car households were almost 50 percent to over 100 percent more likely to obtain 30 minutes of physical activity per day, solely by walking to and from transit, than either the primary driver or non-primary drivers in auto-owning households (Besser and Dannenberg, 2005). Education The 2009 NHTS data presented in Table 16-79 illustrates that both walking and bicycling are most common among the least educated and persons with a college or graduate degree (Kuzmyak et al., 2011). Similar to equivalent mode share versus income cross-classifications, Table 16-79 leaves open the question of whether or not there might be a different pattern for utilitarian trips as com- pared to recreational/exercise trips, such as the intuitively satisfying but possibility dated suppo- sition of higher utilitarian walking and cycling among those with the least education balanced in large measure by higher recreational/exercise walking and cycling in households with more edu- cation. Walking in connection with transit use follows a somewhat different pattern, with by far the highest prevalence among the lesser educated and continued decline with more education until a graduate or professional degree is obtained, at which education level walk-transit use rises. 16-281 Table 16-79 Surveyed U.S. National Walk, Bike, and Transit Mode Shares for All Travel Purposes Combined by Education Level of Household Head, 2009 NHTS Mode of Transportation Highest Educational Level Attained by Household Head Less Than High School Graduate High School Graduate or GED Some College or Vocational/ Associate Degree Bachelor’s Degree (BA, AB, BS) Graduate or Professional Degree Walk-only 13.3% 8.7% 8.6% 10.2% 12.5% Bicycle 1.1 0.5 0.5 1.0 1.1 Walk-Transit a 3.4 1.9 1.3 1.0 1.4 Note: All NMT trips, both urban and rural and of any length, are included in this tabulation. a “Walk-Transit” includes only those transit trips involving walking to/from the transit stop or station. Source: Derived from 2009 NHTS by Kuzmyak et al. (2011). Analyses of 2001 NHTS data using multivariate modeling provided statistical insights into the rela- tionships between the odds of walking and measures of education and ethnicity. The educational level findings did not exactly parallel those for income. Aside from minor inconsistencies, the higher the education and with other factors controlled for, the more likely a person is to choose utilitarian walk- ing. The relationship is similar and even clearer for recreational/exercise walking. For both cate- gories of walk-only trips, a person with a graduate degree is twice as likely to choose to walk on any given day as a person with only a high school diploma. Making particular reference to recreational/ exercise walking, the researchers conclude that: “Although education and income are generally highly correlated, it is clear from these results that educational attainment is a much more important factor than income in determining the odds of walking” (Agrawal and Schimek, 2007). Such findings imply that it is other factors, such as larger proportions of zero-car households or high NMT acces- sibility characteristics of low-income neighborhoods, that result in larger NMT mode shares for the least educated.

In contrast, bivariate analysis suggests that as education level increases, the likelihood decreases of walking 30 minutes or more a day in connection with using public transit. The estimated likeli- hood was half for graduate degree holders when compared to those without a high school diploma. Multivariate analysis, however, controlling simultaneously for other variables, estimated equal propensity for such walk-transit activity by persons with less than a high school diploma and persons with a graduate degree. A lesser propensity was estimated for persons with interme- diate levels of education (Besser and Dannenberg, 2005). Ethnicity The effects of ethnicity on walking are complex, as illustrated by NHTS results. Moreover, observed rates of walking are likely to be influenced by the housing patterns of minorities, includ- ing concentration in inner-city areas and older suburbs more likely to feature sidewalks, higher densities, neighborhood shopping, neighborhood schools, and more transit stops and service. An illustrative example is provided by descriptive and multivariate analysis of choice of mode for school access by elementary and middle school children. Descriptive analysis of 2001 NHTS results indicates that while 10 percent of white children walk to school, 22 percent of African American children do so. However, when included in a multivariate analysis using a multinomial logit model with variables such as auto and walk travel times, population density, income, and vehicles per driver (all in addition to race), the race and ethnicity variables make little explanatory contri- bution (McDonald, 2008). For all age groups taken together, descriptive statistics for 2001 showed all minorities, and partic- ularly African Americans, to be engaging in both utilitarian-trip walking and walking in connec- tion with transit trips more often than non-Hispanic whites. Whites and Asians, on the other hand, were shown to be engaging in recreational/exercise walking more than other groups (Agrawal and Schimek, 2007). Aggregate data for 2009 indicate bicycling is most common amongst whites, fol- lowed closely by African Americans (Kuzmyak et al., 2011). When ethnicity was included in multivariate models of 2001 walk-only trip making, however, all minorities were estimated to have lower propensities for both utilitarian walking and recreational/ exercise walking than non-Hispanic whites. In other words, it was estimated that if faced with equiv- alent transportation and income trade-offs, minority groups inclusive of all ages would walk less. The researchers concluded that the discrepancy in the case of utilitarian trips reveals that factors like lower car ownership, lesser income, and residential area characteristics are what leads to the observed higher incidence of utilitarian walking by minorities. They also speculate that what appears to be lower propensities to walk could, to the extent that minorities are segregated into less safe neighborhoods, be a reaction to neighborhood crime (Agrawal and Schimek, 2007). Walking in connection with transit trips is, in any case, strongly associated with minority status by any available measure (Kuzmyak et al., 2011). For the most part, either in bivariate or multivari- ate analysis, the individual minority groups are estimated to be two to three times more likely to engage in 30 minutes or more a day of transit-associated walking than non-Hispanic whites (Besser and Dannenberg, 2005). The 2002 national survey of pedestrian and bicyclist attitudes and behaviors found non-Hispanic blacks least likely to have walked for any reason during the last 30 days (75 percent walked com- pared to roughly 79 percent for other groups). The same was true for bicycling, but there was also more variability among all groups investigated (non-Hispanic whites, 28 percent cycled; non- 16-282

Hispanic blacks, 22 percent; non-Hispanic other, 25 percent; Hispanic, 29 percent) (NHTSA and BTS, 2002). As previously highlighted, this type of survey overweights the sample of recreational NMT trip making. In terms of the journey to work, as measured in the 2007 ACS, there appear to be few notable differentials by ethnicity. The largest were observed in walking to work by persons of Asian ethnicity (7 percent of walk commutes versus 4 percent of the U.S. population) and in bicy- cling to work by persons of Hispanic ethnicity (22 percent of bicycle commutes versus 15 percent of the population). None of the other similarly measured differentials exceeded 1 percent except for non-Hispanic whites, where the larger percentages involved showed this group taking 64 per- cent of the walk commutes and 61 percent of the bicycle commutes compared to representing 66 percent of the surveyed population (Alliance for Biking & Walking, 2010). Other Factors and Factor Combinations Other separate factors and factor combinations influence the choice to walk or bicycle, but do not fit neatly into a typology limited to environmental, trip, and user factors. This subsection looks at four: security and safety, university affiliation, factor combinations involving trip purpose, and attitudes and modal biases. Security and Safety Safety is a potentially significant but poorly understood travel choice factor for both pedestrian and bicycle trips. Concerns about personal safety from crime and street traffic safety are believed to impose some degree of deterrence on the use of non-motorized modes. Two commissioned litera- ture reviews in 2004, however, failed to find evidence of any strong correlation between safety and NMT travel choice. Several possible reasons were suggested, primarily related to limitations in study design (Committee on Physical Activity, Health, Transportation, and Land Use, 2005). Issues of safety and security not only may affect choices adults make concerning their own travel, they also fairly obviously impact the decisions parents and guardians make about how their children should travel to and from school and around their neighborhood. Among pedestrian environment factors, concerns about safety and security have been identified in at least one study as outranking comfort, convenience, attractiveness, system coherence, and system continuity combined (Khisty, 1994). Other factor ranking findings are provided below. For bicyclists, issues of both traffic safety and bicycle theft are important forms of disutility (Everett, 1990). In an apparent contradiction, safety-related attitudinal questions asked in the Nonmotorized Transportation Pilot Program Evaluation Study found concern about crime affecting daytime walking to be least prevalent of any pedestrian safety concern, yet “free from crime” was the safety-related attribute ranked highest for likelihood to increase the respondents’ walking. Also ranked important, for both walking and cycling, were “free from fast-moving traffic” and certain other traffic safety issues (Krizek et al., 2007). Personal Security. Behavioral Risk Factor Surveillance System (BRFSS) researchers, using data from five states, were the first to estimate rigorous quantitative relationships between physical inactivity and a perception of neighborhood dangerousness. The elderly, and racial/ethnic minori- ties, were found to be the most sensitive to perceived danger. (Children were not studied.) Physical inactivity for persons aged 65 and older ranged from 63 percent for those who reported their neigh- borhood was “not at all safe” down to 39 percent for “extremely safe.” Inactivity percentages for racial/ethnic minorities were 45 percent for adults reporting “not at all safe” neighborhoods down to 30 percent for “extremely safe” neighborhoods. Male non-elderly adults and persons with more 16-283

than a high school education showed little sensitivity. Physical inactivity was defined as no reported physical activity or exercise, which would include walking, within the previous month (Centers for Disease Control and Prevention, 1999). Since then, a number of additional studies have identified physical inactivity as being related to neighborhood danger perceptions for adults, and the same association has been found for children. However, numerous other studies have found no relationship. A study of Boston public housing residents, specific to walking activity, used pedometers for quantitative measurement of response to neighborhood safety perceptions. No association was found with daytime safety perceptions for either males or females, or with nighttime safety perceptions for males. Women who reported feel- ing unsafe at night, however, averaged only 4,302 steps per day as compared to 5,178 for women who felt safe at night (Bennett et al., 2007). Other studies specifically focused on subpopulations of older adults, women, and children have tended to show significant positive correlation between sedentary behavior and real or perceived dangers to personal safety. The relationships appear to be strongest for women, particularly minority women (Committee on Physical Activity, Health, Transportation, and Land Use, 2005). Pedestrian trip generation modeling done using the 2001 NHTS Baltimore add-on travel data sam- ple, augmented with Census- and GIS-derived household and urban form variables, identified a physical negative factor for walking that may well be a surrogate for high crime areas. Higher pro- portions of vacant household units, measured at the Census-block level, were estimated to be sig- nificantly related to lower rates of walking activity (Targa and Clifton, 2004). Traffic Safety. Presence of traffic control devices and perceptions that traffic moves at safe speeds were the second- and third-highest positive factors in survey-respondent rankings of high impor- tance in route choice, as noted earlier in discussing “Walk Route Choice” by pedestrians accessing rail transit stations. Only “Shortest route” outranked these factors in a list of 11 choices. When asked to volunteer route choice factors earlier in the survey, 28 percent mentioned safety, as com- pared to 64 percent mentioning shortest/fastest and 9 percent or less mentioning any other con- sideration (Weinstein et al., 2007). Fear of traffic dangers associated with non-motorized travel has received particular attention in the case of bicycling. Bicyclists who regularly cycle in traffic are not as concerned as non-riders, but this greater confidence could result from more cycling by less fearful cyclists (self-selection) as much as it could result from cycling experience (cause and effect). Multiple surveys have shown many people to be averse to bicycling because of traffic or lack of safe bikeways. For example, 58 percent of bike owners in Philadelphia and 55 percent of adults in Portland, Oregon responded accordingly in surveys (David Evans and Associates, 1992). As part of the much newer GPS-and- network-analysis study in Portland, cyclists apparently more concerned with cycling safety than others were identified with a survey question concerning the relative safety of driving and bicy- cling. Persons so identified gave added importance to riding on a bike lane or off-road trail. Cyclists in the study ranked the importance of avoiding “street with lots of traffic” second only to minimizing distance. Higher-than-average concern with traffic avoidance was recorded for women and infrequent cyclists (Dill and Gliebe, 2008). A review of several U.S. and Australian studies concluded that female cyclists have different percep- tions of road safety relative to males, irrespective of experience levels. Findings were reported of a higher aversion to risk among female cyclists, paired with a higher likelihood of being discouraged from bicycling when required to share space with vehicular traffic. The accompanying original research, staged in six small western cities in the United States, involved a survey of perceptions and bicycling activity (12.6 percent overall survey response rate) together with multivariate analysis. It 16-284

found women to report a significantly lower “comfort score” than men for all types of road/bicycle facilities except quiet streets, which scored highest and essentially the same for both sexes. As male and female comfort levels decreased across facility types, the percentage by which women were less comfortable than men increased until stabilizing at −17 percent. Table 16-80 lists the facility types examined and the corresponding scores obtained, in order of progression from highest score and least difference between women and men to lowest score and greatest difference. The researchers hypothesized that quiet streets offer female bicyclists good visibility to assuage personal safety con- cerns together with vehicular volumes low enough to mitigate traffic safety unease (Emond, Tang, and Handy, 2009). 16-285 Table 16-80 Male Versus Female Bicycling Comfort Scores by Facility Type Facility Type Male Score Female Score Pct. Difference Quiet street 2.92 2.91 0% Off-road shared use path 2.85 2.74 -4% Two-lane local street with bike lane 2.84 2.70 -5% Two-lane local street without bike lane 2.59 2.38 -8% Four-lane street with bike lane 1.97 1.65 -17% Four-lane street without bike lane 1.63 1.36 -17% Note: Scores are means calculated on a three-point scale: 1 = “uncomfortable and I wouldn’t ride on it” 2 = “uncomfortable but I’d ride there anyway” 3 = “comfortable.” Source: Emond, Tang, and Handy (2009). Traffic counts/observations conducted in connection with bicycle safety enhancements to Burrard Bridge in Vancouver, BC, provide empirical support of the thesis that female cyclists are the more concerned with traffic safety and will respond positively to improvements. After cyclists crossing Burrard Bridge were provided with a barrier-protected exclusive lane (outbound) and a full side- walk reserved for cyclists (inbound), use by cyclists increased 26 percent. Broken down by gender, the increases were 23 percent for men and 31 percent for women (City of Vancouver, 2009a). Not only perceived safety and comfort but also actual safety were increased, with a 75 percent reduc- tion in bicycle crashes requiring hospital emergency room visits (Mustel Group Market Research, 2009, City of Vancouver, 2010). (For additional information see “Response by Type of NMT Strategy”—“Pedestrian/Bicycle Systems and Interconnections”—“River Bridges and Other Linkages”—“Other River Bridges.”) The frequently mentioned, aggregate data, cross-sectional analysis of journey-to-work cycling in 90 of the 100 largest U.S. cities included a safety variable in the bicycling-rate analysis. Data limitations forced use of statewide bicyclist fatality averages as the measure. Even so, the safety variable was significant and indicated a bicycle commuting rate that was higher by 2.3 percent for each 10 percent by which the fatality rate was lower (Buehler and Pucher, 2011, Kuzmyak et al., 2011). Traffic safety concerns are, to a degree, addressable through infrastructure, operational, and institu- tional initiatives. Traffic engineering techniques available for safety enhancements are largely beyond the scope of this “Traveler Response” Handbook, but are summarized and cross-referenced in the “Related Information and Impacts” section under “Safety Information and Comparisons.” The subsections under “Response by Type of NMT Strategy” on “NMT Policies and Programs” and “Walking/Bicycling Promotion and Information” contain relevant information on institutional

initiatives such as the Safe Routes to School programs. Bicycle security can be addressed through parking security provisions. Trip-maker response to bicycle parking provisions is covered under “Point-of-Destination Facilities” within the “Response by Type of NMT Strategy” section. A better understanding of the traffic dangers which exist can help put things in perspective. It makes no sense, for example, to restrain children from walking independently based solely on gen- eralized crash rates affected heavily by adult-pedestrian alcohol intoxication. Bicycling does tend to be somewhat less safe than auto travel with an adult driver, and walking may be marginally so, but the only wildly unsafe travel mode relative to others is the motorcycle. The statistics of crash analysis are such that rate comparisons among modes can vary substantially depending on the measure of exposure used. A major focus of the “Safety Information and Comparisons” subsection of the “Related Information and Impacts” section is examination of crash statistics from more than one perspective in an effort to provide a balanced overview of traffic safety as pertains to walking and bicycling choices. University Affiliation University affiliation produces a special combination of environmental, user, and attitudinal fac- tors that heightens campus and citywide non-motorized mode shares. Overall, “college towns” have higher levels of bicycle commuting than non-university locales, especially when relatively large campuses are involved. Davis, California, Boulder, Colorado, and Madison, Wisconsin, are among the most cited examples with high levels of bicycle usage. Although each have significant bicycle infrastructure and bicycle-friendly development in place, the level of cycling is such that it most likely cannot be solely attributed to these accommodations. Perhaps students and university staff are joined in choice of the bicycle mode by other townspeople with no formal university affil- iation, simply because of the visible acceptability of the mode. Nevertheless, many college towns without the cycling infrastructure do not experience equivalent cycling. It is a combination of fac- tors responsible (Goldsmith, 1992, Victoria Transport Policy Institute, 2007). It certainly helps that locales with universities have a large population of young, healthy individ- uals living close by who may dress informally. In addition, most campuses limit or charge for park- ing and are otherwise congested to the point where the bicycle often has a time advantage over other modes. Trips are generally short and many schools are located with relatively bicycle- friendly surrounding streets, generally perceived as “safe” even if there are no dedicated bicycle facilities. Automobile ownership is low and bicycle ownership is high among students, and the culture and area motorists tend to be supportive of their use (Goldsmith, 1992, Everett, 1990). A majority of national studies making use of aggregate city or regional data to attempt isolation of factors linked with higher levels of bicycling have found elevated presence of college students to be associated with more commuter cycling within a city. (None of these studies examined travel purposes not fitting the U.S. Census definition of the “journey to work.”) The first such study to go beyond descriptive analysis was an early 1990s effort by Baltes employing commute mode data from the 1990 U.S. Census. Regression analysis covering a wide range of socio-demographic, travel time, and workplace factors isolated the proportion of college students as one of three or four key factors (Kuzmyak et al., 2011). Later in the decade a similar approach, expanded to include bicycle facility extent and weather data, found higher ratios of college students to be one of three primary variables strongly and pos- itively related to more commuter cycling. The other two were bikeway miles per 100,000 popula- tion and fewer rain days/year (Nelson and Allen, 1997). A subsequent regression analysis, 16-286

utilizing U.S. Census 2000 Supplemental Survey travel data, did not incorporate the percentage of college students within the population into the final research models. The 43 cities covered repre- sented a more uniform assemblage, with no small cities and no college towns (Dill and Carr, 2003). Finally, drawing on augmented 2006–2008 3-year average ACS cycling data covering 90 of the 100 largest U.S. cities, cross-sectional analysis identified student ratios as one of a half-dozen signifi- cant explanatory variables and one of those exhibiting a positive relationship (Buehler and Pucher, 2011) (See “Bicycle Lane System Coverage” within the “Bicycle Lanes and Routes” subsection of the “Response by Type of NMT Strategy” section for more on these three studies.) At the beginning of this “Underlying Traveler Response Factors” section, a potential three-step hier- archy of transportation decisionmaking was introduced, consisting of “initial barriers,” “trip barri- ers,” and “destination barriers.” In a college town such as Davis, Boulder, or Madison, clearly all of these barriers are relatively low. The infrastructure and institutional climate enables both students and non-students to be drawn to bicycling for a large percentage of their travel needs. Davis may be unique among contemporary U.S. cities in acceptance and use of cycling for meeting daily trans- portation needs, but college town and university district bicycle usage is often sufficient to engender a “virtuous circle.” Infrastructure and operational improvements are supported not just by promise of increased NMT activity, but also by readily evident day-to-day current volumes of bicyclists. The improvements, in turn, engender additional bicycling and other active transportation. Factor Combinations Involving Trip Purpose In the preceding “factors” discussions, and elsewhere in this chapter, there are some presentations and speculations about the role of trip purpose in the response or relationship to other factors and to facility improvements. For example, research using 2001 NHTS data found persons in the higher income categories about 40 percent less likely to make utilitarian walk-only trips than persons in the lowest income category, but almost 25 percent more likely to walk for recreation and exercise (Agrawal and Schimek, 2007). A similar relationship has been suggested for bicycling but not established (Alliance for Biking & Walking, 2010). In the aggregate, bicycling appears roughly uni- form across incomes (Kuzmyak et al., 2011), but the aggregation may obscure information impor- tant to planning and equity determinations. Such interrelationships are among the least studied of pedestrian and bicycle choice factors. In using data covering all or multiple trip purposes, the possibility that trip purpose aggregation masks important relationships must constantly be kept in mind. Cross-classifications by trip purpose of self-reported mode and route choice influences have pro- duced quite informative insights. One such case is the discovery that self-reported exercise and enjoy- ment motivational factors were almost as important to commuters as to non-commuters in their choice to use a new pedestrian/bicycle-only river crossing in Brisbane (Abrahams, 2002). A more comprehensive matching of path choice factors by trip destination purpose is provided by the Portland, Oregon, bicyclist route choice research. Participants were asked to rank, in order of impor- tance, seven factors that influenced their choice of route. These were then separately tallied for each of four categories of trip destination purposes. The results, provided in Table 16-81, provide instruc- tive similarities and certain key differences. Route choice factor rankings were virtually identical for the two categories encompassing purely utilitarian travel, with a one-point ranking difference show- ing up only in the factors ranked last or next-to-last. Distance minimization slipped from first place for utilitarian trips to second place for social/recreational trips and sixth place for exercise. Traffic avoidance rose from second place importance ranking in the case of utilitarian trips to first place ranking for social/recreational and exercise activity (Kuzmyak et al., 2011). 16-287

Attitudes and Modal Biases Effects of attitudes and modal biases on travel choices are subject to much debate. To whatever extent they exist, they are related. Modal biases are logically a function of both attitudes and experiences. The subject of attitudes is introduced, in a context relevant to NMT choice, within Chapter 15, “Land Use and Site Design.” In particular, see “Attitudes and Predispositions” within that chapter’s “Underlying Traveler Response Factors” section. Modal biases are frequently cited as having a role in travel choices. Although habit may play a part in transportation decisionmaking, most modal biases are probably explained by reactions to specific attributes of travel mode options. For example, in attitude surveys some people make clear that they do not like the bus and will not ride a bus. However, in most cases the problem relates to a dislike of actual or perceived attributes of bus riding, such as time spent waiting, the type of waiting environ- ment, harassment or crime concerns, unpleasant noise or odors, and so on, rather than some inher- ent dislike of buses per se. Many people make clear that they like to travel in private vehicles for a variety of reasons. They may perceive that private vehicles offer a more protected environment than other modes, and they appreciate having individual control over such factors as climate, radio, route, speed, destination, and people encountered enroute. For “modal biases” that deter use of active transportation modes to be properly addressed through public policy actions or marketing, inherent characteristics, perceptions, barriers to use, and other issues must be understood. Some have suggested that it is personal motivation rather than physical or rational factors that con- trols the decision to undertake active transportation (David Evans and Associates, 1992). A mode- judgment experiment examining the role of habit in active-transportation decisionmaking found subjects who normally used bicycles for transportation to be at least somewhat more likely to choose to bicycle in various hypothetical situations than those who did not as frequently use bicy- cles. Perhaps more crucially, the habitual riders demonstrably streamlined their decision process and used fewer attributes of the circumstances to determine their course of action (Aarts, Verplanken, and van Knippenberg, 1997). The implication is that mode choice is “sticky” and that, in addition to addressing the decision-driving attributes, significant forces of habit and predisposition must be overcome to make a change. Since a majority of U.S. commuters and other trip-makers are motorists, there is a fair amount of collective inertia to overcome. A 1981 FHWA study asked respondents which mode they preferred 16-288 Table 16-81 Importance Ranking of Factors affecting Bicycle Route Choice for Different Destination Purposes, Portland, Oregon Destination Purpose Work, Work- Related, School Shopping, Dining, Per– sonal Business Social and Recreational Exercise Route Choice Factor Minimize distance 1 1 2 6 Avoid traffic 2 2 1 1 Use on-road bike lane 3 3 3 2 Reduce intersection delays 4 4 5 5 Take a signed route 5 5 4 4 Use off-road bike path 6 7 7 3 Avoid hills 7 6 6 7 Source: Derived from Dill and Gliebe (2008) by Kuzmyak et al. (2011).

to use to make trips of various purposes and also which mode they actually used. For commuting, 72 percent of the people preferred using an automobile, but 75 percent actually did. Of the remain- der, 14 percent preferred walking and 7 percent preferred bicycling, although only 11 percent actu- ally walked and only 3 percent actually bicycled. These responses were viewed as being related to the package of attributes generally associated with each mode rather than the specific attributes underlying a specific circumstance. Thus the results were seen as highlighting the significant iner- tia surrounding use of the automobile (David Evans and Associates, 1992). Another interpretation might be that the actual-choice responses reflect the realities faced in actual trips, including both facility adequacies and inadequacies. A model of behavioral change, originally developed in the context of smoking cessation cam- paigns, identifies five stages relevant for overcoming inertia and inducing shifts to active trans- portation (Rose and Marfurt, 2007): 1. Pre-contemplation (not yet thinking about changing). 2. Contemplation (consciously contemplating change). 3. Preparation (preparing to make the change). 4. Action (taking action to change). 5. Maintenance (sustaining the change). Voluntary travel behavior change (VTBC) (see “Individualized Marketing” within “Response by Type of NMT Strategy”—“Walking/Bicycling Promotion and Information”) raises interesting questions concerning the role of attitudes. Individualized marketing in the interests of VTBC is designed to raise awareness, improve information availability, and offer support for people trying alternatives to driving (Brög and Ker, 2008). These thrusts serve to address inertia and the behav- ioral stages listed above. Improving information is most relevant to the contemplation and especially the preparation stages. The role of information availability is perhaps best illustrated by this approximated response of a Bellingham, Washington, recipient of individualized marketing: “Oh! I didn’t know I could take that trail to downtown—I thought it was just for exercising!” In this instance, the information recipient was simply placed in a better position to make an informed utilitarian decision. On the other hand, VTBC is said to work also through empowerment and motivation. A key component is getting people to actually try an appropriate alternative travel mode and to reward the new behavior. This would seem to fall more in the sphere of changing attitudes, particularly when social learning based on the exam- ple of others—including early adopters—is a factor (Horst, 2010b, Brög and Ker, 2008). Clearly environmental and user factors, facilities and services, time and cost parameters, availability of information about them, and attitudes are all important in travel behavior. There is of yet no agreement on the relative importance of “hard” versus “soft” factors, and indeed, their relative import no doubt depends on circumstances. Nevertheless, many explanatory research models—plus applied travel demand modeling experience—suggest it would be incorrect to ascribe the bulk of travel behavior out- comes to attitudes or related factors such as neighborhood choice (see next subsection). At the same time, it would be wrong to assume that the roles of inertia and attitudes are unimportant. Attitudes, to the extent they apply, can affect adult travel behavior in two different ways: they can directly affect short-term travel choices such as choice of mode and they can affect long-term 16-289

underlying mobility choices such as residential location, employment location, and vehicle own- ership. The characteristics of the neighborhood chosen will in turn help define the attractiveness of alternative travel options and thereby influence short-term travel choices (Cao, Mokhtarian, and Handy, 2009, Federal Highway Administration, 1974). Implications of neighborhood choice are further explored in the next subsection. The effect of attitudes on child travel behavior is somewhat different, since parental attitudes shape the travel of children. Parents presumably evaluate the trade-off between perceived child safety and benefits of independent movement and active transportation—an assessment likely influenced heavily by attitudes and social norms—determining whether the child is allowed to travel by walk- ing, cycling, or public bus without adult supervision, or only with supervision, or gets chauffeured by private vehicle (Mackett et al., 2007a, Mackett et al., 2007b). (Within the earlier “Behavioral Paradigms” subsection, see “The Travel Choice Making of and for Children” discussion of atti- tudes as “moderating factors” in parental decisionmaking.) This decisionmaking is also signifi- cantly impacted by neighborhood characteristics (Davison and Lawson, 2006, Moudon, Stewart, and Lin, 2010); so again, choice of neighborhood has its effect. Choice of Neighborhood/Self-Selection Predispositions, in the form of attitudes and modal biases, may directly impact the immediate travel choice of whether or not to walk or bike or use public transit for any given trip, as discussed above. Alternatively, they may impact long-term mobility choices that will affect future short-term travel choices. This subsection examines travel demand interplay with neighborhood choice—primarily choice of pedestrian-and-bicycle-friendly neighborhoods over auto-oriented neighborhoods. Though not always thought of this way, selection of home neighborhood is a prime example of a mobility choice affecting subsequent travel choices such as using or not choosing active transportation for util- itarian trips. Neighborhood choice can also affect (or be affected by) auto ownership, which in turn is another mobility choice factor impacting travel decisions (Federal Highway Administration, 1974). Neighborhood characteristics likewise affect personal decisions concerning exercise and recreation (see “Public Health Issues and Relationships” under “Related Information and Impacts.”) Self-Selection Investigations Research on self-selection at the Institute of Transportation Studies, University of California at Davis, has included an examination of 38 studies focusing either on self-selection or attitudinal effects along with built environment effects on travel behavior. The individual studies address a broad array of built environment descriptors covering land use and transportation features postulated to be sup- portive (or non-supportive) of walking, bicycling, transit use, and reduction in vehicle miles of travel (VMT). The researchers conclude: “Virtually every quantitative study . . . after controlling for self- selection . . . identified a statistically significant influence of one or more built environment measures on the travel behavior variable of interest” (Cao, Mokhtarian, and Handy, 2007 and 2009). Table 16-82 outlines analytical approaches and empirical findings for 14 of these studies that directly addressed non-motorized travel while also explicitly examining residential self-selection and/or attitu- dinal effects. Some 3 to 5 of the 14 studies may be interpreted to have found self-selection or attitudinal effects more important for NMT choice decisions than built environment effects. Four to six of the stud- ies found self-selection and built environment effects to be of roughly equivalent importance, and five found the built environment effects to be more important. Several of the studies concluded that residen- tial preference effects were more important in NMT choices than in transit mode or auto-related choices. 16-290

16-291 Table 16-82 Summary of Findings about NMT Effects of Residential Self-Selection (SS) Relative to Direct Impacts of the Built Environment (BE) Study Process Key Findings 1. Handy and Clifton – 2001 Direct questioning of some 1,400 participants in Austin, TX, [1995] with descriptive and correlational analysis of walk-to-store frequency. Both SS and BE effects identified: Having the option to walk to store was “to some extent an effect of the desire to walk” while perceived store characteristics influenced frequency. 2. Cao, Handy, and Mokhtarian – 2006 Analysis with statistical control of walk-to-store and strolling frequencies of 1,368 individuals in Austin, TX, [1995] using negative binomial regression. Both SS and BE effects identified: Residential preference “most important single factor explaining walk-to-store frequency” but objective and perceived area characteristics had separate influences on walk-to-store and strolling frequencies. 3. Cao, Mokhtarian, and Handy – 2005 Analysis with statistical control of nonwork trip frequencies by mode of 1,682 individuals from Northern California [2003] using seemingly unrelated regression. Both SS and BE effects identified: SS “more likely to influence walking/biking trips than auto and transit trips,” but the BE was nevertheless also found to influence all trips. 4. Chatman – 2009 Analysis with statistical control of numbers by mode of nonwork activities accessed by 1,114 adults in the San Francisco and San Diego regions [2003] using negative binomial regression. Mode preference effects found to be less than BE effects: Both transit and walk/bike trips affected by mode preferences but the BE (as expressed in transit quality and street connectivity measures) also independently affected these alternative modes. Auto travel showed no effects. 5. Frank et al. – 2007 Analysis with statistical control of VMT and percent taking walking trips, with walkability index and residential preference variables, using linear regression and two subsamples of 2,056 and 1,466 from 2001-02 Atlanta SMARTRAQ data. Residential preference and BE both affected driving and walking prevalence. BE effects were stronger for VMT and residential preference effects were stronger for walking. (See also the discussion to follow of residents locationally matched and mismatched with their area-type preferences.) 6. Kitamura, Mokhtarian, and Laidet – 1997 Analysis with statistical control of numbers and fractions of trips by mode across 963 households in San Francisco Bay Area [1993] using linear regression. The attitudinal measures carried more explanatory power than the measures used for BE characteristics (see Chapter 15 — “Underlying Traveler Response Factors” — “Attitudes and Predispositions” for more). 7. Schwanen and Mokhtarian – 2005 (two papers) Analysis with statistical control of commute trip shares and weekly miles by mode (personal vehicle, public transit, walk/jog/bike) for 1,358 workers in San Francisco Bay Area [1998] using multinomial logit and Tobit models. Neighborhood (NBH) preference effects found to be less than BE effects: Preferences had less effect in suburban environments than differences within each preference group between suburban and traditional urban environments. (See Chapter 17 — “Underlying… Factors” — “Self-Selection of Residents” — “Self-Selection Effects on TOD Regional Travel…” for more). 8. Khattak and Rodriguez – 2005 Regression analysis with instrumental variables models of various trip-type frequencies and durations for 453 households in Chapel Hill and Carrboro, NC. BE dominant: In contrast to 8 measures of residential preference, the BE (identified in terms of neo-traditional and suburban NBHs) “influenced most measures of travel behavior.” 9. Boer et al. – 2007 Analysis, with propensity score (probability of self-selection) matching, of choice of walking in 10 metro areas in 1995 NPTS. Both SS and BE effects identified, including land use mix, density, and parking pressure. Most BE influences became insignificant with propensity score matching. (continued on next page)

16-292 Table 16-82 (Continued) Study Process Key Findings 10. Cao – 2008 Analysis using propensity score stratification of walking frequency (2 measures) and VMT for 1,553 Northern CA residents [2003]. BE dominant: SS (residential preferences and travel attitudes) estimated to account for 14% of effect on strolling frequency, 39% for walk- to-store frequency, and 22% for VMT. 11. Bagley and Mokhtarian – 2002 Analysis with simultaneous models (structural equations) of vehicle, transit, and walk/bike miles by 515 individuals in San Francisco Bay Area [1993]. SS effects found to be more important than BE effects: “Residential location type had little separate impact on travel behavior; attitudes and lifestyles were the most important predictors of travel behavior.” 12. Salon – 2006 Analysis with simultaneous models (nested logit) of residential choice, auto ownership, and walking levels for 4,382 New York City regional- travel-survey respondents. Both SS and BE effects identified: “Self- selection accounted for 1/3 to 1/2 the total influence of the built environment” using density as the neighborhood characteristics measure. 13. Cao, Mokhtarian, and Handy – 2007 Longitudinal analysis, using a structural equations model, of changes in auto ownership, driving, and walking/biking for 547 movers in Northern California [2003]. Both SS and BE effects: “Attitudes influenced auto ownership and travel behavior” while the BE had separate effects, isolated using both objective and perceived NBH measures for previous and new residence locations. 14. Handy, Cao, and Mokhtarian – 2006 Longitudinal analysis, using an ordered probit model, of strolling and walking-to-store frequencies and of walking changes and biking changes for 1,682 individuals in Northern California [2003]. Both SS and BE effects: Cross-sectional analyses showed “influence of attitudes on walking” while “longitudinal analysis showed separate effects of BE on walking and biking behavior” based on objective and perceived NBH measures and perceived changes. Note: Drawn from summaries of 38 studies, omitting those not directly addressing non-motorized travel or with no apparent examination of residential self-selection or attitudinal effects. Where substantial additional information on individual studies is provided in text and tables or figures, this is noted — and the location within the TCRP Report 95 “Traveler Response” Handbook is given — in the third column. Source: Cao, Mokhtarian, and Handy (2007 and 2009). Not included in Table 16-82—along with research not explicitly investigating NMT use—are stud- ies which exclude residential choice effects by their very nature, such as before-and-after investi- gations and certain longitudinal studies. The impact estimates produced by these types of studies are of built-environment effects alone, or built-facility or executed-program effects alone, though such estimates have their own sets of issues such as confounding events and multiple causations. It is cross-sectional studies that inherently present the knottiest problems for isolation of residen- tial choice influences. Cross-sectional studies tend to dominate travel behavior research address- ing land use and site design, and are also found in applications such as public health research on sidewalk and other NMT facility effects. Table 16-83 summarizes five additional studies of special interest in the consideration of residen- tial choice even though they do not specifically address non-motorized travel overall. They focus on either private vehicle use, with alternative mode effects inferred, or on bicycle ownership and use. Two of these five studies found self-selection effects to be the more important, one found self- selection and built environment effects to be of roughly equivalent importance, and two found the built environment effects to be more important.

Most of the studies listed in Tables 16-82 and 16-83 employed some form of cross-sectional analy- sis as the investigative technique. Three, however, had the advantage of data on travel behavior before and after residence relocation. Two of those three studies (the 13th and 14th in Table 16-82) found self-selection and built environment effects to be of roughly equivalent importance, and one (the 3rd in Table 16-83) found the built environment effects to be of prime importance (Cao, Mokhtarian, and Handy, 2007, Krizek, 2003). In the latter instance, using a model developed on data from 7 waves of the Puget Sound Transportation Panel survey, it was estimated that the daily household average VMT was 5 miles more—irrespective of preferences—when households that moved were located in a representative suburban location as compared to a representative urban location (Krizek, 2003). Corresponding active transportation effects have to be inferred. Three of the newer studies (the 9th and 10th in Table 16-82 and the 1st in Table 16-83) have explored approaches drawing upon medical treatment analysis procedures. They addressed self-selection as a treatment bias, similar to potentially greater use of a preventative medicine by health-conscious indi- viduals. One of these found both self-selection and built-environment effects to be of comparable importance and two concluded that built environment effects were more important (Cao, Mokhtarian, and Handy, 2009, Zhou and Kockelman, 2008). 16-293

Five studies have been encountered that undertook to put numbers on the proportions of built envi- ronment effects on travel behavior versus self-selection or attitudes. Table 16-84 consolidates the quantitative findings. Notably, all five of these studies indicate that built environment effects tend to substantially exceed, or at least roughly equal, self-selection and attitudinal effects. Both types of influences were, however, found in each study (Ewing and Cervero, 2010). 16-294 Table 16-83 Summary of Selected Research on Interrelationships among Residential Self-Selection (SS), the Built Environment (BE), Auto VMT, and Bicycle Ownership and Use Study (Date) Process (Limitations) Key Findings 1. Zhou and Kockelman (2008) Formulated SS as a sample selection bias, drawing on medical treatment statistics and utilizing Heckman’s latent index model. Tested with 1,903 household sample. (Highly aggregate area-type BE indicator.) Estimated difference of 17.0 to 20.2 daily VMT between central/CBD and suburban/rural Austin, TX, with 58% or more (up to 90%) of the difference attributable to BE (the “treatment”) rather than SS (the “bias”). 2. Circella, Mokhtarian, and Handy (2008) Structural equations modeling, using survey results for 1,217 workers in 8 Northern CA communities [2003], to examine residential location, auto ownership, and VMT. (Direction of causality was found difficult to determine.) Travel and land use attitudes strongly associated with travel and location behavior. SS confirmed for persons preferring alternative travel solutions. BE found to matter also, with higher neighborhood relative to regional ac- cessibility favoring alternative modes. 3. Krizek (2003) SS issues bypassed by examining VMT, person miles traveled (PMT), and tour characteristics changes by 430 households who moved during the Puget Sound Transportation Panel’s 7 waves, which provide disaggregate and longitudinal travel data. (Assumed that relocations were mainly for reasons other than travel environment self-selection.) Households alter their travel in response to differing built environ- ments. Relocating to residences with higher neighborhood accessibility increases the number of daily tours but decreases the trips per tour, PMT, and VMT. Higher regional accessibility is also associated with decreased trips, PMT, and VMT, but with statistically insignificant effects on number of tours. 4. Pinjari, Bhat, and Hensher (2008) Modeled residential location and activity time-use choices of 2,793 households in Alameda Co., CA, using a multinomial logit formula- tion that accommodates attributes both observed and unobserved and controls for SS. (Bicycle ownership levels treated as given.) Individuals with “a preference for physically active pure recreation” and higher bicycle ownership tend to locate in neighborhoods with good bicycle facility density, nevertheless, modify- ing the activity-travel environment can produce small activity level changes (facility density x 10 = 17% increase). 5. Xing, Handy, and Buehler (2008) Cross-sectional analysis, using nested logistic models, of relative influence of individual, social-envi- ronment, and physical-environment factors on bicycle ownership and use in 6 small Western U.S. cities. (13% overall survey response rate, mismatch with U.S. Census.) Aside from socio-demographics, attitudes were dominant in explaining bicycle ownership and use. No BE effect on ownership was found. Two BE factors showed significance as bike use descriptors: perception of safety in reaching selected destinations, and transit access (for reasons not obvious). Sources: As indicated in the first column.

Neighborhood Preference Matches and Mismatches Two research efforts included in Table 16-82 looked beyond self-selection concerns, probing what the net effects are—with and/or without self-selection—of different types of built environments. A San Francisco Bay Area series of studies (the 6th entry in Table 16-82 and covered further in Chapter 17, “Transit Oriented Development,” as noted) studied persons in different residential environments matched and mismatched with their area-type preferences. Neither suburban- oriented nor urban-oriented individuals residing in the suburbs exhibited walk/jog/bike commute mode shares averaging over 0.4 percent, but suburban-oriented individuals in an urban environment averaged a corresponding 3 percent walk/jog/bike commute share, with urban-oriented individuals similarly located averaging 5 percent. The suburban versus urban weekly VMT differential was 82 miles less driven for urban-located suburban-oriented individuals and 100 miles less for urban-located-and- oriented individuals (Schwanen and Mokhtarian, 2005a and b). An Atlanta-based study of residential self-selection (the 5th entry in Table 16-82) carried out a sim- ilar analysis of matched and mismatched residents, albeit with somewhat different criteria and parameters. This analysis was supplemental to the primary statistically controlled modeling and was in essence an illustrative descriptive analysis. Samples were drawn from 2-day trip diary results (85 percent weekdays) for a subset of the regional SMARTRAQ travel survey in which neighborhood preference and walkability score data had been obtained. Only cases in quartiles reflecting substantive preferences and walkability differentials were used in the analysis. Occurrence of one or more walk trips and VMT were both examined. For walk trips, the preference differentials were most striking, but whichever the preference category, high walk- ability neighborhoods had twice the incidence of persons walking than low walkability neighbor- hoods. Of persons preferring environments that tended to be less walkable, 3.3 percent in low walkability neighborhoods took at least one walk trip versus 7.0 percent of persons in high walk- ability neighborhoods. For persons desiring high walkability, 16.0 percent in low walkability neighborhoods took at least one walk trip, while 33.9 percent in high walkability neighborhoods 16-295 Table 16-84 Studies Quantifying the Relative Contributions to Travel Differences of Built Environment (BE) Versus Residential Self-Selection (SS) or Attitudinal Effects Study (Date) Quantitative Estimate of BE Versus SS or Attitudinal Effects 1. Salon – 2006 BE effects accounted for 1/2 to 2/3 of differences in New York City walking levels associated with density (see also Table 16-82, 12th study). 2. Cao – 2008 BE, for Northern California residents in a 2003 survey, accounted for 86% of effect on recreational walking frequency, 61% of effect on walk-to-store frequency, and 78% of effect on VMT (see also Table 16-82, 10th study). 3. Zhou and Kockelman – 2008 BE, depending on assumptions, accounted for 58% to as high as 90% of the VMT difference associated with central area versus suburban/rural housing location per the 1998-99 Austin travel survey (see also Table 16-83, 1st study). 4. Cao, Xu, and Fan – 2009 From 48% to 98% of differences in VMT identified in a Raleigh, NC, regional travel diary survey due to direct BE influences — remainder attributable to SS. 5. Bhat and Eluru – 2009 Some 87% of household VMT differences between conventional suburban and traditional urban neighborhoods, observed in a 2000 San Francisco Bay Area travel survey, found due to “true” BE effects — remainder due to housing SS. Sources: Cao, Mokhtarian, and Handy (2009), Ewing and Cervero (2010).

did so. In the context of overall average driving of 33 VMT per day per individual, the low versus high walkability neighborhood VMT differential was 17 fewer miles driven for residents of highly walkable neighborhoods who were not seeking walkability and 11 miles less for residents of highly walkable neighborhoods strongly preferring high walkability (Frank et al., 2007). Both the San Francisco and Atlanta area research efforts suggest that “What is the extent of self- selection?” is a question that does not have to be fully resolved to know that pedestrian-and-bicycle- friendly urban design achieves more walking and cycling activity and less VMT. On the other hand, in estimating the travel demand differentials associated with different neighborhood designs, self- selection effects do appear to deserve addressing. Importantly, if there is an unmet demand for more compact, mixed use, walkable neighborhoods, then increasing their supply may enable the most sus- tainable of all transportation-supportive housing selection outcomes: the movement into such areas of persons attuned to them, allowing the relocated residents to better “act on their preferences by walk- ing more and driving less” (Cao, Mokhtarian, and Handy, 2009). An impression that there is insufficient housing stock to meet demand for compact, mixed use, walk- able areas is given support by studies in metropolitan Atlanta. A survey sample of 1,455 residents from the SMARTRAQ research program were queried about preferences and categorized into those prefer- ring such “alternative development” and persons preferring the characteristics of auto-oriented neigh- borhoods. Those who preferred alternative development and had a desire to change from the land use and transportation characteristics of their current neighborhood outnumbered those preferring an auto-oriented environment and desiring change by over 2 to 1. This was taken as a definitive indicator of unmet demand in greater Atlanta for alternative development (Levine and Frank, 2007). A related perspective is offered by the authors of the travel and built environment meta-analysis cov- ered in the “Response by Type of NMT Strategy” section (see “Pedestrian/Bicycle Friendly Neighborhoods”—“Walk Elasticities for Land Use and Site Design Parameters”). Their meta-analysis used both many studies that did not control for self-selection and a smaller number of newer studies that did. It was found that elasticities of travel shifts in response to built environment characteristics that were derived from studies controlling for self-selection were either little different from or higher than those derived from studies not examining self-selection. As an explanation for this unexpected result, the authors hypothesize that many residents of higher density, mixed use, walk-friendly areas are indeed self-selecting and fulfilling a latent demand for active transportation and transit use brought about by insufficient supply of alternative development. They identify supporting research by others, but acknowledge that their hypothesis does not mesh with the larger body of literature that finds a degree of attenuation in built environment effects, and links it to self-selection (Ewing and Cervero, 2010). An alternative explanation could be that the studies not controlling for self-selection represent not only older but also less well specified research. As discussed under “Analytical Considerations” in the “Overview and Summary,” less well executed research has been shown to be associated with less frequent findings of statistically significant differences in walking activity (Ogilvie et al., 2007). Working with Self-Selection A relatively recent analytical development pertaining to attitudes and “self-selection” is the use of statistical methods to control for these factors, internal to the core modeling approach, when esti- mating built environment effects on active transportation. This is accomplished using one of a number of forms of “joint models” or models derived from medical and epidemiological practice, including joint discrete score models, structural equations models, mutually dependent discrete choice models, sample selection models, and application of propensity scores (Cao, Mokhtarian, and Handy, 2009). The approach accepts that there likely are attitudinal and residential self-selection 16-296

Next: Related Information and Impacts »
Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities Get This Book
×
 Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 16, Pedestrian and Bicycle Facilities
Buy Paperback | $98.00
MyNAP members save 10% online.
Login or Register to save!
Download Free PDF

TRB’s Transit Cooperative Research Program (TCRP) Report 95: Traveler Response to Transportation System Changes Handbook, Third Edition; Chapter 16, Pedestrian and Bicycle Facilities examines pedestrian and bicyclist behavior and travel demand outcomes in a relatively broad sense.

The report covers traveler response to non-motorized transportation (NMT) facilities both in isolation and as part of the total urban fabric, along with the effects of associated programs and promotion. The report looks not only at transportation outcomes, but also recreational and public health outcomes.

TCRP Report 95, Chapter 16 focuses on the travel behavior and public health implications of pedestrian/bicycle area-wide systems; NMT-link facilities such as sidewalks, bicycle lanes, and on-transit accommodation of bicycles; and node-specific facilities such as street-crossing treatments, bicycle parking, and showers.

The report also includes discussion of the implications of pedestrian and bicycle “friendly” neighborhoods, policies, programs, and promotion.

The report is complemented by illustrative photographs provided as a “Photo Gallery” at the conclusion of the report. In addition, PowerPoint slides of the photographs are available for download..

The Traveler Response to Transportation System Changes Handbook consists of these Chapter 1 introductory materials and 15 stand-alone published topic area chapters. Each topic area chapter provides traveler response findings including supportive information and interpretation, and also includes case studies and a bibliography consisting of the references utilized as sources.

READ FREE ONLINE

  1. ×

    Welcome to OpenBook!

    You're looking at OpenBook, NAP.edu's online reading room since 1999. Based on feedback from you, our users, we've made some improvements that make it easier than ever to read thousands of publications on our website.

    Do you want to take a quick tour of the OpenBook's features?

    No Thanks Take a Tour »
  2. ×

    Show this book's table of contents, where you can jump to any chapter by name.

    « Back Next »
  3. ×

    ...or use these buttons to go back to the previous chapter or skip to the next one.

    « Back Next »
  4. ×

    Jump up to the previous page or down to the next one. Also, you can type in a page number and press Enter to go directly to that page in the book.

    « Back Next »
  5. ×

    To search the entire text of this book, type in your search term here and press Enter.

    « Back Next »
  6. ×

    Share a link to this book page on your preferred social network or via email.

    « Back Next »
  7. ×

    View our suggested citation for this chapter.

    « Back Next »
  8. ×

    Ready to take your reading offline? Click here to buy this book in print or download it as a free PDF, if available.

    « Back Next »
Stay Connected!