National Academies Press: OpenBook
« Previous: Front Matter
Page 1
Suggested Citation:"UNDERSTANDING TRAVEL ISSUES." National Academies of Sciences, Engineering, and Medicine. 2005. Research on Women's Issues in Transportation - Volume 2: Technical Papers. Washington, DC: The National Academies Press. doi: 10.17226/23299.
×
Page 1
Page 2
Suggested Citation:"UNDERSTANDING TRAVEL ISSUES." National Academies of Sciences, Engineering, and Medicine. 2005. Research on Women's Issues in Transportation - Volume 2: Technical Papers. Washington, DC: The National Academies Press. doi: 10.17226/23299.
×
Page 2
Page 3
Suggested Citation:"UNDERSTANDING TRAVEL ISSUES." National Academies of Sciences, Engineering, and Medicine. 2005. Research on Women's Issues in Transportation - Volume 2: Technical Papers. Washington, DC: The National Academies Press. doi: 10.17226/23299.
×
Page 3
Page 4
Suggested Citation:"UNDERSTANDING TRAVEL ISSUES." National Academies of Sciences, Engineering, and Medicine. 2005. Research on Women's Issues in Transportation - Volume 2: Technical Papers. Washington, DC: The National Academies Press. doi: 10.17226/23299.
×
Page 4
Page 5
Suggested Citation:"UNDERSTANDING TRAVEL ISSUES." National Academies of Sciences, Engineering, and Medicine. 2005. Research on Women's Issues in Transportation - Volume 2: Technical Papers. Washington, DC: The National Academies Press. doi: 10.17226/23299.
×
Page 5
Page 6
Suggested Citation:"UNDERSTANDING TRAVEL ISSUES." National Academies of Sciences, Engineering, and Medicine. 2005. Research on Women's Issues in Transportation - Volume 2: Technical Papers. Washington, DC: The National Academies Press. doi: 10.17226/23299.
×
Page 6
Page 7
Suggested Citation:"UNDERSTANDING TRAVEL ISSUES." National Academies of Sciences, Engineering, and Medicine. 2005. Research on Women's Issues in Transportation - Volume 2: Technical Papers. Washington, DC: The National Academies Press. doi: 10.17226/23299.
×
Page 7
Page 8
Suggested Citation:"UNDERSTANDING TRAVEL ISSUES." National Academies of Sciences, Engineering, and Medicine. 2005. Research on Women's Issues in Transportation - Volume 2: Technical Papers. Washington, DC: The National Academies Press. doi: 10.17226/23299.
×
Page 8
Page 9
Suggested Citation:"UNDERSTANDING TRAVEL ISSUES." National Academies of Sciences, Engineering, and Medicine. 2005. Research on Women's Issues in Transportation - Volume 2: Technical Papers. Washington, DC: The National Academies Press. doi: 10.17226/23299.
×
Page 9
Page 10
Suggested Citation:"UNDERSTANDING TRAVEL ISSUES." National Academies of Sciences, Engineering, and Medicine. 2005. Research on Women's Issues in Transportation - Volume 2: Technical Papers. Washington, DC: The National Academies Press. doi: 10.17226/23299.
×
Page 10
Page 11
Suggested Citation:"UNDERSTANDING TRAVEL ISSUES." National Academies of Sciences, Engineering, and Medicine. 2005. Research on Women's Issues in Transportation - Volume 2: Technical Papers. Washington, DC: The National Academies Press. doi: 10.17226/23299.
×
Page 11
Page 12
Suggested Citation:"UNDERSTANDING TRAVEL ISSUES." National Academies of Sciences, Engineering, and Medicine. 2005. Research on Women's Issues in Transportation - Volume 2: Technical Papers. Washington, DC: The National Academies Press. doi: 10.17226/23299.
×
Page 12
Page 13
Suggested Citation:"UNDERSTANDING TRAVEL ISSUES." National Academies of Sciences, Engineering, and Medicine. 2005. Research on Women's Issues in Transportation - Volume 2: Technical Papers. Washington, DC: The National Academies Press. doi: 10.17226/23299.
×
Page 13
Page 14
Suggested Citation:"UNDERSTANDING TRAVEL ISSUES." National Academies of Sciences, Engineering, and Medicine. 2005. Research on Women's Issues in Transportation - Volume 2: Technical Papers. Washington, DC: The National Academies Press. doi: 10.17226/23299.
×
Page 14
Page 15
Suggested Citation:"UNDERSTANDING TRAVEL ISSUES." National Academies of Sciences, Engineering, and Medicine. 2005. Research on Women's Issues in Transportation - Volume 2: Technical Papers. Washington, DC: The National Academies Press. doi: 10.17226/23299.
×
Page 15
Page 16
Suggested Citation:"UNDERSTANDING TRAVEL ISSUES." National Academies of Sciences, Engineering, and Medicine. 2005. Research on Women's Issues in Transportation - Volume 2: Technical Papers. Washington, DC: The National Academies Press. doi: 10.17226/23299.
×
Page 16
Page 17
Suggested Citation:"UNDERSTANDING TRAVEL ISSUES." National Academies of Sciences, Engineering, and Medicine. 2005. Research on Women's Issues in Transportation - Volume 2: Technical Papers. Washington, DC: The National Academies Press. doi: 10.17226/23299.
×
Page 17
Page 18
Suggested Citation:"UNDERSTANDING TRAVEL ISSUES." National Academies of Sciences, Engineering, and Medicine. 2005. Research on Women's Issues in Transportation - Volume 2: Technical Papers. Washington, DC: The National Academies Press. doi: 10.17226/23299.
×
Page 18
Page 19
Suggested Citation:"UNDERSTANDING TRAVEL ISSUES." National Academies of Sciences, Engineering, and Medicine. 2005. Research on Women's Issues in Transportation - Volume 2: Technical Papers. Washington, DC: The National Academies Press. doi: 10.17226/23299.
×
Page 19
Page 20
Suggested Citation:"UNDERSTANDING TRAVEL ISSUES." National Academies of Sciences, Engineering, and Medicine. 2005. Research on Women's Issues in Transportation - Volume 2: Technical Papers. Washington, DC: The National Academies Press. doi: 10.17226/23299.
×
Page 20
Page 21
Suggested Citation:"UNDERSTANDING TRAVEL ISSUES." National Academies of Sciences, Engineering, and Medicine. 2005. Research on Women's Issues in Transportation - Volume 2: Technical Papers. Washington, DC: The National Academies Press. doi: 10.17226/23299.
×
Page 21
Page 22
Suggested Citation:"UNDERSTANDING TRAVEL ISSUES." National Academies of Sciences, Engineering, and Medicine. 2005. Research on Women's Issues in Transportation - Volume 2: Technical Papers. Washington, DC: The National Academies Press. doi: 10.17226/23299.
×
Page 22
Page 23
Suggested Citation:"UNDERSTANDING TRAVEL ISSUES." National Academies of Sciences, Engineering, and Medicine. 2005. Research on Women's Issues in Transportation - Volume 2: Technical Papers. Washington, DC: The National Academies Press. doi: 10.17226/23299.
×
Page 23
Page 24
Suggested Citation:"UNDERSTANDING TRAVEL ISSUES." National Academies of Sciences, Engineering, and Medicine. 2005. Research on Women's Issues in Transportation - Volume 2: Technical Papers. Washington, DC: The National Academies Press. doi: 10.17226/23299.
×
Page 24
Page 25
Suggested Citation:"UNDERSTANDING TRAVEL ISSUES." National Academies of Sciences, Engineering, and Medicine. 2005. Research on Women's Issues in Transportation - Volume 2: Technical Papers. Washington, DC: The National Academies Press. doi: 10.17226/23299.
×
Page 25
Page 26
Suggested Citation:"UNDERSTANDING TRAVEL ISSUES." National Academies of Sciences, Engineering, and Medicine. 2005. Research on Women's Issues in Transportation - Volume 2: Technical Papers. Washington, DC: The National Academies Press. doi: 10.17226/23299.
×
Page 26
Page 27
Suggested Citation:"UNDERSTANDING TRAVEL ISSUES." National Academies of Sciences, Engineering, and Medicine. 2005. Research on Women's Issues in Transportation - Volume 2: Technical Papers. Washington, DC: The National Academies Press. doi: 10.17226/23299.
×
Page 27
Page 28
Suggested Citation:"UNDERSTANDING TRAVEL ISSUES." National Academies of Sciences, Engineering, and Medicine. 2005. Research on Women's Issues in Transportation - Volume 2: Technical Papers. Washington, DC: The National Academies Press. doi: 10.17226/23299.
×
Page 28
Page 29
Suggested Citation:"UNDERSTANDING TRAVEL ISSUES." National Academies of Sciences, Engineering, and Medicine. 2005. Research on Women's Issues in Transportation - Volume 2: Technical Papers. Washington, DC: The National Academies Press. doi: 10.17226/23299.
×
Page 29
Page 30
Suggested Citation:"UNDERSTANDING TRAVEL ISSUES." National Academies of Sciences, Engineering, and Medicine. 2005. Research on Women's Issues in Transportation - Volume 2: Technical Papers. Washington, DC: The National Academies Press. doi: 10.17226/23299.
×
Page 30
Page 31
Suggested Citation:"UNDERSTANDING TRAVEL ISSUES." National Academies of Sciences, Engineering, and Medicine. 2005. Research on Women's Issues in Transportation - Volume 2: Technical Papers. Washington, DC: The National Academies Press. doi: 10.17226/23299.
×
Page 31
Page 32
Suggested Citation:"UNDERSTANDING TRAVEL ISSUES." National Academies of Sciences, Engineering, and Medicine. 2005. Research on Women's Issues in Transportation - Volume 2: Technical Papers. Washington, DC: The National Academies Press. doi: 10.17226/23299.
×
Page 32
Page 33
Suggested Citation:"UNDERSTANDING TRAVEL ISSUES." National Academies of Sciences, Engineering, and Medicine. 2005. Research on Women's Issues in Transportation - Volume 2: Technical Papers. Washington, DC: The National Academies Press. doi: 10.17226/23299.
×
Page 33
Page 34
Suggested Citation:"UNDERSTANDING TRAVEL ISSUES." National Academies of Sciences, Engineering, and Medicine. 2005. Research on Women's Issues in Transportation - Volume 2: Technical Papers. Washington, DC: The National Academies Press. doi: 10.17226/23299.
×
Page 34
Page 35
Suggested Citation:"UNDERSTANDING TRAVEL ISSUES." National Academies of Sciences, Engineering, and Medicine. 2005. Research on Women's Issues in Transportation - Volume 2: Technical Papers. Washington, DC: The National Academies Press. doi: 10.17226/23299.
×
Page 35
Page 36
Suggested Citation:"UNDERSTANDING TRAVEL ISSUES." National Academies of Sciences, Engineering, and Medicine. 2005. Research on Women's Issues in Transportation - Volume 2: Technical Papers. Washington, DC: The National Academies Press. doi: 10.17226/23299.
×
Page 36
Page 37
Suggested Citation:"UNDERSTANDING TRAVEL ISSUES." National Academies of Sciences, Engineering, and Medicine. 2005. Research on Women's Issues in Transportation - Volume 2: Technical Papers. Washington, DC: The National Academies Press. doi: 10.17226/23299.
×
Page 37
Page 38
Suggested Citation:"UNDERSTANDING TRAVEL ISSUES." National Academies of Sciences, Engineering, and Medicine. 2005. Research on Women's Issues in Transportation - Volume 2: Technical Papers. Washington, DC: The National Academies Press. doi: 10.17226/23299.
×
Page 38
Page 39
Suggested Citation:"UNDERSTANDING TRAVEL ISSUES." National Academies of Sciences, Engineering, and Medicine. 2005. Research on Women's Issues in Transportation - Volume 2: Technical Papers. Washington, DC: The National Academies Press. doi: 10.17226/23299.
×
Page 39
Page 40
Suggested Citation:"UNDERSTANDING TRAVEL ISSUES." National Academies of Sciences, Engineering, and Medicine. 2005. Research on Women's Issues in Transportation - Volume 2: Technical Papers. Washington, DC: The National Academies Press. doi: 10.17226/23299.
×
Page 40
Page 41
Suggested Citation:"UNDERSTANDING TRAVEL ISSUES." National Academies of Sciences, Engineering, and Medicine. 2005. Research on Women's Issues in Transportation - Volume 2: Technical Papers. Washington, DC: The National Academies Press. doi: 10.17226/23299.
×
Page 41
Page 42
Suggested Citation:"UNDERSTANDING TRAVEL ISSUES." National Academies of Sciences, Engineering, and Medicine. 2005. Research on Women's Issues in Transportation - Volume 2: Technical Papers. Washington, DC: The National Academies Press. doi: 10.17226/23299.
×
Page 42
Page 43
Suggested Citation:"UNDERSTANDING TRAVEL ISSUES." National Academies of Sciences, Engineering, and Medicine. 2005. Research on Women's Issues in Transportation - Volume 2: Technical Papers. Washington, DC: The National Academies Press. doi: 10.17226/23299.
×
Page 43
Page 44
Suggested Citation:"UNDERSTANDING TRAVEL ISSUES." National Academies of Sciences, Engineering, and Medicine. 2005. Research on Women's Issues in Transportation - Volume 2: Technical Papers. Washington, DC: The National Academies Press. doi: 10.17226/23299.
×
Page 44
Page 45
Suggested Citation:"UNDERSTANDING TRAVEL ISSUES." National Academies of Sciences, Engineering, and Medicine. 2005. Research on Women's Issues in Transportation - Volume 2: Technical Papers. Washington, DC: The National Academies Press. doi: 10.17226/23299.
×
Page 45
Page 46
Suggested Citation:"UNDERSTANDING TRAVEL ISSUES." National Academies of Sciences, Engineering, and Medicine. 2005. Research on Women's Issues in Transportation - Volume 2: Technical Papers. Washington, DC: The National Academies Press. doi: 10.17226/23299.
×
Page 46
Page 47
Suggested Citation:"UNDERSTANDING TRAVEL ISSUES." National Academies of Sciences, Engineering, and Medicine. 2005. Research on Women's Issues in Transportation - Volume 2: Technical Papers. Washington, DC: The National Academies Press. doi: 10.17226/23299.
×
Page 47
Page 48
Suggested Citation:"UNDERSTANDING TRAVEL ISSUES." National Academies of Sciences, Engineering, and Medicine. 2005. Research on Women's Issues in Transportation - Volume 2: Technical Papers. Washington, DC: The National Academies Press. doi: 10.17226/23299.
×
Page 48
Page 49
Suggested Citation:"UNDERSTANDING TRAVEL ISSUES." National Academies of Sciences, Engineering, and Medicine. 2005. Research on Women's Issues in Transportation - Volume 2: Technical Papers. Washington, DC: The National Academies Press. doi: 10.17226/23299.
×
Page 49
Page 50
Suggested Citation:"UNDERSTANDING TRAVEL ISSUES." National Academies of Sciences, Engineering, and Medicine. 2005. Research on Women's Issues in Transportation - Volume 2: Technical Papers. Washington, DC: The National Academies Press. doi: 10.17226/23299.
×
Page 50
Page 51
Suggested Citation:"UNDERSTANDING TRAVEL ISSUES." National Academies of Sciences, Engineering, and Medicine. 2005. Research on Women's Issues in Transportation - Volume 2: Technical Papers. Washington, DC: The National Academies Press. doi: 10.17226/23299.
×
Page 51
Page 52
Suggested Citation:"UNDERSTANDING TRAVEL ISSUES." National Academies of Sciences, Engineering, and Medicine. 2005. Research on Women's Issues in Transportation - Volume 2: Technical Papers. Washington, DC: The National Academies Press. doi: 10.17226/23299.
×
Page 52
Page 53
Suggested Citation:"UNDERSTANDING TRAVEL ISSUES." National Academies of Sciences, Engineering, and Medicine. 2005. Research on Women's Issues in Transportation - Volume 2: Technical Papers. Washington, DC: The National Academies Press. doi: 10.17226/23299.
×
Page 53
Page 54
Suggested Citation:"UNDERSTANDING TRAVEL ISSUES." National Academies of Sciences, Engineering, and Medicine. 2005. Research on Women's Issues in Transportation - Volume 2: Technical Papers. Washington, DC: The National Academies Press. doi: 10.17226/23299.
×
Page 54
Page 55
Suggested Citation:"UNDERSTANDING TRAVEL ISSUES." National Academies of Sciences, Engineering, and Medicine. 2005. Research on Women's Issues in Transportation - Volume 2: Technical Papers. Washington, DC: The National Academies Press. doi: 10.17226/23299.
×
Page 55
Page 56
Suggested Citation:"UNDERSTANDING TRAVEL ISSUES." National Academies of Sciences, Engineering, and Medicine. 2005. Research on Women's Issues in Transportation - Volume 2: Technical Papers. Washington, DC: The National Academies Press. doi: 10.17226/23299.
×
Page 56

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.

UNDERSTANDING TRAVEL ISSUES 98709mvpTxt 1_40 9/20/05 4:57 PM Page 1

98709mvpTxt 1_40 9/20/05 4:57 PM Page 2

3This study used the 2001 U.S. National House-hold Travel Survey (NHTS) to examine thetravel patterns of foreign-born adult women liv- ing in the United States and to compare their patterns with those of native-born women. Typically, foreign- born women and their households are different from those of the native born; they themselves are younger, less educated, and less likely to be employed than the native born. They are more likely to live in a rental property, have more members in the household, and live in a household with children. However, even when these demographic factors were controlled for, the authors found differences in the travel patterns of foreign- and native-born women. Foreign-born women are less likely to be drivers, but even among those who are drivers, foreign-born women are less likely to use an automobile and are more likely to use public tran- sit. Although foreign-born women live in larger house- holds, their households have fewer personal vehicles. In fact, a greater percentage of foreign-born women live in households with no vehicles. Both groups spent similar amounts of time traveling on the travel day, but foreign-born women took fewer trips and traveled fewer miles. Not surprisingly, foreign-born women take more international trips and travel with more household members. Foreign-born drivers are also more likely to be concerned about road conditions such as involvement in a traffic accident, highway con- gestion, and distracted drivers. The study findings clearly show important differences in travel behavior between foreign-born and native-born women, differ- ences that persist even when other salient variables are controlled for. Unfortunately, the data cannot indicate why these differences exist. Future research should focus on the impact on the travel patterns of foreign- born women of acculturation over time (i.e., the length of time spent in the United States, for which data are available in the 2001 NHTS) and answer several ques- tions: What will have a greater impact over time, the country of residence or citizenship or the country of birth? Do foreign-born women and men have compa- rable travel patterns, patterns that differ from those of native-born men and women, and if so, do these dif- ferences continue or converge over time? How influen- tial are standard socioeconomic variables like education, income, employment, and number and age of children? How much will the travel behaviors and patterns of aging foreign-born women mimic the greater culture of which they are a part and how much will they reflect the lifestyles these women left behind? Travel Characteristics of Native- and Foreign-Born Women in the United States Jonaki Bose and June Taylor Jones, Bureau of Transportation Statistics, U.S. Department of Transportation Abstract prepared by Sandra Rosenbloom, University of Arizona. 98709mvpTxt 1_40 9/20/05 4:57 PM Page 3

4Influence of Residential Location on Travel Behavior of Women in Chennai, India Sumeeta Srinivasan, Harvard University The visible impact of urban transportation is in access to employment. However, transportation also affects access to other services such as shopping and social service facili- ties. Past research in Chennai, a large city in India, indicates that the relocation of the very poor in peripheral informal settlements severely affects their accessibility to jobs and services because of the commuting distances involved when employment opportunities continue to remain highly cen- tralized. In this study an attempt was made to understand the influence of relative location within the city on travel behavior by using a sample of 116 low-income households from a variety of locations in Chennai. In particular, the travel behavior of women as affected by location was assessed. Models estimated to determine the influence of location characteristics on household travel behavior indi- cate that availability of transportation choices did affect the travel behavior of women even after differences in their life-cycle stage are accounted for. Recently, Chennai has been investing heavily in rail for public transportation without estimating current travel demand by spatial loca- tion within the city. The implications of this policy for inte- grated land use and transportation planning are especially pertinent in this context. The debate on the underlying relationships betweentravel demand and land use patterns continues inNorth America (1, 2). However, cities in develop- ing countries like India are growing rapidly, and land use and travel behavior interactions are of increasing impor- tance in planning for sustainable growth. Chennai is among the largest cities in India (3) and unlike some other cities in India has invested and continues to invest in public transportation. The current planning strategy in Chennai, described as a “minimally directed organic strategy to manage market-led development” (4), has not worked to the benefit of the low-income residents in the city. The 2001 census estimate suggests that low- income residents constitute over 25% of the total popu- lation of Chennai. Large investments in public transit like the Mass Rapid Transit System (the heavy-rail sys- tem under construction in Chennai) have failed to attract trips by these low-income households (both men and women) at the levels that were predicted (5). The poor continue to depend on nonmotorized modes and the rel- atively inexpensive modes like buses as their primary choice for travel to work (6). In this study, differences in travel behavior are investigated on the basis of differ- ences in accessibility to employment and the availability of transportation choices. In particular, the travel behav- ior of women in terms of frequency of trips, travel time, and mode choice is compared with the travel behavior of men. These differences in travel behavior have implica- tions not only for transportation planners who decide bus and train routes but also for land use planners who designate investments in infrastructure that will decide future employment growth. BACKGROUND Accessibility is an essential quality of cities. Substantial literature exists dating back at least 30 years on the def- 98709mvpTxt 1_40 9/20/05 4:57 PM Page 4

inition of accessibility (7). Accessibility can be defined as proximity to jobs, to shopping, to recreation, and ultimately to a decent quality of life. In this study, because of the lack of supplementary transportation studies for Chennai, accessibility is measured as average travel time to work and nonwork opportunities avail- able within one location relative to other locations in the city. Shen (8) estimates models (for the Boston metropolitan area in the United States) that suggest that an increase in general employment accessibility leads to a decrease in average commute time. Thangavel (9) finds that accessi- bility (to employment), average land value, social environ- ment, and average population density affect land develop- ment in peripheral Chennai. He suggests that accessibility is expected to play a greater role in shaping the urban structure of Madras (now known as Chennai). An earlier study in Chennai (6) found that women in a location with better accessibility were more likely to make more trips and travel farther for work trips. Travel behavior of resi- dents who are otherwise similar (in terms of socioeco- nomic status) is likely to be different if they live in locations with differing employment and transportation opportunities. Measures of travel behavior that have been studied in the United States and other developed countries include trip time, trip length, mode choice, trip fre- quency, and route choice (both spatial as well as time choice). A recent study in Seattle, Washington, by Limanond and Niemeier (10) suggests that land use patterns are associated with decisions about the type of shopping tours undertaken. Their study indicates that households with poorer accessibility tend to make fewer one-stop shopping tours and are more likely to combine nonwork trips with other trips and that households with greater accessibility are more likely to use nonautomobile modes for one-stop shopping tours. Thus, location affects both work and nonwork travel behavior. In a study in Jordan, Hamed and Olaywah (11) find that bus commuters are less likely to pursue social activities as compared with private vehicle com- muters. The choice of mode affects trip frequency, espe- cially in the case of low-income households because of the cost of travel. Gender-related accessibility disadvantages are also of concern in Chennai. Many activities performed by women (including child care, school drop-off, shopping, and jobs in the informal sector) are different from typical male activities. Women in both developed and develop- ing countries tend to make more trips than men, which are often more complex trip chains (12–14). In Chennai, a previous study (6) found that women conducted more trips and tended to use the least expensive mode (usually walking). Glick (15) notes in a West African study that women devote a substantial amount of time to house- hold work while engaging in income-earning activities, and the hours spent on these activities are outcomes of an optimization process in which allocations of time to “home work,” “market work,” and leisure are jointly determined. His study also notes that the cost of trans- portation to the city commercial center was key in affect- ing women’s entry into self-employment. It is in that context that this study tries to link the travel behavior of women to their relative location within the city. Context of Chennai Chennai has a population of 4.2 million (3) and is the capital city of the southern state of Tamil Nadu in India. The city itself (excluding the greater metropolitan area of over 1,000 km2) is spread over an area of 174 km2 that is administered by the Chennai Metropolitan Develop- ment Authority (CMDA). The larger metropolitan area includes a total population estimated to be about 6 mil- lion. Chennai is the fourth most populous city in India, and an estimated 1 million people live in shanty towns (or slums) in the city according to the most recent census. The city has had severe water and other infrastructure- related problems in recent years (16), especially in the peripheral locations. In many ways it is typical of rapidly growing urban areas in India. However, one way in which it diverges from some of the other large cities in India is that it also has a long tradition of investment in public transit. Data and Methodology Research on the travel behavior of those living in the cities of developing countries is scarce. Most studies tend to focus on limited data collected for large-scale trans- portation models (4). Further, these data are restricted to one point of time (1991 for Chennai) and do not include travel behavior variables at the individual level. Surveys are not conducted at regular intervals by local public agencies. Therefore the transportation data for Chennai, a rapidly growing city, are outdated, and primary sur- veys have to be conducted to obtain detailed data that can link location and travel behavior. Chennai is not unusual in this regard. As an example, Delhi (the capital of India) has not had a large-scale transportation survey for nearly 15 years. The data for this study come from a survey of 116 households (with a total of 509 persons conducting 1,862 trips) selected through geographically stratified sampling. The 41 geographical locations are based on a 1984 census conducted by the Slum Clearance Board (a public agency in Chennai) supplemented by the latest available census data for the city. The survey recorded 5INFLUENCE OF RESIDENTIAL LOCATION ON TRAVEL BEHAVIOR OF WOMEN IN CHENNAI, INDIA 98709mvpTxt 1_40 9/20/05 4:57 PM Page 5

one working day of the week for each household and included both work and nonwork activities conducted by the households. A separate location survey was also carried out that recorded the distance and time to travel to the nearest available services (schools, hospitals, etc.). These 116 households are from within the CMDA boundaries (174 km2) and exclude the larger metropol- itan area beyond the boundaries (about 1,000 km2). The data for this study are part of a larger sample that col- lected travel diaries for 160 low-income households from various locations in the Chennai metropolitan area. The focus on quantitative methods rather than qualitative methods for the study was mainly because of the cultural context. A qualitative study is less likely to influence the planning agencies in Chennai, especially transportation agencies, which are strongly dominated by engineers. LOCATION AND TRAVEL BEHAVIOR Ideally, travel behavior characteristics should be linked to location at the individual level. However, there were no publicly available spatial (geographic information system) data for Chennai. Therefore, an aggregated spa- tial location variable, or zone, was used to classify the location characteristics of Chennai. There are 10 zones in Chennai as designated by the local planning author- ity (the CMDA) aggregated from 155 census wards, which are the most disaggregated geographical unit (Figure 1). However, the CMDA zones are not represen- tative of relative accessibility to employment. For this study, the 155 wards were aggregated to seven zones based on their location within the city and the availabil- ity of employment and transportation opportunities. The modified zones were congruent with the 10 CMDA zones in the northern, central, and western sections of the city. However, they were different in the south (which has seen most of the population growth and pub- lic infrastructure investment in the past 10 years). The highly accessible central zones (Zones 6, 7, and parts of 8 and 10 of the CMDA zones) were aggregated into one zone (designated Zone 5). Table 1 indicates the differ- ences between households sampled from the CMDA zones and those from the modified zones. As mentioned earlier, accessibility is measured as average travel time to work and nonwork opportuni- ties. Average location characteristics are shown in Tables 2, 3, 4, and 5. There are marked differences in travel time to the major employment and commercial centers in Chennai among the zones (Table 2). Loca- tions in the central and western parts of Chennai have better bus-based accessibility than other zones do. Over- all, the northernmost zones (1 and 2) have the poorest accessibility to employment centers. This trend is appar- ent in services including schools, markets, and medical facilities (Table 3). Bus services are also dissimilar, with central, western, and southern locations having shorter walks to bus stops, better bus frequency, and larger numbers of bus route choices (Table 4). For this study, train-based accessibility was not examined because the 6 RESEARCH ON WOMEN’S ISSUES IN TRANSPORTATION FIGURE 1 Zones in Chennai as designated by CMDA (dot- ted boundaries and gray numbers indicate modified zone numbers) (map for thematic purposes and not to scale). TABLE 1 Number of Households, Persons, and Trips by CMDA Zone and Modified Zones CMDA Number of Number of Number of Modified Number of Number of Number of Zone Households Persons Trips Zone Households Persons Trips 1 10 44 115 1 (North) 10 44 115 2 19 96 296 2 (North) 19 96 296 3 19 79 306 3 (North West) 19 79 306 4 5 16 70 4 (West) 16 64 255 5 3 13 64 5 (Central) 21 87 353 6 14 65 258 6 (South) 14 66 330 7 5 13 57 7 (South) 17 73 409 8 3 14 59 Total 116 509 2064 9 7 30 100 10 31 139 739 Total 116 509 2064 98709mvpTxt 1_40 9/20/05 4:57 PM Page 6

sample showed limited use of trains (they are much more expensive modes than buses in Chennai and are limited in their availability). Table 5 suggests poor side- walk access overall except in the central zone and some older locations in the north. Household Characteristics The data indicate that over 80% of the households lived in rented housing (as perceived by the residents since owner- ship is not always clear). Housing stock in the north had relatively high ownership (as perceived by the household). In the peripheral zones of the north, west, and south, 60% of the families consisted of four or more people. In con- trast, the central zone had relatively small household sizes. Income did not show any distinct variations across zones in the sample. The northern zones had higher-than-overall proportions of the lowest-income households but also had higher-than-overall levels of the highest-income house- holds (Table 6). The southern zones had higher-than- overall proportions of the middle-income households. It should be noted that these ranges fall within an overall low-income range [less than 2,000 rupees (Rs) per month] for Chennai, where the per capita income was Rs 1,800 per month in 2003. Table 6 also suggests that the propor- tion of the sample employed was higher in the central zones than in the peripheral zones. Peripheral zones had relatively high vehicle ownership (Table 7), and the central zones had a higher proportion of households with no vehi- cles (not even bicycles). However, the families in the peripheral zones also tended to drive older vehicles as compared with those in the central zones. Travel Behavior Indicators This study focuses on the travel behavior of persons over the age of 15, since they make most of the work- and non-work-related trips. The indicators of time and 7INFLUENCE OF RESIDENTIAL LOCATION ON TRAVEL BEHAVIOR OF WOMEN IN CHENNAI, INDIA TABLE 2 Average Travel Times by Bus from Zones to Various Centers in Chennai To To To To To To To T Nagar Nungumbakkam Ambattur Luz Anna Salai Purusawakkam Parrys From Zone (South) (South) (West) (Central) (Center) (North) (North) 1 (North periphery) 115 80 96 78 68 75 49 2 (North) 75 90 127 57 67 60 52 3 (North Central) 58 51 62 49 39 26 31 4 (West) 36 38 47 41 28 38 34 5 (Central) 35 33 70 28 22 44 31 6 (South Central) 32 38 88 17 23 47 43 7 (South periphery) 40 44 98 33 45 56 54 Note: Times are given in minutes; shading indicates lower-than-average travel time. TABLE 3 Average Travel Times by Walking in Zones to Nearest Location Zone Grocery Store Primary School Temple Doctor Market Autorikshaw Stop 1 (North periphery) 12 20 18 18 26 16 2 (North) 13 22 16 22 28 14 3 (North Central) 11 26 14 13 18 12 4 (West) 11 23 13 14 14 13 5 (Central) 11 20 13 11 15 11 6 (South Central) 11 15 16 17 18 11 7 (South periphery) 11 21 10 11 13 11 Note: Times are given in minutes; shading indicates lower-than-average travel time. TABLE 4 Average Bus-Related Variables in Zones Average Bus Average Time by Walk Frequency Number of to Bus Stop Zone (Minutes) Bus Routes (Minutes) 1 (North periphery) 15 2 18 2 (North) 24 2 17 3 (North Central) 20 7 17 4 (West) 10 14 14 5 (Central) 11 12 13 6 (South Central) 9 10 12 7 (South periphery) 12 11 13 Note: Shading indicates lower-than-average travel time and higher- than-average number of bus routes from zone. TABLE 5 Safety-Related Variables by Zones Locations Locations Locations with with with Sidewalk Safe Walk Sidewalk to to School to School Market Zone (%) (%) (%) 1 (North periphery) 12.5 12.5 37.5 2 (North) 16.7 25 25 3 (North Central) 7.1 0 21.4 4 (West) 12.5 0 12.5 5 (Central) 33.3 25 8.3 6 (South Central) 11.1 0 22.2 7 (South periphery) 16.7 8.3 41.7 98709mvpTxt 1_40 9/20/05 4:57 PM Page 7

cost are among the most basic measures of travel behav- ior because they indicate the utility of the activities to the trip makers. The relationship between travel to work and accessibility is of immediate concern to transporta- tion planners in Chennai since they are in the process of designating employment zones and planning future transportation routes. Understanding mode choice is also important because the planners are in the process of designating funds for development of bus versus train routes. In all the zones, men tend to have higher travel costs than do women. However, the ratio of average travel cost of men and women is highest in the peripheral zones. Only the central zone shows parity in travel costs between the genders. Men in peripheral zones spend more than two times as much as women do in traveling to work. Overall men spend Rs 11.5 every day for their travel to work as compared with Rs 8 for women. Aver- age travel costs for men are highest in the north (Rs 20) followed by the central zone (Rs 16). The travel times for men followed the same pattern as the costs. How- ever, women in the peripheral zones tend to spend less on travel to work than the women in the central (Rs 16) and western zones (Rs 10). This finding suggests that both men and women on the periphery tend to have fewer work opportunities and transportation choices to get to the jobs in the business districts, but men are forced to spend more to get to work. Both men and women appear to stay close to home beyond a fixed cost threshold. Figure 2 indicates that men and women in the periph- eral zones (especially northern Zones 1, 2, and 3 and southern Zones 6 and 7) tend to walk more than those in the other zones do. However, their lack of transit choices (Table 4) and their higher-than-average travel time reflect the fact that walking is often the only mode available to them. Although men in the central zone also have higher-than-average travel times, a larger propor- tion travel by bus than do those on the periphery. In all zones women tend to walk more than men. Trip frequency also shows distinct differences based on location as well as gender. Women tend to make more trips then men do in all the zones since they per- form most of the household-related tasks: they conduct 76% of all grocery shopping trips, 85% of all drop-off trips, and all trips to fetch water. The ratio of male and female trips to work was almost equal in all the zones except in the northern and western periphery, where the number of work tours by men was over two times that by women. This finding follows the patterns indicated in the travel cost and time variations between genders. The ratio of shopping tours by women in the central zones was two times the number of shopping tours done by men. However, in the peripheral zones women con- 8 RESEARCH ON WOMEN’S ISSUES IN TRANSPORTATION TABLE 6 Household Characteristics: Income and Jobs Percentage Percentage in Each Income Group Women Men Zone Rs 0–1000 Rs 1000–2000 Rs 2000+ (Employed) (Employed) Difference 1 (North periphery) 30 30 40 42 80 38 2 (North) 21 47 32 33 90 57 3 (North Central) 53 47 0 69 93 24 4 (West) 40 20 40 52 92 40 5 (Central) 29 52 19 66 76 10 6 (South Central) 14 57 29 68 95 27 7 (South periphery) 12 71 18 53 91 38 Overall 29 48 23 55 88 33 Note: Shading indicates higher-than-overall averages. TABLE 7 Household Characteristics: Vehicle Ownership Percentage of Households Percentage of with Households Percentage of Two-Wheelers with Bicycles Households with Vehicles per Bicycles per Two-Wheelers Less Than 3 Less Than 3 Zone No Vehicles Household Household per Household Years Old Years Old 1 (North periphery) 40 1 0.4 0.6 17 0 2 (North) 53 0.6 0.4 0.1 100 38 3 (North Central) 74 0.3 0.3 0 0 20 4 (West) 50 0.6 0.4 0.2 100 67 5 (Central) 62 0.4 0.3 0.1 100 67 6 (South Central) 64 0.6 0.4 0.2 67 33 7 (South periphery) 47 0.8 0.6 0.1 0 30 Overall 57 0.6 0.4 0.1 59 38 Note: Shading indicates higher-than-overall averages. 98709mvpTxt 1_40 9/20/05 4:58 PM Page 8

ducted over five times the number of shopping tours as did men. Further, peripheral zones (especially the north- ern zone) tend to have a higher proportion of trips devoted to work (47% of all the trips conducted in the northern zone as compared with 33% in the central zone). In the central zones, the proportion of social or recreational trips was higher than that of shopping or drop-off trips. In the peripheral zones (1, 2, and 7) most of the nonwork trips were shopping or drop-offs. Figure 3 summarizes some of the disparities in showing that women tend to have higher trip frequencies than do men in all the zones but a higher proportion of men and women in the peripheral zones make two or fewer trips a day. These patterns are modeled in the next section in order to assess the statistical significance of location in affecting the travel behavior of an area’s residents. Modeling Travel Behavior Travel Time Travel time (in minutes) for trips made by those who work was regressed over several independent variables (Table 8). The model had an adjusted R2 of 31%. The results indicate that activity time (the number of min- utes spent at the activity for which the trip was being made) was significant in estimating travel time. Several location variables were also significant. Bus trips made by workers living on the periphery took 14 min longer than did private-vehicle trips, confirming the lack of bus service. Walk trips (which have to be restricted to shorter distances) were about 8 min less than private- vehicle trips. Thus a peripheral worker given the cur- rent transit choices would probably turn to private transportation with a rise in personal income. In con- trast, bus trips made by male workers living in the cen- ter were about 10 min shorter than private-vehicle trips were, whereas walk trips were about 23 min longer. All work trips made by workers living on the periphery were longer than the trips made by workers living in the central areas (trips from the west were about 17.5 min longer, whereas trips originating from the north or south were about 7 min longer). This find- ing suggests that employment, like public transporta- tion choices, is not evenly distributed over the city. Work trips made by women living in the central zones were about 7.5 min longer than work trips made by women living in the periphery. As discussed in the sec- tion on travel behavior indicators, women in the cen- tral zones have better opportunities to find work and tended to use the public transportation choices that were available to them. 9INFLUENCE OF RESIDENTIAL LOCATION ON TRAVEL BEHAVIOR OF WOMEN IN CHENNAI, INDIA FIGURE 2 Travel behavior characteristics: proportion of persons who conduct work and school trips in each mode choice category. 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% F M F M F M F M F M F M F M 1 1 2 2 3 3 4 4 5 5 6 6 7 7 Walk Bicycle Rickshaw Bus Train 3 wheeler 2 wheeler Shared 3 wheeler 98709mvpTxt 1_40 9/20/05 4:58 PM Page 9

Trip Frequency Trip frequency was categorized as follows: less than average (fewer than three), average (three to four), and higher than average (over four trips). Nearly 50% of the persons in the sample made an average number of trips per day, and the rest were evenly divided into the other two categories: Trip Frequency No. of Trips Percentage of Trips Less than average 66 23 Average 143 50 Higher than average 75 27 A discrete choice model (17) was estimated by regressing trip frequency choice on socioeconomic and location variables (Table 9). The household characteris- tics that were significant were the number of vehicles in the household and the income level. Persons in house- holds with more vehicles were more likely to have an average or higher-than-average number of trips per day. Women living in the central and southern areas were likely to have higher-than-average trip frequency. Also, the coefficient in the central zone was much higher than that for the southern zone, confirming that trip frequen- cies for women in the central zones are higher. Men who lived in the north were significantly likely to have a less- than-average trip frequency, which is perhaps related to 1 0 RESEARCH ON WOMEN’S ISSUES IN TRANSPORTATION 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% F M F M F M F M F M F M F M 1 1 2 2 3 3 4 4 5 5 6 6 7 7 Number of trips 0–2 Number of trips 3–4 Number of trips 5–6 Number of trips 7+ FIGURE 3 Travel behavior characteristics: proportion of persons who conduct work and school trips in each trip frequency category. TABLE 8 Regression of Travel Time in Minutes of Trips Made by Persons over Age 15 Who Work Independent Variable Estimated Coefficient t-Statistic P-value Constant 22.56** 15.48 0.00 Activity time (minutes) 0.01** 2.44 0.01 Middle income –0.55 –0.35 0.72 Bus trips from the periphery 14.21** 5.14 0.00 Walk trips from the periphery –7.80** –4.42 0.00 Work trips from north 7.09** 2.57 0.01 Work trips from south 7.59** 3.24 0.00 Work trips from west 17.47** 4.69 0.00 Men in center who take the bus (work trip) –9.71** –3.09 0.00 Men in center who walk (work trip) 23.13** 7.03 0.00 Women in center (work trip) 7.50** 2.21 0.03 Women in west (work trip) –3.04 –0.52 0.60 Note: N = 588; R2 = 32.4%; R2 (adjusted) = 31.1%; F = 25.1 (P = 0.00). **Significant at the 5% level. 98709mvpTxt 1_40 9/20/05 4:58 PM Page 10

their longer travel times. The adjusted r2 for this model was, however, very low (0.1). Mode Choice Mode choice by workers over the age of 15 was also estimated within a discrete choice framework by using a multinomial logit (MNL) model with four choices: walk, bicycle, bus or train, and a fourth category called “gasoline operated,” which included two-wheeled, three-wheeled, and shared three-wheeled vehicles. About 40% of the trips made by the persons over 15 in the sample who worked were by walking, about 10% were bicycle trips, about 40% were bus or train trips, and about 10% of the trips were by gasoline-operated vehicles: Mode No. of Trips Percentage of Trips Walk 297 42 Bicycle 65 9 Bus or train 270 38 Gasoline- operated vehicles 74 11 Table 10 shows the results of the MNL mode choice model. Travel time for the trip was significant and neg- ative, indicating that choices involving the least travel time were made. Travel cost was not significant although it had the expected negative sign, indicating the choice of the least travel cost. Households without vehicles were significantly more likely to use walk or transit for their trips. Work trips made by women living in the periphery as well as at the center were likely to be walk-based trips. However, the coefficient was much higher for working women living in the center, which may indicate better availability of job opportunities within walking distance. Men living in the periphery were significantly more likely to walk, although the coefficient was smaller than the coefficient for women. Men living in the central zones were significantly more likely to use transit. The adjusted r2 for this model was relatively high (0.45) although the location variables accounted only for a 0.02 increase in the adjusted r2. However, the significance of location-related variables suggested similar patterns in conjunction with other measures of travel behavior. SUMMARY OF RESULTS The better distribution of employment opportunities and wider range of transportation choices within the central zone in Chennai improved the transportation choices available to the woman residents who were employed. It also allowed both men and women to con- duct activities besides maintenance activities (work, shopping, and drop-off trips). Further, it was more likely that residents of the central zone could walk to these nonwork activities, unlike the residents on the periph- ery. Living in the central zone also allowed for more parity in the travel costs and times between men and women. This finding could also be linked to the fact that the women were now able to bring in more of the household share of earnings. Some of these differences 1 1INFLUENCE OF RESIDENTIAL LOCATION ON TRAVEL BEHAVIOR OF WOMEN IN CHENNAI, INDIA TABLE 9 Discrete Choice Model of Trip Frequency Category of Persons over Age 15 Who Work Independent Variable Estimated Coefficient Standard Error t-Statistic Constant (less than average trip frequency) 0.56* 0.29 1.88 Constant (average trip frequency) 0.96** 0.19 5.05 Number of vehicles in household (less than average trip frequency) –0.51** 0.26 –1.99 Household income relatively high (less than average trip frequency) –0.09 0.37 –0.26 Household income relatively low (less than average trip frequency) –1.05** 0.39 –2.64 Person lives in the central zone and is female (higher than average trip frequency) 1.69** 0.47 3.57 Person lives in the northern zone and is female (average trip frequency) –0.25 0.41 –0.61 Person lives in the southern zone and is female (higher than average trip frequency) 0.69* 0.38 1.78 Person lives in the central zone and is male (average or below average trip frequency) –0.28 0.48 –0.58 Person lives in the northern zone and is male (less than average trip frequency) 0.74* 0.39 1.87 Person lives in the southern zone and is male (less than average trip frequency) 0.08 0.42 –0.19 Note: r2 = 0.10; N = 284; percent correctly predicted = 51.8%. **Significant at the 5% level. *Significant at the 1% level. 98709mvpTxt 1_40 9/20/05 4:58 PM Page 11

in travel behavior are probably linked to the fact that households in the central zone are at a different life- cycle stage. However, the relatively job-rich southern zones (with larger families with children) also tended to have travel behavior more like that of the central zone than of the job-poor northern zones. Other studies of the travel patterns of low-income communities and women in the United States (18) indi- cate that employment opportunities are affected not only by distance from the city center but also by the travel times and availability of transit. In the case of Chennai, better public transportation in the central and southern zones appears to help both men and women in getting to work even if the jobs are not located close to them. Even though the northern peripheries have some employment opportunities, they are restricted to fewer jobs by the lack of inexpensive transportation choices. Further, even the relatively job-poor locations in the western and southern periphery appear to be able to use the availability of buses to their advantage. Although local planning agencies (including the Slum Clearance Board) have been less interested in the travel behavior of low-income women when the data are presented quali- tatively, the estimation of statistical models has gener- ated some interest among local planners. IMPLICATIONS FOR POLICY Indian cities like Chennai continue to have highly cen- tralized employment. This factor can be an advantage if public transportation to the center of the city is given priority and land use planners focus on sustaining the high densities at the center. Chennai has recently invested in heavy rail at the high cost of over Rs 7 bil- lion to improve the transit-based accessibility of the southern periphery and its connections to the central zone and the northern zones (19). However, the new ele- vated railway is not patronized by low-income travelers for several reasons: fares are much higher than current bus fares, the fares are not integrated with bus fares, and there are no connections to bus service at any of the stops. Middle-income travelers do not use the train either because there are no park-and-ride facilities and the stations are badly linked with bus and taxi services. Integrating the railway lines and the bus service instead of having them compete with each other would help improve overall transit-based accessibility in Chennai. As in many other developing cities (20), the general focus of transportation policy in Chennai has been on improving travel times for automobiles through mar- ginal improvements to roads. Planners have tended to focus on mitigating congestion by mitigating traffic con- ditions at selected locations and building large over- passes that only increase overall congestion in the system. Policy makers have not investigated changes in the structure or distribution of employment and other services. Several trends in vehicle ownership are also worrying in that the lack of public transportation and jobs will create more dependency in the periphery on private forms of transportation. As incomes rise this dependency may lead to higher two-wheeler ownership in locations that are not designed for their use. Many of the peripheral residents in Chennai, regardless of income, have no choice other than three-wheelers and two-wheelers for travel to work. Also, regardless of gender, the lack of integration of transit and land use planning has meant that job oppor- tunities are unevenly distributed throughout the city. The northern periphery in Chennai has been the worst affected by the lack of investment in infrastructure; such investment could lead to more employment opportuni- ties in that zone. The better infrastructure in the south- ern and western periphery has been able to attract more middle-income and upper-income residents. The infor- mal sector jobs that the low-income residents need occur in locations where the upper-income residents live. In this situation, the planners in Chennai need to intervene rather than let “minimally directed market-led forces” dictate the future urban structure of the city. 1 2 RESEARCH ON WOMEN’S ISSUES IN TRANSPORTATION TABLE 10 MNL Model of Mode Choice of Trips Made by Persons over Age 15 Who Work Independent Variable Estimated Coefficient Standard Error t-Statistic Constant (walk) 3.66** 0.34 10.71 Constant (bicycle) 2.78** 0.32 8.79 Constant (transit) 3.05** 0.25 12.12 Cost –0.003 0.004 –0.78 Time –0.06** 0.006 –10.10 Zero-vehicle household (walk and transit) –0.57** 0.24 –2.38 High-income household (transit or gasoline operated) 0.32 0.29 1.08 Work trip made by female living on the periphery (walk) 0.95** 0.36 2.66 Work trip made by female living in the center (walk) 2.07** 0.60 3.45 Trip made by male living in the periphery (walk) 0.55* 0.29 1.89 Trip made by male living in the center (transit) 0.84** 0.41 2.05 Note: r2 = 0.45; N = 706; percent correctly predicted = 79.6%. **Significant at the 5% level. *Significant at the 1% level. 98709mvpTxt 1_40 9/20/05 4:58 PM Page 12

ACKNOWLEDGMENTS The author thanks Josephine Edward and Angel Anthony for their assistance in conducting the survey. This research was funded by the American Institute for Indian Studies, the Harvard University Asia Center, and the Milton Fund of Harvard University Medical School. REFERENCES 1. Cervero, R., and J. Landis. The Transportation–Land Use Connection Still Matters. Access, Fall, 1995, pp. 3–9. 2. Badoe, D. A., and E. J. Miller. Transportation–Land- Use Interaction: Empirical Findings in North America and Their Implications for Modeling. Transportation Research, Vol. 5D, 2000, pp. 235–263. 3. Census of India: 2001 Census Results. www.census india.net. Accessed June 2005. 4. Master Plan for Madras (Chennai). CMDA, Chennai, India, 1991. 5. Mitric, S. Urban Transport in Chennai and Bangalore: Energy and Infrastructure Unit. South Asia Regional Report. World Bank, Washington, D.C., 2004. 6. Srinivasan, S., and P. Rogers. Travel Behavior of Low- Income Residents: Studying Two Contrasting Locations in the City of Chennai, India. Journal of Transport Geography, in press (2005). 7. Bhat, C., S. Handy, K. Kockelman, H. Mahmassani, Q. Chen, and L. Weston. Development of an Urban Accessibility Index: Literature Review. Center for Transportation Research, University of Texas, Austin, 2000. 8. Shen, Q. Location Characteristics of Inner-City Neighborhoods and Employment Accessibility of Low Wage Workers. Environment and Planning B, Vol. 25, 1998, pp. 345–365. 9. Thangavel, C. An Empirical Estimation of the Effect of Some Variables on Land Subdivision in Madras. Urban Studies, Vol. 37, 2000, pp. 1145–1156. 10. Limanond, T., and D. A. Niemeier. Effect of Land Use on Decisions of Shopping Tour Generation: A Case Study of Three Traditional Neighborhoods in Washington. Transportation, Vol. 31, 2004, pp. 153–181. 11. Hamed, M. M., and H. H. Olaywah. Travel-Related Decisions by Bus, Servis Taxi, and Private Car Commuters in the City of Amman, Jordan. Cities, Vol. 17, 2000, pp. 63–71. 12. Brun, J., and J. Fagnani. Lifestyles and Locational Choices—Tradeoffs and Compromises: A Case Study of Middle Class Couples Living in the Ile de France Region. Urban Studies, Vol. 31, 1994, pp. 921–934. 13. Astrop, A. The Urban Travel Behavior and Constraints of Low-Income Households and Females in Pune, India. In Women’s Travel Issues: Proceedings from the Second National Conference, October 1996, FHWA, U.S. Department of Transportation, 1997, pp. 215–246. 14. Helling, A. The Effect of Residential Accessibility to Employment on Men’s and Women’s Travel. In Women’s Travel Issues: Proceedings from the Second National Conference, October 1996, FHWA, U.S. Department of Transportation, 1997, pp. 147–163. 15. Glick, P. Simultaneous Determination of Home Work and Market Work of Women in Urban West Africa. Oxford Bulletin of Economics and Statistics, Vol. 61, 1999, pp. 57–83. 16. Dahiya, B. Peri-Urban Environments and Community Driven Development: Chennai, India. Cities, Vol. 20, 2003, pp. 341–352. 17. Ben-Akiva, M., and S. Lerman. Discrete Choice Analysis: Theory and Application to Travel Demand. MIT Press, Cambridge, Mass., 1985. 18. Blumenberg, E., and M. Manville. Beyond the Spatial Mismatch: Welfare Recipients and Transportation Policy. Journal of Planning Literature, Vol. 19, 2004, pp. 183–205. 19. CMDA Home Page. www.cmdachennai.org/english/ projects/mrtsph1.htm. Accessed Dec. 2004. 20. Turner, J., M. Grieco, and E. A. Kwakye. Subverting Sustainability? Infrastructural and Cultural Barriers to Cycle Use in Accra. World Transport Policy and Practice, Vol. 2/3, 1996, pp. 18–23. 1 3INFLUENCE OF RESIDENTIAL LOCATION ON TRAVEL BEHAVIOR OF WOMEN IN CHENNAI, INDIA 98709mvpTxt 1_40 9/20/05 4:58 PM Page 13

1 4 Comparing Women’s and Men’s Morning Commute Trip Chaining in Atlanta, Georgia, by Using Instrumented Vehicle Activity Data Hainan Li, Randall Guensler, and Jennifer Ogle, Georgia Institute of Technology Differences between women’s and men’s morning com- mute trip-chaining patterns are examined by using a subset of instrumented vehicle activity observations for 10 days of morning commute journeys made by 182 drivers from 138 households in Atlanta, Georgia. Morning commute trips that involve trip chaining are longer both in distance and in duration for both men and women compared with morning commutes without trip chaining. On the basis of analysis of the Atlanta data reported, overall gender differences in the morning commute trip-chaining patterns for men and women appear to exist. Men traveled a greater distance and spent more time in the morning commute than did women. Men stopped more frequently than women, and women tended to have shorter stop durations than did men. Some of the findings contradict previous research. It is not clear whether the differences reported here are specific to Atlanta, to the households involved in the sample, or perhaps to the specific time frame in which the analyses were undertaken. A larger sampling of the instrumented vehicle data (1 year of commute travel for 250+ households in the Commute Atlanta project) is currently being prepared to further assess these differ- ences and to examine whether gender roles may be changing, at least in Atlanta. Empirical evidence in previous research efforts indi-cates that a secondary role of the commute jour-ney is to provide an opportunity to link nonwork travel with the commute itself (1). Commuting trips are becoming increasingly complex as workers incorporate personal, household, and child-care activities into their trips (2). Since women’s participation in the labor force is at an all-time high, many working women fulfill house- hold and family responsibilities as well as their work duties. Given the gender roles in many households, women’s commute patterns are potentially different from men’s and may be affected by a typically greater share of household and family responsibilities. These differences in commute behavior may also vary depending on their socioeconomic and life-cycle status. This study compares men’s and women’s morning commute trip-chaining pat- terns by using a subset of instrumented vehicle activity observations. LITERATURE REVIEW Previous research (3–6) indicates that women are more likely than men to trip-chain on the way to and from work. On the basis of the 1990 Nationwide Personal Transportation Survey (NPTS) data, Strathman and Dueker (5) found that women make stops on their way to and from work or during work 42% of the time, whereas men make stops 30% of the time. Wegmann and Jang (7) examined the trip-chaining behavior of workers and developed nine work-related trip-chaining patterns from the 1990 NPTS data. They found that women have a higher total number of trip-chaining activities per day than men. Yet they did not find signif- icant differences in the amount of home-to-work trip chaining of men and women. 98709mvpTxt 1_40 9/20/05 4:58 PM Page 14

On the basis of the 1990 NTPS data, Strathman and Dueker (5) found that the trip purpose “other family/personal business” is the most heavily repre- sented in chains that are made both to work and from work. Wegmann and Jang (7) compared activity types pursued by men and women during morning commute trip chaining. They found that family and personal busi- ness trips and school or church trips account for 60% to 70% of the morning commute trip-chaining activities. Men and women made a comparable number of work- related business trips, shopping trips, school or church trips, and trips to visit friends or relatives. In these stud- ies, men made significantly more other social and recre- ation trips compared with women, and women made significantly more other family and personal business trips compared with men. McGuckin and Murakami (8) compared the trend of trip-chaining patterns noted between 1995 and 2001 by using the 1995 NPTS and 2001 National Household Travel Survey (NHTS) data. Thus research identified a robust growth in trip chaining that occurred between 1995 and 2001, nearly all in the direction of home to work. Men increased their trip chaining more than women, and a robust amount of the increase was due to stops for coffee. On the basis of trip purpose in the 2001 NHTS data, for workers who made stops on the way to work, the most common type of trip embedded in the home-to-work chain was a serve-passenger trip (33%), followed by fam- ily or personal business (16%) and stops for a meal or cof- fee (14%). In families in which both parents worked on weekdays, 61.3% of the trips to drop off a child were made by women compared with 38.7% made by men. Gender effects on trip chaining may differ across households in different life-cycle stages. Strathman et al. (9) determined that certain household types contributed the largest amounts of peak-period trip-chaining behav- ior. Single adults with young children have the highest propensity to form complex trip chains on the way to and from work, followed by single adults with school- age children, dual-income couples without children, and dual-income couples with preschoolers. Working moth- ers are more likely to link trips than working fathers (and they are more likely to link trips when the children are younger). On the basis of 1982 and 1985 data from France, the Netherlands, and the United States, Rosen- bloom (4) determined that 65% of working women with children under 6 years old linked trips to work, whereas only 42% of comparable men did. From the literature review, it appears likely that a dif- ference exists in commute trip-chaining behavior between men and women and among different house- hold structures. In previous studies, women were more likely to trip-chain on the way to and from work com- pared with men, and women made more serve-passenger trip chains compared with men. Previous research results were mostly based on household travel surveys. How- ever, one of the problems with household travel surveys is misreporting, as revealed in previous research (10–14). Advancements in Global Positioning System (GPS) technology provide a new method for multiday data col- lection for travel diary studies and other transportation applications. On the basis of the summary by Pendyala (15), GPS technologies capture travel behavior better during a long period of time and eliminate the survey fatigue problem of the multiday travel diary survey. GPS-based travel data can capture short and infrequent trips that may not be obtained in a traditional travel diary survey. Yalamanchili et al. (16) compared the trip- chaining indications provided by the GPS data with those provided by the recall data. Results of their study show that the GPS-based data performed in a manner superior to the recall data in capturing multistop chains in that the former captured more than twice as many multistop chains as the latter when comparisons were made in the context of a 1-day travel period. On the basis of the GPS study carried out in the California statewide household travel survey, Zmud and Wolf (14) found, on an aggregated level, that travel survey data underreport 27.4% of trips compared with the GPS- measured data. Especially for short-duration trips (between 0 and 10 min), 70.9% of the trips captured by GPS technology were missed in the travel survey. GPS-BASED VEHICLE ACTIVITY STUDIES The data used in this study were taken from the Georgia Institute of Technology Commute Atlanta project. This project instrumented approximately 487 vehicles from 268 representative households in the 13-county Atlanta metro- politan area with event data recorders (EDRs). The EDR provides an accurate itinerary of vehicle trips, including those short, intermediate, and infrequent stops that would otherwise be missed with traditional travel diary data col- lection methods. The network of EDR-equipped vehicles logs more than 2 million vehicle-seconds of activity each day. The research team collected second-by-second speed and position data for more than 600,000 trips during the first 10 months of the project. The Commute Atlanta research included standard household sociodemographic interviews and the collection of standard 2-day travel diaries (via computer-assisted telephone interview meth- ods) for the participating households (17). SAMPLE SUMMARY Ten days’ worth of morning commute journeys for 182 drivers from 138 households make up the data subset 1 5WOMEN’S AND MEN’S MORNING COMMUTE TRIP CHAINING 98709mvpTxt 1_40 9/20/05 4:58 PM Page 15

used for the analyses presented here. To meet the research goal of this study, only the 182 drivers whose gender information was known and who work full time at a fixed location and do not share their vehicle with another household member were included in the data subset. Significantly fewer lower-income households meet all of these conditions. The household recruitment strata used in the Commute Atlanta project and the sub- set of these households used in the analyses reported here are provided in Table 1. The recruitment process and study refusal rates are detailed elsewhere (17). For the data subset employed in the gender-based analyses reported here, the average household size is 2.86 persons. The average age of the drivers is 43. Most of the drivers have resided at their current residence loca- tion for more than 3 years, indicating a good level of familiarity with their travel areas. The respondents are divided fairly equally between men and women, with 49.5% being men. Children less than 16 years of age are present in 52 households (70 commuters) and children 5 years or younger are present in 20 households (25 com- muters). The ratio of workers per household is 1.45, which is comparable with 1.37 from the U.S. census data in 2000 for the Atlanta Metropolitan Statistical Area (MSA). Household vehicle ownership of the sample is higher than the average value in the 2000 census for the Atlanta MSA (2.37 vehicles per household compared with the 1.8 vehicles per household). This difference is expected since the objective of the project is to determine effects of by-the-mile congestion pricing on commute travel behavior, and only households that own vehicles were recruited. At least 55% of the drivers have either undergradu- ate or postgraduate educations, and the median house- hold income of the sample is between $75,000 and $99,000. Household income in the sample is signifi- cantly higher than the median household income of the Atlanta MSA ($51,948 in the 2000 census) because of higher-than-expected refusals and opt-outs of lower- income households and higher-than-expected retention of upper-income households (17). It may also be due to the fact that the commuters with white-collar occupa- tions usually have a higher salary and a fixed working schedule, whereas commuters with blue-collar occupa- tions who work in shifts may have commute schedules different from the traditional morning and afternoon peak periods. Hence household incomes for the com- muters identified during the morning peak periods are higher than those of the overall working population. The net result, however, is that upper-income house- holds and more educated individuals are overrepre- sented in the sample when compared with census demographic profiles of the Atlanta MSA population. Conclusions regarding behavior with demographics need to be restricted to each sample stratum in which sufficient data are available (see Table 1). The home address of each household and the work address of each worker were geocoded. The series of trips in which the first trip starts at home, the last trip ends at the workplace, and all intermediate trips are included that take place during the morning commute period (weekdays from 5:00 to 10:00 a.m.) on a given day is considered a single morning journey to work. Because drivers may or may not turn off the car’s engine when they stop, stops made during the morning com- mute were divided into two types. Engine-off stops take place when the driver turns off the engine during the stop; such trips are captured automatically in the data stream since one data file records activities between engine-on and engine-off stops. Occasionally, drivers will turn the engine on and off without moving and gen- erate a false trip. These false trips were screened out from the data set. Engine-on stops take place when the driver does not turn off the engine during the stop; these stops are detected by a script that examines the travel trace in detail. An engine-on stop is detected if the vehi- cle’s position falls outside of the 75-ft buffer of the road network and the vehicle speed is less than 5 mph for a duration longer than 1 min. A manual check of the detection results was tested against a set of sample trips. The algorithm detected the stops successfully under most situations. Figure 1 shows an example of one 1 6 RESEARCH ON WOMEN’S ISSUES IN TRANSPORTATION TABLE 1 Household Recruitment Strata Atlanta Household Households Households Sampling Annual Household Vehicles per Population Sample Recruited Used Strata Income Size Household (percent) Target No. (percent) (percent) 0 Any Any 0 7.4 0 0 (0%) 0 (0%) 1 <$30,000 Any 1+ 18.4 35–40 20 (7.46%) 4 (2.90%) 2 $30,000–$75,000 1 1+ 11.3 35–40 34 (12.69%) 17 (12.32%) 3 $30,000–$75,000 2+ 1 6.8 35–40 18 (6.72%) 7 (5.07%) 4 $30,000–$75,000 2 2+ 10.6 35–40 38 (14.18%) 13 (9.42%) 5 $30,000–$75,000 3+ 2+ 13.9 35–40 34 (12.69%) 14 (10.14%) 6 $75,000+ 1 1+ 2.8 0 5 (1.87%) 4 (2.9%) 7 $75,000–$100,000 2+ 1+ 12.1 35–40 41 (15.30%) 26 (18.84%) 8 $100,000+ 2+ 1+ 16.8 35–40 73 (27.24%) 51 (36.96%) 99 Unknown Any Any na 0 5 (1.87%) 2 (1.45%) Total 100 280 268 (100%) 138 (100%) 98709mvpTxt 1_40 9/20/05 4:58 PM Page 16

engine-off stop (at a daycare center) and one engine-on stop (at a video store). Among the 1,820 commute jour- neys from 182 commuters during the 10-day period analyzed, a total of 722 vehicle stops were detected in the sample. Among them, 460 were engine-off stops and 262 were engine-on stops. FINDINGS Commute Time and Distance Mean values of morning commute distance, travel time, and commute duration (stopping time plus travel time) of men and women in the sample are summarized in Table 2. The results of t-tests conducted to compare the mean val- ues between men and women indicate that men traveled longer distances and spent more time in their morning commute than did women. This result is consistent with previous research results summarized by Sarmiento (18). A previous study indicated that a large percentage of households’ total travel is undertaken in conjunction with the journey to and from work and that the growth of nonwork vehicle trips made during the commute con- tributes to traffic congestion (19). Work trips with non- work stops contribute to the vehicle miles and vehicle hours traveled in an urban area (20). In this study, t- tests of paired sample means (Table 3) indicate that for both men and women, commute journeys with trip chaining tend to be longer in distance than those with no chaining. However, trip chaining adds less distance to women’s morning commutes than to men’s. 1 7WOMEN’S AND MEN’S MORNING COMMUTE TRIP CHAINING FIGURE 1 Morning commute: trip-chaining example. TABLE 2 Gender Comparison of Average Commute Distance, Duration, and Travel Time Men Women Difference t-Statistics Significance (two-tailed) Travel distance (miles) 16.42 14.77 1.65 3.143 0.002 Commute duration (minutes) 40.80 36.02 4.78 3.319 0.001 Travel time (minutes) 32.05 29.58 2.47 2.872 0.004 98709mvpTxt 1_40 9/20/05 4:58 PM Page 17

Stop Frequency In the data subset, the research team detected slightly more stopping than was found in previous research. A total of 537 (30.5%) out of 1,820 morning commute journeys had one or more stops. Similarly, Hanson (21) found 29.4% of passenger vehicle trips having one or more stops between home and work. In a survey of 164 respondents, Mahmassani et al. (22) found that 24.3% of morning commute trips had one or more stops. On the basis of an empirical analysis with data from an activity survey conducted in the Boston metropolitan area and San Francisco Bay Area, Bhat and Singh (23) determined that 85.2% of the morning commute jour- neys had no stop, with the remaining 14.8% having one or more stops. Although commuting may be significantly different in Atlanta, GPS-based data collection methods may simply be more effective in capturing trip-chaining behavior. The frequency of nonwork stops during the morning commute by gender is shown in Figure 2. Of the 1,820 commutes, 90 men made 900 commutes and 92 women made the remaining 920 commutes. Chi-square test results at the 0.05 level indicate that in the sample, men are more likely to make one or more stops than are women. For each commuter, a stops ratio was calculated by dividing the number of commute journeys with stops by the total number of commute journeys for each driver. Among the 182 commuters, 50 never stopped, and the remaining 132 commuters (66 men and 66 women) stopped at least once during the 10-day period. Approx- imately one-third of the drivers stop during at least half of the commute journeys, and 5.49% of the drivers stop every day during their morning commute journey. These results indicate that making nonwork stops during the morning commute is a common phenomenon among a large percentage of commuters. The stops ratio during the morning commute grouped by gender is shown in Figure 3. Chi-square test results at the 0.05 level indi- cate that the men in the sample generally had a higher stop ratio than the women. Stop Locations In this study, stop locations were recorded in latitude and longitude format. For the 132 commuters (66 men and 66 women) who stopped at least once during the journey to work, the number of stop locations was com- pared across genders. If two stop locations were within 600 ft, they were considered to be the same. On aver- age, men commuters stopped at an average of 3.05 loca- tions compared with 2.86 for women. The t-test that assumes men and women have the same number of stop locations is not rejected at the 0.05 significance level. If the stop locations are divided into two groups— routine locations, at which a commuter stopped at least twice during the 10-day commute period, and nonrou- tine locations, at which a commuter only stopped once during the 10-day commute period—77 out of the 132 commuters stopped at routine locations. The male com- muters have an average of 0.89 routine stop locations compared with 0.92 for women. The t-test that assumes men and women have the same number of routine stop locations is not rejected at the 0.05 significance level. 1 8 RESEARCH ON WOMEN’S ISSUES IN TRANSPORTATION 0.00 10.00 20.00 30.00 40.00 50.00 60.00 70.00 80.00 0 1 2 3 4+ Number of stops per commute Percent of commutes Male Female FIGURE 2 Morning commute: number of stops by gender. 0.00 5.00 10.00 15.00 20.00 25.00 30.00 35.00 40.00 0 0.1–0.2 0.3–0.4 0.5–0.6 0.7–0.8 0.9–1 Stops ratio Percent of commuters Male Female FIGURE 3 Morning commute: stops ratio by gender. TABLE 3 Travel Distance Comparison of Commutes With and Without Trip Chaining With Without Trip Chaining Trip Chaining Difference t-Statistics Sig. (two-tailed) Distance (miles) 18.19 16.33 –1.8618 –4.996 0.000 Distance (miles) (men) 19.56 17.18 –2.1669 –3.280 0.002 Distance (miles) (women) 16.72 15.42 –1.5568 –4.484 0.000 98709mvpTxt 1_40 9/20/05 4:58 PM Page 18

Stop Duration Few previous studies examined trip-chaining stop dura- tions. In this study, the median durations for engine-off stops are 315 s for women and 375 s for men. The median durations for engine-on stops are 146 s for women and 150 s for men. The stop duration distribu- tion of men versus women is shown in Figure 4. Women tend to have shorter stop durations than men (finding significant at the 0.05 level by the chi-square test). CONCLUSIONS The research team conducted a cross-classification analysis of repeated behavioral data to examine the morning com- mute trip-chaining patterns for 182 men and women in Atlanta. This study employed a 10-day subset of on-road travel observations collected by GPS-equipped vehicles in the Commute Atlanta project. On the basis of the sample in this study, the existence of nonwork stops during the morning commute is a common phenomenon for both men and women. Some significant gender differences in morning commute trip- chaining patterns were noted in this analysis. For exam- ple, men traveled longer distances and spent more time in their morning commute than did women. Men also made more stops and stopped for longer durations than did women in morning commutes. However, the num- ber of stop locations did not differ significantly across genders. Some of the research findings here contradict previ- ous research results reported in the literature. Because the analytical results reported here are constrained to the household sample employed in the study (a higher presence of relatively affluent, car-owning households in the Atlanta commuting environment), it is not clear whether the differences identified also hold true for other sociodemographic groups and across regions. Hence, one should exercise caution in directly compar- ing the results in this paper with previous research results based on national travel surveys. However, it is also important to note that the Atlanta results did not rely on user-reported data but on revealed travel data collected by means of vehicle instrumentation. Hence, some of the differences may be associated with differ- ences in underreporting of travel by men and women and the characteristics of the trips that go unreported. Additional research into the underreporting issue is cur- rently under way in Atlanta through comparisons of instrumented vehicle data and travel diary data. Travel behavior of demographic groups is con- strained by different circumstances. Working women, in particular, often face constraints arising from their mul- tiple roles in the workplace and in the household. As the division of labor between men and women equalizes, corresponding changes in the division of household responsibilities should also occur. Although women continue to retain primary responsibility for house- work, the gap may be narrowing over time. One impor- tant piece of information that is missing in this study is the trip purposes for trip-chaining stops. Until this infor- mation is collected in the household travel diary surveys and until the parcel-level land use database is integrated into the analysis, it will be difficult to further evaluate the division of household labor between genders with this sample. However, once the new data are available, it will be possible to examine whether the differences reported here are likely due to increased sharing of household and family responsibilities between men and women workers in the same household, at least in Atlanta. A larger sampling of the instrumented vehicle data (1 year of commute travel for 250+ households in the Commute Atlanta project) is currently being prepared for more detailed analysis. More than 1 year’s instru- mented vehicle data have been collected in Atlanta. Such detailed commute data, over such a long period of time, have never been previously available to travel behavior researchers. As instrumented vehicle sampling programs become more pervasive and data are collected across multiple cities and in larger sociodemographic seg- ments, the research community will be able to expand and improve the core body of knowledge associated with trip-chaining behavior significantly. REFERENCES 1. Nishii, K., K. Kondo, and R. Kitamura. Empirical Analysis of Trip Chaining Behavior. In Transportation Research Record 1203, TRB, National Research Council, Washington, D.C., 1988, pp. 48–59. 2. Bianco, M., and C. Lawson. Trip Chaining, Childcare, and Personal Safety. In Women’s Travel Issues: Proceedings from the Second National Conference, 1 9WOMEN’S AND MEN’S MORNING COMMUTE TRIP CHAINING 0 10 20 30 40 50 60 70 80 90 £5 min 5–10 min 10–15 min 15–30 min >30 min Stop durations Percent of stops Male Female FIGURE 4 Stop duration (engine-on and engine-off stops combined) by gender. 98709mvpTxt 1_40 9/20/05 4:58 PM Page 19

October 1996, FHWA, U.S. Department of Transportation, 1997, pp. 124–142. http://www. fhwa.dot.gov/ohim/womens/wtipage.htm. 3. Rosenbloom, S. The Impact of Growing Children on Their Parents’ Travel Behavior: A Comparative Analysis. In Transportation Research Record 1135, TRB, National Research Council, Washington, D.C., 1987, pp. 17–25. 4. Rosenbloom, S. Trip-Chain Behaviour: A Comparative and Cross-Cultural Analysis of the Travel Patterns of Working Mothers. In Gender, Transport and Employment (M. Grieco, L. Pickup, and R. Whipp, eds.), Gower Publishing Company, Aldershot, England, 1989. 5. Strathman, J., and K. Dueker. Understanding Trip Chaining. In 1990 NPTS Special Reports on Trip and Vehicle Attributes, FHWA, U.S. Department of Transportation, 1995, pp. 1–27. 6. Al-Kazily, J., C. Barnes, and N. Coontz. Household Structure and Travel Behavior. In 1990 NPTS Demographic Special Reports, FHWA, U.S. Department of Transportation, 1995. 7. Wegmann, F. J., and T. Y. Jang. Trip Linkage Patterns for Workers. Journal of Transportation Engineering, May-June 1998. 8. McGuckin, N., and E. Murakami. Examining Trip- Chaining Behavior: Comparison of Travel by Men and Women. In Transportation Research Record: Journal of the Transportation Research Board, No. 1693, TRB, National Research Council, Washington, D.C., 1999, pp. 79–85. 9. Strathman, J. G., K. J. Dueker, and J. S. Davis. Congestion Management: Travel Behavior and the Use of Impact Fees. In Effects of Household Structure and Selected Travel Characteristics on Trip Chaining, Transportation Northwest (TransNow), Seattle, Wash., 1993, Vol. 1. 10. Willmot, C., and T. Adler. Item Nonresponse. In Transport Survey Quality and Innovation (P. Stopher and P. Jones, eds.), Elsevier, London, 2003, pp. 555–567. 11. Wolf, J., M. Loechl, M. Thompson, and C. Arce. Trip Rate Analysis in GPS-Enhanced Personal Travel Surveys. In Transport Survey Quality and Innovation, Elsevier, London, 2003, pp. 483–498. 12. Wolf, J., M. Oliveira, and M. Thompson. Impact of Underreporting on Mileage and Travel Time Estimates: Results from Global Positioning System–Enhanced Household Travel Survey. In Transportation Research Record: Journal of the Transportation Research Board, No. 1854 , Transportation Research Board of the National Academies, Washington, D.C., 2003, pp. 189–198. 13. Zmud, J. P., and C. H. Arce. Item Nonresponse in Travel Surveys: Causes and Solutions. In Transportation Research Circular E-C008: Transport Surveys: Raising the Standard, TRB, National Research Council, Washington, D.C., 2000, pp. II-D/20–II-D/34. 14. Zmud, J., and J. Wolf. Identifying the Correlates of Trip Misreporting: Results from the California Statewide Household Travel Survey GPS Study. Presented at 10th International Conference on Travel Behavior Research, Lucerne, Aug. 10–15, 2003. 15. Pendyala, R. Measuring Day-to-Day Variability in Travel Behavior Using GPS Data. Final Report. FHWA, U.S. Department of Transportation, 1999. 16. Yalamanchili, L., R. M. Pendyala, N. Prabaharan, and P. Chakravarthy. Analysis of Global Positioning System–Based Data Collection Methods for Capturing Multistop Trip Chaining Behavior. In Transportation Research Record: Journal of the Transportation Research Board, No. 1660, TRB, National Research Council, Washington, D.C., 1999, pp. 58–65. 17. Ogle, J., R. Guensler, and V. Elango. Commute Atlanta Value Pricing Program: Recruitment Methods and Travel Diary Response Rates. Presented at 83rd Annual Meeting of the Transportation Research Board, Washington, D.C., 2004. 18. Sarmiento, S. Household, Gender, and Travel. In Women’s Travel Issues: Proceedings from the Second National Conference, October 1996, FHWA, U.S. Department of Transportation, 1997. http://www. fhwa.dot.gov/ohim/womens/wtipage.htm. 19. Strathman, J., and K. Dueker. Understanding Trip Chaining. In NPTS Special Reports on Trip and Vehicle Attributes, FHWA, U.S. Department of Transportation, 1990, pp. 1–7. 20 Mahmassani, H. Dynamics of Commuter Behaviour: Recent Research and Continuing Challenges. In Understanding Travel Behaviour in an Era of Change (P. Stopher and M. Lee-Gosselin, eds.), Elsevier, London, 1996, pp. 279–313. 21. Hanson, S. The Importance of the Multi-purpose Journey to Work in Urban Travel Behavior. Transportation, Vol. 9, 1980, pp. 229–248. 22. Mahmassani, H., G. Hatcher, and C. Caplice. Daily Variation of Trip Chaining, Scheduling, and Path Selection Behaviour of Work Commuters. In Understanding Travel Behaviour in an Era of Change (P. Stopher and M. Lee-Gosselin, eds.), Elsevier, London, 1996, pp. 351–379. 23. Bhat, C., and S. Singh. A Comprehensive Daily Activity-Travel Generation Model System for Workers. Transportation Research, Vol. 34A, 2000, pp. 1–22. 2 0 RESEARCH ON WOMEN’S ISSUES IN TRANSPORTATION 98709mvpTxt 1_40 9/20/05 4:58 PM Page 20

2 1 Activities, Time, and Travel Changes in Women’s Travel Time Expenditures, 1990–2000 Rachel Gossen and Charles L. Purvis, Metropolitan Transportation Commission, Oakland, California This study examines the constancy and change in total travel time expenditures of women and men in the San Francisco Bay Area across the 10-year period from 1990 to 2000. The data sets analyzed are the 1990 and 2000 Bay Area Travel Surveys. Total travel time expenditures for women and men are examined across various sociodemographic and household attributes including age, race and ethnicity, employment status, and house- hold life-cycle category. The results show that for both women and men, reported daily travel time expendi- tures increased significantly from 1990 to 2000. In addi- tion, the results show that for some subgroups of women and men, differences in travel time expenditures have equalized from 1990 to 2000, whereas differences between other subgroups have increased. The evolution of the labor force, which includesmore women and working mothers than everbefore, has increased the interest in the travel behavior of women, particularly the unique needs, bur- dens, and patterns associated with women’s travel. A growing body of research has surfaced to address the variation in travel behavior between women and men, and from this research, several behavioral patterns have emerged. Results have been well documented over the past few decades, but as equality in the work force is approaching, is this translating over time into equality within the household? Specifically, are travel patterns equalizing between women and men? This study addresses these questions by examining the constancy and change in total travel time expenditures of women and men across the 10-year period from 1990 to 2000 in the San Francisco Bay Area. Studies on women’s and men’s travel have found many significant differences between the behavior and patterns of the two genders. The most pronounced find- ing is the increase of working women (and in particular, working mothers) in the labor force over the past few decades (1–3). However, this increase has not translated into an equal share of household maintenance and child- care activities between women and men, though the dis- parity has perhaps become less stark over the past 30 years (4–6). Specific to the San Francisco Bay Area, Tay- lor and Mauch (7) found that white, Hispanic, and low- income women were particularly burdened with household maintenance activities. Another consistent difference in the literature is that women typically have shorter trip durations but make more trips than men (8–10). In particular, women’s work commutes tend to be shorter (11, 12). Despite these consistencies, there is evidence of change. A few studies have shown that women and men are becoming more alike in their travel for certain markets and for certain trip characteristics. McGuckin and Murakami (13) found that single adult women and men without children are more similar than different in their travel, and Pucher and Renne (14) show that, at the aggregate level, women and men are becoming more alike in their travel on the basis of travel mode distribu- tions. Robinson and Godbey (15) report that from 1965 to 1985, total time spent on travel increased for both women and men, but women’s travel time was consis- 98709mvpTxt 1_40 9/20/05 4:58 PM Page 21

tently lower. For employed individuals, however, Robin- son and Godbey found that total travel time in 1985 for working women was actually longer than that for work- ing men. The current research extends these efforts to determine what changes have occurred in the Bay Area relative to travel time expenditures of women and men by using the 1990 and 2000 Bay Area Travel Surveys (BATS). METHODOLOGY AND DATA Two household travel surveys from the San Francisco Bay Area are used in this analysis to characterize and compare the constancy and change in women’s and men’s travel time expenditures: BATS1990 and BATS2000. The 1990 survey was trip-based and collected only weekday travel information from individuals aged 5 and over in more than 9,000 sample households. The most recent Bay Area household travel survey is BATS2000. More than 15,000 households participated. BATS2000 is an activity-based travel survey that collected informa- tion on all in-home and out-of-home activities over a 2- day period, including weekday and weekend pursuits. Unlike the 1990 survey, BATS2000 collected travel information from all members of the household, regard- less of age. For the purposes of this study, only individ- uals aged 5 and over are included. In addition, weekend, interregional, and external trips are excluded. For both data sets, survey results were weighted and expanded on the basis of census data, and trips were linked to produce the results discussed here. A detailed explanation of sample weighting, expansion, and trip- linking procedures may be found elsewhere (16, 17). As mentioned previously, only weekday travel within the nine-county Bay Area is reviewed. The result is 16.9 mil- lion trips made in 1990 by more than 5 million persons. Just over 51% of respondents were women, and they made nearly 52% of 1990 trips. BATS2000 includes 19.6 million trips made by 6.1 million individuals. Approximately 52% of BATS2000 participants were women, and they made over 53% of the trips in 2000. NOTE ON TRAVEL TIMES At the onset of this analysis, the most significant change found between the 1990 and 2000 surveys was in reported durations, which are significantly higher in the 2000 survey. Figure 1 shows that 30% of the trips in the 2000 survey were reported with durations greater than 30 min compared with only 17% of the 1990 trips. The average total travel time per person from the 2000 sur- vey is 92 min, a 48.5% increase from 1990 (see Table 1). This large increase in travel times was unexpected since past work by Kollo and Purvis (18) and Purvis (19) shows only modest increases in total travel times and average travel time per trip for the San Francisco Bay Area. Purvis (19) shows that increases in average trip dura- tions by trip purpose from 1981 to 1990 ranged from 7.8% to 11.0%. However, increases in average trip duration ranged from 23.0% to 62.0% from 1990 to 2000. Purvis also found that average total travel time per person decreased from 64 min in 1981 to 62 min in 1990 (a 3.7% decrease). In these examples, average travel time and total travel time are not increasing as significantly as the duration results of BATS2000 sug- gest. However, when BATS2000 data are compared with national surveys, the increase in travel times appears to be more reasonable. 2 2 RESEARCH ON WOMEN’S ISSUES IN TRANSPORTATION (a) (b) £15 min 57%15.1– 30.0 min 15.1– 30.0 min 26% >30 min 17% £15 min 43% 27% >30 min 30% FIGURE 1 Distribution of reported trip durations in (a) BATS1990 and (b) BATS2000. 98709mvpTxt 1_40 9/20/05 4:58 PM Page 22

Toole-Holt et al. (20) studied trends in the Nation- wide Personal Transportation Survey (NPTS) and the 2001 National Household Travel Survey (NHTS) and found that the amount of time spent on daily travel has steadily increased by approximately 2 min per person per year. If this rate is applied to total travel time expen- ditures in 1990 to estimate daily travel in 2000, the results are within 2 and 5 min of the reported daily travel times in BATS2000. In addition, the average travel time per capita reported in the 1990 NPTS for the San Francisco–Oakland–San Jose area was 60.1 min. Data from the 2001 NHTS show that average travel time per person per day for the Bay Area was 80.2 min. Although BATS2000 averages are still a bit higher than those of the NPTS and NHTS data, travel time expendi- tures in these surveys are comparable and make the BATS2000 duration data credible. Another factor that may be contributing to the signif- icantly higher travel times is the different survey method- ologies used in BATS1990 and BATS2000. It should be recalled that the 1990 survey was trip-based, whereas the 2000 survey was activity-based. Research suggests that intermediate stops are better captured with activity- based surveys (21). If this is indeed the case, the increase in recorded (or captured) trips would obviously add to the amount of travel time that individuals pursue on a daily basis. The difference between trip-based and activity-based surveys is an area that has not been ade- quately explored, and as noted by other researchers, studying the differences between these survey method- ologies is an area for future research (22). Since the heart of this research is an analysis of travel time expenditures, it is important that this substantial change in reported durations be considered. However, the authors believe that using the travel time informa- tion from the BATS2000 survey is valid to determine shifts in travel time expenditures that have (or have not) occurred between women and men from 1990 to 2000. TOTAL TRAVEL TIME EXPENDITURES Travel time expenditures for women and men are exam- ined in this paper by trip purpose and select sociodemo- graphic and household characteristics: age, race or ethnicity, employment status, and life cycle. This research began with an investigation of additional sociodemographic variables, including household income and vehicle availability. However, results for only those variables that showed the most significant trends are provided and discussed. Trip Rates, Total Travel Time, and Average Trip Time Total travel time, trips per capita, and average trip times for women and men in 1990 and 2000 are provided in Table 1. The difference between women’s and men’s trip rates increased slightly from 1990 to 2000; women aver- aged nearly 5% more trips per day than men in 2000. Total travel time per capita increased from 1990 to 2000 for both women and men by more than 20 min per per- son. In 2000, men spent only 3.3 min more per day on travel compared with a 6.5-min travel time gap between women and men in 1990. Average travel times for both men and women increased from 1990 to 2000 by approx- imately 7.5 min. However, the difference in average trip travel times between women and men remained relatively stable in the two survey years, with men traveling roughly 2 min longer per trip than women. Travel Time Shares by Trip Purpose Five different trip purpose categories are analyzed with relation to total travel time expenditures from 1990 to 2000: home-based work, home-based shop (other), home-based social-recreational, home-based school, and non-home-based. A more detailed description of the groupings used for each trip purpose follows. Home-based work, home-based school, and non- home-based trips are traditional trip-based definitions. Several activities are incorporated in the home-based shop (other) category such as shopping, household chores and personal care, sleep, personal services (banking, dry cleaning), time spent sick or at a medical appointment, nonwork or nonshop Internet use, picking up or drop- ping off passengers, or changing modes. Home-based social-recreational trips encompass activities such as meals, entertainment, hobbies, exercise, social activities, relaxing, volunteer work, and religious activities. 2 3ACTIVITIES, TIME, AND TRAVEL TABLE 1 Trip Rates, Total Travel Time, and Average Travel Time by Gender Percent Difference Between 1990 2000 Women and Men Women Men Women Men 1990 2000 Trips per capita 3.23 3.17 3.25 3.11 1.9% 4.5% ** Total travel time per capita (minutes) 65.1 71.6 90.3 93.6 –9.1% ** –3.5% ** Average trip time (minutes) 20.1 22.6 27.8 30.1 –11.1% ** –7.6% ** **Significant at the 0.01 level. 98709mvpTxt 1_40 9/20/05 4:58 PM Page 23

Figure 2 shows travel time shares by trip purpose for men and women in 1990 and 2000. Some interesting trends are evident. First, Figure 2 indicates that in 1990, men spent 10% more of their total travel time budgets on work trips than women did. In 2000, the difference in travel time work shares between women and men is still roughly 10%, but the share of travel time spent on home-based work trips decreased for both sexes. This decrease in the travel time shares for home-based work trips is balanced by an increase in the portion of travel time spent for social-recreational and shopping trips. The decrease in time spent on work trips does not imply that men and women are spending less time traveling for work. Average travel times for work trips increased for both women and men, and in 2000, women’s commute times were much more comparable with those of men. In 1990, men averaged 30 min between home and work, whereas women commuted for only 20 min. In 2000, men spent 39 min commuting, whereas women averaged a 36- min commute. These average commute times paired with the travel time shares shown in Figure 2 indicate that even though women in 2000 are commuting for almost the same amount of time as men, they still bear more of the burden for home-based shop (other) trips. Women spent roughly 9% more of their travel budget on home-based shop (other) activities than men in both survey years. It must be recalled that within the home-based shop (other) category are activities like household chores, shopping, child care, and serving passengers. This result reinforces the idea that women are disproportionately burdened with household maintenance and child-care responsibilities. Effects of Sociodemographic Variables Travel time expenditures are analyzed next by various sociodemographic characteristics and household attri- butes, which include age, race or ethnicity, employment status, and household life-cycle category. Gender as well as employment status are used as controlling factors in the analysis of each sociodemographic and household characteristic. The effects of each attribute on total travel time expenditures between working women and men and nonworking women and men are discussed in the following sections. Income and vehicle availability were also investigated, but the results were not as significant as those for the variables included in this section. 2 4 RESEARCH ON WOMEN’S ISSUES IN TRANSPORTATION Men Men (b)(a) Women Women 31% 41% 25% 36% 25% 16% 28% 20% 11% 10% 16% 15% 10% 9% 11% 9% 23% 24% 20% 20% NHB HBSC HBSR HBSH HBW NHB HBSC HBSR HBSH HBW FIGURE 2 Total travel time shares by trip purpose and gender: (a) 1990; (b) 2000. (HBW = home-based work; HBSH = home-based shop (other); HBSR = home-based social/recreational; HBSC = home-based school; NHB = non-home-based.) 98709mvpTxt 1_40 9/20/05 4:58 PM Page 24

Age The first sociodemographic variable reviewed relative to travel time expenditures is age of the trip maker. Table 2 provides results for working and nonworking women and men in eight age categories. Differences in travel time expenditures between working women and men equalized from 1990 to 2000, and in fact, these differences in travel time expenditures disappeared except for working women in their fifties, who still spent less time traveling per day than their male counterparts did. Travel gaps decreased primarily because of the increases in women’s travel. For almost all age groups, women’s daily travel time expenditures increased at a faster rate than those of men; in some cases, women’s travel times increased 1.5 to 2 times as much as men’s. Although age cohorts were not analyzed in this work, there appears to be an interesting trend with individuals who were in their forties in 1990 and in their fifties in 2000. Men in this group spent signifi- cantly more time traveling per day than their female counterparts did in both survey years. For nonworking women and men, there was no sig- nificant difference in travel time expenditures in 1990 for almost all age groups. However, in 2000, changes occurred for several age groups. Young girls reported more time traveling than young boys. Nonworking men in their late twenties spent significantly more time (nearly 30 min) on travel than did nonworking women in this age group, and in fact, these young men averaged the most on travel across all age and employment cate- gories. Another significant change from 1990 is that nonworking women between 40 and 59 years old spent more time traveling than their male counterparts did. The results in Table 2 suggest that, barring the 23-to-29 age group and individuals over 60, nonworking women spent more time on daily travel than unemployed men. For both survey years, workers spent more time on travel than nonworkers, but the nonworkers’ travel time expenditures increased at a higher rate so that the dif- ference in travel time between workers and nonworkers was less pronounced in 2000. Finally, the results in Table 2 suggest that children reported significantly less time on travel than adults in each survey year. To account for this finding and make a more appropriate comparison between workers and nonworkers, age is controlled for in the remainder of the analysis. Race or Ethnicity Travel time expenditures by five race or ethnicity cate- gories are explored in Table 3. As with Table 2, employ- ment status is considered; however, in the case of nonworkers, two groups are reviewed: adult nonworkers and nonworking children. For almost all race or ethnicity and employment cat- egories, there was no significant difference in travel time expenditures between women and men in either survey year. The most interesting results by race or ethnicity are 2 5ACTIVITIES, TIME, AND TRAVEL TABLE 2 Travel Time per Capita in Minutes by Gender, Employment, and Age Group Percent Difference Between 1990 2000 Women and Men Age Group Women Men Women Men 1990 2000 Workers, all ages 5–17 — — — — — — 18–22 65.1 76.1 98.1 94.0 –14.5% * 4.4% 23–29 74.5 76.0 103.3 102.6 –1.9% 0.7% 30–39 85.5 85.5 101.6 105.9 –0.1% –4.0% 40–49 82.9 90.0 110.1 112.2 –7.9% ** –1.9% 50–59 71.4 86.2 95.8 108.2 –17.1% ** –11.5% ** 60–64 64.9 86.6 103.7 106.0 –25.0% ** –2.1% 65–99 67.9 82.5 95.0 102.2 –17.7% –7.0% Nonworkers, all ages 5–17 41.3 39.4 66.2 60.8 4.8% 8.9% ** 18–22 58.5 58.1 84.0 76.5 0.6% 9.8% 23–29 57.5 64.9 82.2 112.0 –11.3% –26.7% ** 30–39 59.6 66.0 88.8 83.0 –9.8% 6.9% 40–49 60.4 50.2 97.3 84.9 20.3% 14.6% * 50–59 58.1 50.3 89.5 76.2 15.4% 17.4% * 60–64 49.0 51.4 91.2 99.5 –4.8% –8.4% 65–99 45.8 51.6 72.3 83.7 –11.4% * –13.6% ** Total, workers and nonworkers Workers, all ages 78.5 84.3 102.9 106.8 –6.9% ** –3.7% ** Nonworkers, age 18 and over 53.9 55.8 84.9 86.0 –3.4% –1.3% Nonworkers, age 17 and under 43.5 41.4 68.2 61.9 5.1% 10.2% ** Note: The dash represents values that could not be calculated and cells with no observations. *Significant at the 0.05 level. **Significant at the 0.01 level. 98709mvpTxt 1_40 9/20/05 4:58 PM Page 25

for Hispanic-Latino women and men. In 2000 working Hispanic-Latino men spent roughly 8 min more per day on travel than did working Hispanic-Latino women. This same trend was evident in 1990 for nonworking Hispanic-Latino women and men, but in 2000, the dif- ference in travel times for this group was negligible because of the increase in travel times for Hispanic- Latino women. In fact, daily travel expenditures for nonworking Hispanic-Latino women increased more than twice as much as those for nonworking Hispanic- Latino men. Household Life-Cycle Category The final sociodemographic factor used to compare travel time expenditures among different groups of women and men is life-cycle category. Household life- cycle categories in this study are based on categories used in the 2001 NHTS. Use of the life-cycle variable allows for the comparison of travel time expenditures for persons living alone, individuals without children, parents in single- and multiadult households with chil- dren of various ages, and retirees. Table 4 shows the dis- tribution of travel time per capita for working and nonworking adults by the 10 life-cycle categories. The results of the household life-cycle analysis show that employed men in multiadult households spent more time on travel than their female counterparts did in both survey years. Aside from this finding, working women and men in each of the different household types in 2000 spent approximately the same time on travel across all life-cycle groups. The exception is for single working mothers with very young children (<6). The sample of single working fathers with young children in 2000 was small (only 37 respondents) and not statisti- cally significant, but the travel time averages suggest that single working mothers with young children spent much more time traveling than single working fathers in this group. In addition, these single working mothers averaged between 10 and 20 min more time on daily travel than almost all other workers; the exception is single working fathers with school-age children, who averaged 117 min per day on travel. Table 4 shows that in general, nonworking adult women in households with children have higher travel time expenditures than do nonworking men in family households. In addition, in both survey years, nonwork- ing men living alone spent nearly 30 additional minutes per day on travel than did nonworking women living alone. This finding also holds for multiadult households in 2000, though the difference in average travel time is only 11 min between nonworking men and women. Retired women and men spent about the same amount of time traveling in each survey year. CONCLUSIONS Travel behavior research on the differences between women and men travelers has yielded interesting and fairly consistent results over the past two decades in gauging the effect of the surge of women in the work force. However, as new policies take effect and society adjusts to the increasing role of women in the labor force, these observed trends in travel behavior are likely 2 6 RESEARCH ON WOMEN’S ISSUES IN TRANSPORTATION TABLE 3 Travel Time per Capita in Minutes by Gender, Employment, and Race or Ethnicity Percent Difference Between 1990 2000 Women and Men Race/Ethnicity Women Men Women Men 1990 2000 Workers, all ages White, not Hispanic 79.1 86.9 103.8 106.3 –9.0% –2.3% Hispanic/Latino, any race 72.1 74.5 97.6 105.9 –3.2% –7.8% * Black/African American, not Hispanic 93.3 88.1 103.4 106.0 5.9% –2.5% Asian/Pacific Islander, not Hispanic 72.5 77.2 104.8 105.8 –6.1% –0.9% Other 75.1 80.5 101.0 120.6 –6.7% –16.3% ** Nonworkers, age 18 and over White, not Hispanic 57.3 58.7 86.8 89.7 –2.3% –3.2% Hispanic/Latino, any race 40.6 56.6 92.9 80.7 –28.2% * 15.2% Black/African American, not Hispanic 52.2 40.7 83.9 80.9 28.3% 3.8% Asian/Pacific Islander, not Hispanic 49.7 52.1 73.9 78.3 –4.5% –5.6% Other 40.7 47.8 91.6 86.0 –14.8% 6.6% Nonworkers, age 17 and under White, not Hispanic 42.9 42.0 64.5 64.1 2.2% 0.7% Hispanic/Latino, any race 38.5 38.6 71.7 53.3 –0.2% 34.4% * Black/African American, not Hispanic 54.1 51.7 88.3 79.0 4.7% 11.8% Asian/Pacific Islander, not Hispanic 44.0 36.7 63.8 56.8 19.9% 12.2% Other 46.4 43.9 60.4 65.1 5.7% –7.1% Total 65.1 71.6 90.3 93.6 –9.1% ** –3.5% ** *Significant at the 0.05 level. **Significant at the 0.01 level. 98709mvpTxt 1_40 9/20/05 4:58 PM Page 26

to change. This study examined the 1990 and 2000 BATS to determine which changes have occurred in the 10-year period from 1990 to 2000. Specifically, this study focused on the constancy and change in travel time expenditures of women and men from 1990 to 2000. The findings suggest that at aggregate levels, the dif- ferences in travel time expenditures between women and men have indeed decreased over time and are less pro- nounced in 2000 than they were in 1990. However, when women and men are stratified by various socioe- conomic and household variables, different patterns emerge. A few key points found in this study are as follows: • Reported daily travel time expenditures signifi- cantly increased from 1990 to 2000 for both women and men. • At the aggregate level, women in 2000 averaged slightly more trips per day than men. • From 1990 to 2000, the share of total travel time spent on home-based work trips decreased, whereas shares for home-based shop (other) and social-recreational trips increased for both women and men. • Women in 1990 spent 10 fewer minutes commut- ing to work than men. In 2000, however, women com- muted for nearly the same amount of time (36 min for women versus 39 min for men) and had a much higher increase in work-trip travel times (19 min versus a 6-min increase for men). • Except for women in their fifties, working women in 2000 across all age groups spent approximately the same amount of time on travel by all purposes as did men. • Although nonworking women and men in 1990 spent roughly the same amount of time traveling across dif- ferent age groups, nonworking women in 2000 generally spent more time traveling than nonworking men. • Working Hispanic-Latino men spent 8 more min- utes per day traveling in 2000 than did working Hispanic- Latino women. • Among workers, Hispanic-Latino women spent the least amount of time on travel. However, nonworking adult Hispanic-Latino women had the highest average travel time of nonworkers, and their travel times increased twice as much as those of nonworking Hispanic-Latino men. • Single working parents with young children spent more time on travel in 2000 than did women and men in other life-cycle groups. • Nonworking adult women in households with children traveled more than nonworking men in family households. The results of this study imply that for some sub- groups, women and men are beginning to approach more equal levels of travel time expenditures. That is, women’s travel time expenditures are increasing at a faster rate than men’s so that the travel time gap has lessened between 1990 and 2000. Further analysis 2 7ACTIVITIES, TIME, AND TRAVEL TABLE 4 Travel Time per Capita in Minutes by Gender, Employment, and Life-Cycle Category Percent Difference Between 1990 2000 Women and Men Life-Cycle Category Women Men Women Men 1990 2000 Workers, all ages Single adult, no children 88.4 92.5 107.1 104.8 –4.5% 2.2% Two or more adults, no children 74.7 81.2 99.5 106.9 –8.0% ** –6.9% ** Single adult, youngest child under 6 70.6 † 105.9 † 117.5 88.9 † –33.4% † 32.2% † Two or more adults, youngest child under 6 82.7 76.5 100.7 105.4 8.1% –4.5% Single adult, youngest child 6–15 89.3 93.3 107.9 117.3 –4.2% –8.0% Two or more adults, youngest child 6–15 82.5 88.1 107.6 111.4 –6.4% * –3.4% Single adult, youngest child 16–21 76.6 81.9 98.3 106.5 –6.5% –7.7% Two or more adults, youngest child 16–21 71.0 82.9 98.6 100.8 –14.3% ** –2.2% Single adult, retired, no children — — — — — — Two or more adults, retired, no children — — — — — — Nonworkers, age 18 and over Single adult, no children 61.0 90.7 77.2 106.3 –32.7% * –27.4% * Two or more adults, no children 50.6 54.4 78.7 89.5 –7.0% –12.1% ** Single adult, youngest child under 6 35.2 † 143.8 † 76.7 † 53.0 † –75.5% † 44.7% † Two or more adults, youngest child under 6 59.9 53.9 † 85.5 72.5 11.1% † 17.9% Single adult, youngest child 6–15 72.7 48.6 † 107.9 † 204.0 † 49.7% † –47.1% † Two or more adults, youngest child 6–15 60.0 55.9 97.7 83.2 7.2% 17.4% * Single adult, youngest child 16–21 44.5 41.7 † 108.3 † 58.9 † 6.9% † 83.7% † Two or more adults, youngest child 16–21 51.1 54.7 73.1 64.4 –6.6% 13.6% Single adult, retired, no children 55.5 54.8 82.3 92.9 1.4% –11.3% Two or more adults, retired, no children 49.9 54.0 83.8 85.6 –7.6% –2.1% Total 65.1 71.6 90.3 93.6 –9.1% ** –3.5% ** Note: The dash represents values that could not be calculated and cells with no observations. †Insufficient sample size (less than 50 individuals). *Significant at the 0.05 level. **Significant at the 0.01 level. 98709mvpTxt 1_40 9/20/05 4:58 PM Page 27

should be undertaken and more data sets should be ana- lyzed to determine whether the changes are unique to the Bay Area or are an artifact of comparing trip-based with activity-based surveys. It is hoped that this style of reporting travel time expenditures will be replicated for other national, statewide, and metropolitan travel sur- veys to gain a better understanding of how women and men spend their time. The cross-classifications exam- ined do show that differences in travel time expendi- tures are higher for some subgroups of women. Therefore, these data should be used to find ways in which these additional burdens can be alleviated. Areas of future research might include examining the effects of age cohorts. In addition, it would be beneficial to study the differences between activity-based and trip- based survey results. Finally, this research used simple means tests to compare women’s and men’s travel behavior; multivariate analyses would be useful in exploring the effects of different combinations of vari- ables. Clearly, this research shows that women’s travel is still on the rise, and as such, warrants further research and consideration. ACKNOWLEDGMENTS The current version of this paper reflects feedback from participants in the Conference on Research on Women’s Issues in Transportation. The authors thank these reviewers as well as Sandra Rosenbloom and three anonymous referees for valuable comments essential in the development of this work. REFERENCES 1. Atkins, S. Women, Travel and Personal Security. In Gender, Transport and Employment (M. Grieco, L. Pickup, and R. Whipp, eds.), Gower Publishing Company Limited, Aldershot, England, 1989, pp. 169–189. 2. Schor, J. The Overworked American: The Unexpected Decline of Leisure. BasicBooks, New York, 1992. 3. Hayghe, H. V. Women’s Labor Force Trends and Women’s Transportation Issues. In Women’s Travel Issues: Proceedings from the Second National Conference (S. Rosenbloom, ed.), Report FHWA-PL- 97-024, FHWA, U.S. Department of Transportation, 1997, pp. 9–14. 4. Hamilton, K., and L. Jenkins. Why Women and Travel? In Gender, Transport and Employment (M. Grieco, L. Pickup, and R. Whipp, eds.), Gower Publishing Company Limited, Aldershot, England, 1989, pp. 17–45. 5. Jones, P. Household Organisation and Travel Behavior. In Gender, Transport and Employment (M. Grieco, L. Pickup, and R. Whipp, eds.), Gower Publishing Company Limited, Aldershot, England, 1989, pp. 46–76. 6. Levinson, D. M. Space, Money, Life-Stage, and the Allocation of Time. Transportation, Vol. 26, No. 2, 1999, pp. 141–171. 7. Taylor, B. D. and M. Mauch. Gender, Race, and Travel Behavior: An Analysis of Household-Serving Travel and Commuting in the San Francisco Bay Area. In Women’s Travel Issues: Proceedings from the Second National Conference (S. Rosenbloom, ed.), Report FHWA-PL- 97-024, FHWA, U.S. Department of Transportation, 1997, pp. 371–406. 8. Hosking, D. Organising the Domestic Portfolio: Gender and Skill. In Gender, Transport and Employment (M. Grieco, L. Pickup, and R. Whipp, eds.), Gower Publishing Company Limited, Aldershot, England, 1989, pp. 115–126. 9. Robinson, J. Americans on the Road. American Demographics, Sept. 1989, p. 10. 10. Chapple, K., and R. Weinberger. Is Shorter Better: An Analysis of Gender, Race, and Industrial Segmentation in San Francisco Bay Area Commuting Patterns. In Women’s Travel Issues: Proceedings from the Second National Conference (S. Rosenbloom, ed.), Report FHWA-PL-97-024, FHWA, U.S. Department of Transportation, 1997, pp. 407–436. 11. MacDonald, H. I. Women’s Employment and Commuting: Explaining the Links. Journal of Planning Literature, Vol. 13, No. 3, 1999, pp. 267–283. 12. Turner, T., and D. Niemeier. Travel to Work and Household Responsibility: New Evidence. Trans- portation, Vol. 24, 1997, pp. 397–419. 13. McGuckin, N., and E. Murakami. Examining Trip- Chaining Behavior: Comparison of Travel by Men and Women. In Transportation Research Record: Journal of the Transportation Research Board, No. 1693, TRB, National Research Council, Washington, D.C., 1999, pp. 79–85. npts.ornl.gov/npts/1995/Doc/chain2.pdf. Accessed Aug. 31, 2004. 14. Pucher, J., and J. L. Renne. Socioeconomics of Urban Travel: Evidence from the 2001 NHTS. Transportation Quarterly, Vol. 57, No. 3, 2003, pp. 49–77. 15. Robinson, J. P., and G. Godbey. Time for Life: The Surprising Ways Americans Use Their Time. Pennsylvania State University Press, University Park, 1997. 16. Purvis, C. L. Sample Weighting and Expansion: Working Paper—Bay Area Travel Survey 2000. Metropolitan Transportation Commission, Oakland, Calif., June 2003. 17. Purvis, C. L. Trip Linking Procedures: Working Paper 2—Bay Area Travel Survey 2000. Metropolitan Transportation Commission, Oakland, Calif., June 2003. 2 8 RESEARCH ON WOMEN’S ISSUES IN TRANSPORTATION 98709mvpTxt 1_40 9/20/05 4:58 PM Page 28

18. Kollo, H. P. H., and C. L. Purvis. Changes in Regional Travel Characteristics in the San Francisco Bay Area: 1960–1981. In Transportation Research Record 987, TRB, National Research Council, Washington, D.C., 1984, pp. 57–66. 19. Purvis, C. L. Changes in Regional Travel Charac- teristics and Travel Time Expenditures in San Francisco Bay Area: 1960–1990. In Transportation Research Record 1466, TRB, National Research Council, Washington, D.C., 1994, pp. 99–110. 20. Toole-Holt, L., S. E. Polzin, and R. M. Pendyala. Two Minutes per Person per Day Each Year: Exploration of Growth in Travel Time Expenditures. In Transportation Research Record: Journal of the Transportation Research Board, No. 1917, Transportation Research Board of the National Academies, Washington, D.C., 2005. 21. Stopher, P. R. Use of an Activity-Based Diary to Collect Household Travel Data. Transportation, Vol. 19, 1992, pp. 159–176. 22. Mokhtarian, P. L., and C. Chen. TTB or not TTB, That Is the Question: A Review and Analysis of the Empirical Literature on Travel Time (and Money) Budgets. Transportation Research, Vol. 38A, 2004. 2 9ACTIVITIES, TIME, AND TRAVEL 98709mvpTxt 1_40 9/20/05 4:58 PM Page 29

3 0 Do High-Occupancy Toll Lanes Serve Women’s Travel Needs? Theresa M. Dau, Parsons Brinckerhoff This study examined differences in how womenand men perceived three high-occupancy toll(HOT) lanes in California: State Route 91 in Orange County, Interstate 15 in San Diego, and pro- posed lanes on Interstate 680 in Alameda County. The literature suggested that women travelers aged 30 to 50 were more likely than other groups to choose to use a toll road, possibly to allow them to better handle their complicated household and employment responsibili- ties. The study found that women were statistically more likely to use SR-91 HOT lanes more frequently than men and more often during off-peak hours; they were also statistically more likely to say that it saved them more than 30 min and to express satisfaction with the lanes. The study came to similar conclusions about I-15 in San Diego; women were more likely to use the HOT lane more frequently than men but less likely to drive alone in the lane; both men and women supported the use of the HOT lanes, but women supported them slightly more strongly. With regard to the proposed HOT lane in Alameda County, an opinion survey found that women were statistically more likely to support a demonstration project to test the concept than were men. Women were also slightly more likely to support certain proposed operational features such as using a FasTrak toll collection system, using toll revenues to support public transit services, and allowing carpools free use of the lane. A higher percentage of women also supported the idea of enhanced enforcement through electronic surveillance, video cameras, and greater Cal- ifornia Highway Patrol involvement. Additional research should focus on evaluating and comparing other HOT lane experiences; exploring equity issues related to women’s use of these lanes versus their abil- ity to pay; analyzing differences in perceptions and use by race, ethnicity, and immigration status; and identify- ing the specific reasons why women tend to use HOT lanes more frequently and evaluate them more highly. Abstract prepared by Sandra Rosenbloom, University of Arizona. 98709mvpTxt 1_40 9/20/05 4:58 PM Page 30

3 1 Gender Differences in Bicycling Behavior and Facility Preferences Kevin J. Krizek, Pamela Jo Johnson, and Nebiyou Tilahun, University of Minnesota This study focuses on bicycling and specifically the differ- ences by gender in terms of use and facility preferences. It is hypothesized that there are observable differences in bicycle use and how bicycle facilities are perceived; the researchers attempt to understand where differences exist and to document these differences in a manner that pro- vides a baseline for future research. Secondary data from five different surveys were used to examine actual cycling behavior (commuting and other), desired amenities, and safety perceptions, as well as cycling facility preferences of women versus men. In general, the research uncovered a number of differences between men and women but also several other important differences. For example, there are distinct gender differences in the purpose of bicycle trips, desired amenities and safety perceptions, and the degree to which separate facilities are valued. This work contributes to the planning, transportation, and public health (physical activity) literature by providing a quanti- tative baseline documentation on which to build future work on a specific but often-glossed-over topic. Travel researchers, transportation professionals,public health practitioners, and policy makershave been steadfast in encouraging increased rates of walking and bicycling. Although most trans- portation analysis aggregates these two modes, there are considerable differences between them in terms of use, facilities, and preferences. Even considering each mode independently, there exist differences across populations. It is unlikely that a single population of current (and potential) walkers or cyclists exhibits sim- ilar characteristics, uses, and preferences (1). Cycling use among youth may differ from that among adults, who differ from the elderly. Likewise, income levels and geographic areas certainly have a role. This study focuses on bicycling and specifically the differences by gender in terms of use and facility preferences. It is hypothesized that there are observable differences in bicycle use and how bicycle facilities are perceived; the researchers attempt to understand where differences exist and to document these differences in a manner that provides a baseline for future research. A key dimension to encouraging heightened bicycle use—for men or for women—is to understand the extent to which it is currently being employed, the purpose, and the preferences that affect its use. An extensive body of research identifies gender as an important predictor of travel. The focus of this litera- ture is relatively broad; most of it examines the journey to work. Little focuses on differences by mode, espe- cially cycling. Part of the difficulty in examining cycling behavior is that it is affected by myriad factors, includ- ing safety along a planned route, the need to carry goods, limitations imposed by schedule or attire, dis- tance, weather, risk, or the need to combine errands. Gender may affect how strongly such factors are weighed. Existing transportation and urban theory liter- ature related to travel and gender, however, offers only general insights to inform the thinking on cycling behav- ior, some of which leads to contradictory expectations. Existing research, for example, is unified in finding that women in the aggregate work closer to home. 98709mvpTxt 1_40 9/20/05 4:58 PM Page 31

Women have shorter commutes than men (2–4). Fur- thermore, they make fewer and shorter trips than men do (5). The consideration that distance is a dominant factor in deciding to bicycle suggests that it is easier for women to cycle to work than for men (6). Higher rates of adoption would be expected for women (7). In addi- tion, lower rates of employment, on average, suggest an increase in discretionary time. This increase would allow greater time for recreational activities, which may or may not result in increased cycling. In contrast, equally compelling reasoning suggests that women would have lower rates of cycling than men. Women typically shoulder typical household responsibilities (8); such trips require serving passen- gers, linking multiple errands, or carrying household goods (e.g., groceries). None are well suited for bicycle travel (9). Such generalities suggest that women would have fewer cycling trips than men. Furthermore, there may also be differences in motivation, attitudes, and preferences for travel between women and men (10). Cycling is well recognized as being among the riskiest of transportation modes (11); there is a considerable body of literature documenting women as tending to be more risk averse than men (12), suggesting lower rates of adoption. Which of the foregoing described theories likely hold true? A central problem in research on bicycle use is that information about cycling, much less about differences between women and men, is scarce. Anecdotal evidence suggests that men are more likely to cycle than women. However, the authors are aware of fewer than a handful of studies to confirm such evidence, hardly a reliable research base. Rodriguez and Joo (13) find that women have between 72% and 83% lower odds of using non- motorized modes than do men (though their analysis combines cycling and walking). Krizek and Johnson (14) conclude that women have 52% lower odds of making a bicycle trip. Cervero and Duncan (15) demonstrate that bicycle trips are more likely to be made by men. In terms of cycling distance, women have longer bicycle com- mutes (in terms of travel time) going from suburb to cen- tral business district and shorter ones for suburb to suburb (16). A relatively recent survey based in San Fran- cisco aimed to understand why low-income women do not ride bicycles (17). Different studies analyze different phenomena, ranging from rates of use to distance to rea- sons for use. Unfortunately, available data prevent robust analysis to reconcile such complexities reliably because cycling is a mode of transportation used by so few, at least in the United States. Its relatively rare use makes it extremely difficult to break down such data by gender, purpose, or geographic area. For a rigorous explanation of why a rare event occurs, a targeted survey design and instrument, a relatively large survey, and a sample able to detect subtle differences are required. The aim in this study is to focus on gender and cycling and document such relationships by exploiting secondary data sources. By using the foregoing studies as a springboard, it is hypothesized that rates of cycling are greater for men than for women—for all types of trips. Furthermore, it is theorized that men make longer cycling trips than women because women typically work closer to home. It is also theorized that part of the reason why women bicycle less is because men are less affected by inferior cycling facilities (e.g., cycling in traf- fic). To shed light on these hypotheses, results are described by using an analysis from five different sur- veys. Each survey is based in Minneapolis, Minnesota, and the accompanying region,1 except for the National Household Travel Survey (NHTS), which is relied on to offer a general perspective. The core of this paper lies in two different analysis sections. The first reports on findings from three differ- ent surveys measuring revealed behavior (two travel sur- veys and the U.S. census). All three data sets are relied on to focus on rates of cycling by gender, commute mode share, and differences between city and suburb. The second analysis section switches to explaining stated-preference data from two other surveys. The first stated-preference survey focuses on cycling infrastruc- ture preferences and safety perceptions; the second is an adaptive stated-preference survey examining the value of different types of bicycle facilities. The central pur- pose throughout this pilot study is to provide baseline information about how different types of bicycle use and facility preference differ by gender and to direct more concentrated work in this area. Employing a com- bination of surveys (revealed behavior and stated pref- erence) helps to establish a stronger empirical base for continued dialogue and future research concerning the unique needs and preferences of women cyclists. REVEALED BEHAVIOR ANALYSIS OF CYCLING BEHAVIOR Data Sources of Revealed Behavior To examine rates of cycling behavior, three comprehen- sive surveys of revealed behavior were used. The first is the 2001 NHTS, which aims to collect a sample of the nation’s daily travel. The survey includes demographic characteristics of households, people, and vehicles, and detailed information on daily travel for all purposes by all modes.2 The other two data sets used in this part of the analysis focus on the Twin Cities, Minnesota, met- ropolitan area. One is the 2000 Twin Cities metropoli- tan area Travel Behavior Inventory (TBI), which contains individual and household-level demographic data as well as travel behavior characteristics for a sam- 3 2 RESEARCH ON WOMEN’S ISSUES IN TRANSPORTATION 98709mvpTxt 1_40 9/20/05 4:58 PM Page 32

ple of Twin Cities metropolitan area residents.3 The other data set is the 5% Public Use Microdata Sample (PUMS) from the 2000 U.S. census. The aim in this part of the analysis is to uncover gen- der differences in cycling across three dimensions: the overall frequency of all cycling trips, commute-only behavior, and cycling behavior of urban versus subur- ban residents by gender. However, each of the foregoing surveys has limitations in its ability to shed light on these questions. For example, the PUMS is a large sam- ple but only reports on commuting. The NHTS includes all trips, but issues of confidentiality prevent detailed analysis of geographical attributes for either sample. The TBI, although it focuses only on the Twin Cities, includes all trips and allows geographical precision but is based on a relatively small sample size. In the follow- ing discussion, trip purpose is examined by using the NHTS, work commute is compared across all three data sources, and city and suburb differences are examined by using the TBI. Employing all three helps provide a comparative picture of relevant differences. Trip Purpose The initial observation on looking at the NHTS data rein- forces the earlier statement that cycling is a relatively rare activity. On average, a mere 0.4% of all reported trips for adults in the United States is made by bicycle.4 Examining how these trips break down by gender reveals interesting differences. Controlling for the number of overall trips within each gender group shows that men are more than twice as likely to complete their trip by bicycle than women (0.66% versus 0.25%). Breaking down cycling trips further by gender and purpose reveals the following differences, each of which is statistically significant at the p = .01 level. Men are more likely to bicycle to work than women (10.2% of men’s cycling trips versus 6.24% of women’s cycling trips) and to bicycle for rest and relax- ation (2.14% versus 0.79%). Conversely, however, women are more likely than men to ride a bicycle to school as a student (1.2% versus 0.58%), to do shopping and errands (2.64% versus 1.11%), and to visit friends and relatives (4.53% versus 2.76%). All other purposes of travel did not reveal statistically significant differences across gender. Work Commute Focusing only on the work commute allows comparison of results across all three surveys. The NHTS provides a national overview. The PUMS and TBI are examined for only Minneapolis to compare similar geographies. Table 1 shows the prevalence and duration of cycling trips for employed persons in each of these surveys. Although there is considerable variation for each measure across the data sets, the consistent pattern shows women’s rates of cycling to be less than those of men. However, only the NHTS and the PUMS revealed the differences to be statistically significant at the p = .01 level. Mean duration in cycling commute times reveals no statistically significant differences; most times did not differ by more than 90 s. Of particular interest here is the ability to compare different survey instruments and samples for exactly the same geographic area (Min- neapolis). Although the differences between cycling rates of men and women are similar in the PUMS and TBI, it is interesting to note that the TBI survey results in higher measures for both prevalence and distance. City Versus Suburb Differences in cycling according to urban or suburban residence, focusing on the Minneapolis–St. Paul region, are examined next. The analysis is restricted to the TBI and to the behavior of those who indicated they had 3 3GENDER DIFFERENCES IN BICYCLING BEHAVIOR TABLE 1 Prevalence and Duration of Cycling Commute Trips by Gender for Employed Persons Data Source Women Men Total Cycling commute frequency in % (n) NHTS (national)1, 3 0.23% (119,659) 0.75% (460,612) 0.51% (580,271) TBI (Minneapolis)2 4.37% (15) 5.80% (21) 5.11% (36) PUMS (Minneapolis)2, 3 1.04% (33) 2.90% (96) 1.99% (129) Commute trip time in minutes (sd) NHTS (national)1 14.28 (9.04) 15.44 (14.78) 15.20 (12.31) TBI (Minneapolis)2 22.93 (12.78) 21.57 (17.25) 22.13 (15.36) PUMS (Minneapolis)2 15.36 (7.35) 16.95 (9.92) 16.55 (9.32) 1 Includes weighted sample of full and part-time workers. 2 Denominator includes only those who are employed residents of Minneapolis (TBI, n = 705; PUMS, n = 6,476). 3 The NHTS and PUMS commute frequencies are the only gender differences shown to be statistically significant: chi-square = 63.16, p = 0.00 and chi-square = 117.24, p = 0.00. 98709mvpTxt 1_40 9/20/05 4:58 PM Page 33

completed a cycling trip during the survey (among those in the TBI, n = 142). Men and women cyclists vary little across sociodemographic characteristics. However, when they are stratified by urban or suburban residence, some gender differences emerge (Table 2).5 The last two rows of Table 2 present summary char- acteristics of cycling behavior for TBI cyclists by gender and household location. Overall, the mean number of bicycle trips in a day is only slightly lower for women than for men, whereas the mean distance traveled by bicycle is about a half kilometer lower for women. This pattern differs for urban residents and suburban resi- dents, however. For the urban residents (defined as resi- dents of Minneapolis or St. Paul), women cyclists traveled nearly a kilometer more than men, whereas the mean distance traveled by bicycle for suburban women was nearly 3 km less than that for suburban men (sub- urban was defined as anyone in the seven-county region except those in Minneapolis or St. Paul). Most striking, however, are the gender differences in the purpose of bicycle trips. For the urban population, 63% of women cyclists made a work- or school-related (i.e., commute) bicycle trip compared with 38% of male cyclists. Conversely, in the suburbs, only 11% of women cyclists made a commute trip compared with 25% of men. In contrast to commute trips, gender differences for recre- ation trips are reversed. In other words, 13% of urban women cyclists made a recreational trip compared with 21% of men. In the suburbs, more women cyclists made a recreational trip compared with men (50% versus 31%). 3 4 RESEARCH ON WOMEN’S ISSUES IN TRANSPORTATION TABLE 2 Characteristics of TBI Cyclists by Gender and Household Location in Seven-County Metropolitan Area, Minnesota Twin Cities Suburbs Total Women Men Women Men Women Men n % n % n % n % n % n % 30 34% 58 66% 18 33% 36 67% 48 34% 94 66% Age category 18–29 years 11 37% 22 38% 5 28% 3 8% 16 33% 25 27% 30–49 years 14 47% 30 52% 6 33% 21 58% 20 42% 51 54% ≥50 years 5 17% 6 10% 7 39% 12 33% 12 25% 18 19% Educational attainment 4-year college degree or more 21 70% 44 76% 11 61% 19 53% 32 67% 63 67% Less than 4-year degree 9 30% 14 24% 7 39% 17 47% 16 33% 31 33% Employment status Employed 30 100% 48 83% 11 61% 28 78% 41 85% 76 81% Unemployed 0 0% 10 17% 7 39% 8 22% 7 15% 18 19% Household income <$50,000 14 47% 34 59% 5 28% 13 36% 19 40% 47 50% $50,000– $74,999 12 40% 9 16% 5 28% 10 28% 17 35% 19 20% ≥$75,000 3 10% 11 19% 3 17% 8 22% 6 13% 19 20% Missing 1 3% 4 7% 5 28% 5 14% 6 13% 9 10% Other cyclist in household Yes 8 27% 7 12% 7 39% 9 25% 15 31% 16 17% No 22 73% 51 88% 11 61% 27 75% 33 69% 78 83% Bicycle trip purpose Work commute 19 63% 22 38% 2 11% 9 25% 21 44% 31 33% No 11 37% 36 62% 16 89% 27 75% 27 56% 63 67% Work or school commute 19 63% 28 48% 3 17% 9 25% 22 46% 37 39% No 11 37% 30 52% 15 83% 27 75% 26 54% 57 61% Recreation/ fitness 4 13% 12 21% 9 50% 11 31% 13 27% 23 24% No 26 87% 46 79% 9 50% 25 69% 35 73% 71 76% Cycling Behavior Characteristics Mean S.D. Mean S.D. Mean S.D. Mean S.D. Mean S.D. Mean S.D. Number of bicycle trips 2.80 (1.40) 2.98 (1.96) 2.06 (0.64) 2.33 (1.60) 2.52 (1.22) 2.73 (1.85) Distance (km) by bicycle 9.13 (5.87) 8.22 (7.71) 6.05 (6.62) 8.75 (11.46) 7.97 (6.26) 8.43 (9.34) Note: Age ≥18 years. 98709mvpTxt 1_40 9/20/05 4:58 PM Page 34

STATED-PREFERENCE ANALYSIS OF CYCLING FACILITIES Omnibus Data In the second part of the analysis, differences were exam- ined by using results from stated-preference surveys, which were obtained from two sources. The first is the Minnesota Department of Transportation Statewide Omnibus Study 2003–2004, which provides data on preferences for cycling facility infrastructure and on perceptions of cycling. The Omnibus data were originally collected by a telephone sur- vey from a random sample of Minnesota residents 18 years or older.6 The preference variables represent importance ratings of cycling facility infrastructure characteristics such as paved shoulders, lighting on bicycle paths, and bicycle racks on buses. The safety perception variables represent general themes that emerged from open-ended responses provided by subjects who reported that Minnesota was less than “very safe” for cyclists. The Omnibus data shed light on two general phe- nomena: gender differences with respect to (a) desired amenities and facilities among current and potential cycling commuters and (b) perceptions of safety for cycling. The Omnibus sample of cyclists is nearly evenly distributed on gender (49% women and 51% men). Only about one-fourth (28%) are central city residents (Minneapolis or St. Paul zip codes), whereas 72% live in the suburbs (Table 3). Because so few Omnibus cyclists are central city residents, all cycling facility infrastruc- ture characteristics reported here are for the pooled sample of urban and suburban cyclists. Among current and potential cycling commuters, few gender differences were noted with respect to amenities 3 5GENDER DIFFERENCES IN BICYCLING BEHAVIOR TABLE 3 Characteristics of Omnibus Cyclists by Gender and Household Location in Seven-County Metropolitan Area, Minnesota Twin Cities Suburbs Total Women Men Women Men Women Men 16 39% 25 61% 55 52% 50 48% 71 49% 75 51% Subject Demographics Age category 18–29 years 5 31% 2 8% 7 13% 8 16% 12 17% 10 13% 30–49 years 8 50% 12 48% 39 71% 31 62% 47 66% 43 57% ≥50 years 3 19% 11 44% 9 16% 11 22% 12 17% 22 29% Educational attainment 4-year college degree or more 12 75% 18 72% 30 55% 26 52% 42 59% 44 59% Less than 4-year degree 4 25% 7 28% 25 45% 24 48% 29 41% 31 41% Employment status Employed 13 81% 23 92% 42 76% 42 84% 55 77% 65 87% Unemployed 3 19% 2 8% 13 24% 8 16% 16 23% 10 13% Household income <$50,000 9 56% 8 32% 14 25% 14 28% 23 32% 22 29% $50,000– $74,999 3 19% 5 20% 11 20% 14 28% 14 20% 19 25% ≥$75,000 3 19% 10 40% 21 38% 17 34% 24 34% 27 36% Missing 1 6% 2 8% 9 16% 5 10% 10 14% 7 9% Cycling Behavior Bicycle trip purpose in past year Work or school-related commute 6 38% 8 32% 6 11% 9 18% 12 17% 17 23% No 10 63% 17 68% 49 89% 41 82% 59 83% 58 77% Nonwork or school trip 16 100% 24 96% 54 98% 49 98% 70 99% 73 97% No 0 0% 1 4% 1 2% 1 2% 1 1% 2 3% How safe for cyclists Very unsafe 1 6% 1 4% 3 5% 3 6% 4 6% 4 5% Somewhat unsafe 4 25% 4 16% 13 24% 9 18% 17 24% 13 17% Somewhat safe 7 44% 15 60% 25 45% 26 52% 32 45% 41 55% Very safe 4 25% 5 20% 13 24% 12 24% 17 24% 17 23% Don’t know 0 0% 0 0% 1 2% 0 0% 1 1% 0 0% Note: Values are in number of cycle trips in past year. 98709mvpTxt 1_40 9/20/05 4:58 PM Page 35

and facilities rated as “very important” to commuting by bicycle. Women and men cyclists were relatively sim- ilar in the proportion who value specific types of bicycle facilities such as on-road bicycle lanes, separate bicycle paths, and a connected system of bicycle routes as well as those who value amenities such as secure storage facilities at work or school. They were also relatively similar with respect to the lower proportions of those who value showers at work or bicycle racks on buses. Some gender differences emerged. While none failed to reach levels of statistical significance, the differences are described briefly. Most notably, women are more likely than are their male counterparts to rate paved shoulders and lighting on bicycle paths as “very impor- tant” to commuting by bicycle (84% versus 71% and 68% versus 45%, respectively). Conversely, men are more likely to rate access to information about com- muting and access to information about bicycle routes as “very important” to commuting by bicycle as com- pared with women (48% versus 36% and 65% versus 56%, respectively). Perceptions of safety varied more dramatically between genders. More men cyclists than women cyclists rated Minnesota as safe for cycling (77% ver- sus 70%). Of those who did not rate Minnesota as “very safe” for cycling (n = 111), four themes were identified: lack of bicycle paths, unsafe driver behav- iors, unsafe cyclist behaviors, and unsafe road condi- tions. Among these reasons, there were marked differences by gender. Women were more likely than men to report lack of paths (55% versus 41%) and poor road conditions (13% versus 3%). In contrast, men were more likely than women to report unsafe behaviors of drivers (53% versus 36%) and unsafe behaviors of cyclists (22% versus 15%). Adaptive Stated-Preference Data The second stated-preference data set was a computer- based adaptive stated-preference (ASP) survey adminis- tered by Tilahun et al. to collect information on people’s valuation of different cycling facilities (18). The ASP survey was primarily used to quantify how much addi- tional time, in minutes, respondents are willing to travel to use an alternate higher-quality bicycle facility and if this valuation varies by gender. It is hypothesized that the additional time people are willing to travel in an alternate facility is a function of the attributes of the base facility they can use, attributes of the alternate facility, and personal attributes such as gender, age, and income.7 Given the attributes of the shortest path (base facility), one can measure how much certain improve- ments are valued (in terms of travel time) by users of that facility. The measures are relative, and the presence of certain attributes of the base facility will affect how much one values a given improvement. Each respondent was presented with nine scenarios comparing two facilities for four sets of travel times (see Figure 1 for infrastructure characteristics). The travel times on the higher-quality facility adapt to the subject’s previous choice; if a facility is rejected at a particular travel time, the next presentation has a lower travel time.8 The algorithm always presents a new travel time that is between the now-rejected and previously accepted or the now-accepted and previously rejected travel times. Presenting choices in this manner allows convergence on the critical travel time difference at which an individual is still willing to choose the higher- quality facility. The ASP sample was composed of civil service employees from the University of Minnesota, aged 18 years or older, who reported using a bicycle in the past year (n = 127, 85 women and 42 men). The results show a preferential hierarchy of facilities (people are willing to trade time for higher-quality facil- ities) and differences between women and men. Both women and men are willing to travel longer for an off- road facility (Facility A), followed by a facility with a bicycle lane and no street parking (Facility B), a bicycle lane with side-street parking facility (Facility C), and an in-traffic facility with no parking (Facility D) (see Figure 2). Assuming a typical 20-min commute, this model pre- dicts that individuals are willing to travel about 7.74 min [95% confidence interval (CI) = 5.85, 9.63] for an off-road facility in comparison with a facility that has no side parking and no bicycle lane (see Table 4 for parameter estimates of the full model).9 A key point from this analysis is that, on average, women are willing to travel more additional minutes than men for a preferred facility. Assuming a 20-min 3 6 RESEARCH ON WOMEN’S ISSUES IN TRANSPORTATION 0% 20% 40% 60% 80% 100% Female Male Bike lanes Paved shoulders Bike paths Share-the-road Lighting Connected routes Storage Showers Bike racks Commute info Route info FIGURE 1 Importance of cycling facility infrastructure char- acteristics to current and potential commuting cyclists by gender (percentage that rated characteristic as very impor- tant to commuting by cycle). 98709mvpTxt 1_40 9/20/05 4:58 PM Page 36

3 7GENDER DIFFERENCES IN BICYCLING BEHAVIOR 1 5 10 15 20 25 Alternate Facility; Solid = Female, Dashed = Male Additional Minutes (T) 2 2 3 3 4 4 4 4 A B C D Base 4 1 2 3 E B C D 1 2 2 3 3 4 4 4 4 (A) (D) (C) (B) (E) FIGURE 2 Types of facilities and average additional time willing to travel for alternate facilities by gender: A, off-road facility; B, bike lane, no parking; C, bike lane with parking; D, no bike lane, no parking; E, no bike lane with parking. TABLE 4 Parameter Estimates of Mixed-Effects Regression Model Linear mixed-effects model fit by maximum likelihood AIC BIC logLik 8119.567 8190.147 –4045.783 Random effects Formula: ~1 subject (Intercept) Residual StdDev: 8.385928 7.230089 Fixed effects: Ti ~ W + P + B + O + DP + DB + A + S + H + I + C Description Value Std. Error t-stat p-value (Intercept) 10.709 4.013 2.669 0.0077 ** W Season Winter = 1 Summer = 0 –5.087 1.561 –3.260 0.0014** P Base parking? Yes = 1 No = 0 4.441 0.526 8.437 0.0000 *** B Base bike lane? Yes = 1 No = 0 –6.663 0.526 –12.658 0.0000 *** O Alternate off road Yes = 1 No = 0 7.742 0.967 8.006 0.0000 *** DP Alternate has no parking, Yes = 1 base has parking No = 0 2.252 0.832 2.706 0.0069 ** DB Alternate has bike lane Yes = 1 Base does not No = 0 3.328 0.832 3.890 0.0001 *** A Age 0.095 0.076 1.255 0.2120 S Sex Male = 1 Female = 0 –5.427 1.673 –3.243 0.0015 ** H Household size –1.281 0.667 –1.919 0.0574 † I Household income (=annual/1000) 0.051 0.022 2.266 0.0252 * Significance ***0.001 **0.01 *0.05 †0.1 98709mvpTxt 1_40 9/20/05 4:58 PM Page 37

commute, men are willing to divert 5.43 fewer minutes (95% CI = 2.13, 8.17) than women for any facility com- pared in the survey. For example, the uppermost solid line in Figure 2 connects the average additional time women would travel when the Base Facility E was com- pared with alternate Facilities A, B, C, and D, respec- tively. The corresponding dashed line shows the same comparison for men. In all cases the solid line is above its dashed counterpart, indicating that the average addi- tional travel time that women are willing to expend for a better facility is higher than that for men. The data also suggest that within gender groups, Base Facility E leads to a greater willingness to travel on any other alternate route than when the base is D, followed by when the base is C and B, which suggests a hierarchy in preference for these facilities.10, 11 INTERPRETATION AND SUMMARY An extensive body of literature identifies gender as an important predictor of travel patterns; little of this research, however, examines how cycling patterns and preferences differ between men and women. The research presented here serves to reinforce many expectations of differences between men and women, challenge others, and provide an empirical foundation on which to base future work. It reinforces some expectations by docu- menting that in general, overall rates of cycling for women are less than those for men both in absolute terms and after the number of trips is controlled for. It is shown that rates of cycling across gender differ by type of trip; in particular, women are more likely than men to cycle for shopping and errands or visiting friends. With reported behavior from three different surveys, it is shown that the prevalence of commuting by bicycle is less for women. Furthermore, the bulk of the stated-preference work pre- sented suggests that when only risk is considered, women perceive risks differently from men (12); in particular, women demonstrate a stronger preference for safer forms of cycling infrastructure. In some respects, however, this work challenges or clouds other assumptions or expectations, namely, that women have shorter distances between home and work. Some work suggests that this might not be the case (20). Some of the descriptive statistics presented show that the difference in commute distance between men and women is not statistically significant. Furthermore, some of the prevailing literature suggests that women make fewer recreational trips (21, 22). The descriptive analysis hints that women may have lower rates of com- muting; they may pass men in terms of recreational use, particularly in suburban environments. This finding tends to support those of other studies showing that women have higher rates of leisure travel (8). The findings reported here, however, need to be con- sidered in light of several study limitations. First, the samples used were relatively small, especially after strat- ification on gender and household location. All summary characteristics must therefore be viewed with caution since small changes in any given value could change the described patterns substantially. Second, data sets except for the NHTS are subsets from larger data sets, none of which is representative of cyclists in the Twin Cities met- ropolitan area.12 Third, it is difficult to compare cyclists across subsamples of the data analyzed. However, this study provides empirical documenta- tion of an often-glossed-over but important phenome- non—women’s cycling. The paper therefore contributes to the planning, transportation, and public health (physi- cal activity) literature by providing a quantitative valua- tion of how women demonstrate different patterns of cycling, may prefer different bicycle facilities, and have different safety considerations. These findings—in con- cert with more refined investigation—will inevitably aid policy discussions. For example, they draw attention to the fact that different infrastructure decisions likely have varying impacts on difference audiences in terms of mak- ing cycling environments safer (23) or more attractive to different users. From a practical standpoint, such infor- mation may be useful for marketing or for directing seg- mented and targeted policies. If women have different use patterns, make different route choice decisions, or prefer different cycling facilities, these factors are likely to have important implications for provision of different facilities and the use that planners and other policy officials can expect from them. For example, women may prize lighted paths and paved shoulders more than do men. Future research could be oriented toward understand- ing how these patterns play out by age and location and moreover what the underlying behavioral reasons for these patterns are. This study could be done through a combina- tion of more extensive and focused analysis of available data sets (e.g., the NHTS) and direct questionnaires to both current and potential women cyclists. It would be interest- ing to learn whether such relationships hold true across metropolitan settings. This work could be used in combi- nation with conceptual frameworks (14) to further refine future research. This study therefore offers a first step in describing gender differences in cycling behaviors and pref- erences. Such an understanding can be incorporated into the planning process and contribute to policy dialogues regarding optimal investment decisions on bicycle facilities for different market segments. ACKNOWLEDGMENTS The authors are grateful to Gary Barnes, who assisted with the analysis of the NHTS data for this project, and 3 8 RESEARCH ON WOMEN’S ISSUES IN TRANSPORTATION 98709mvpTxt 1_40 9/20/05 4:58 PM Page 38

to Robin McWaters, who assisted with diligent editing. Financial support was provided in part by the National Cooperative Highway Research Program and by the Minnesota Department of Transportation. NOTES 1. The Twin Cities of Minneapolis and St. Paul provide a suitable setting for such research. The metropolitan area enjoys an unparalleled system for off-street bicycling, and the city of Minneapolis ranks among the top cities in the percentage of workers commuting by bicycle. 2. NHTS data are collected from 60,282 persons in 26,038 households that make up the national sample. The survey asked respondents (or their adult proxies) to report all trips taken during a specified 24-h travel day. The response rate was approximately 41%, and weighted results were used here to reflect the travel behavior of the whole sample population. 3. The TBI data were originally collected through 24-h travel diaries and household telephone interviews from randomly selected households across the seven-county metropolitan area and 13 surrounding counties. All subjects from the TBI database that were residents of the seven-county metropolitan area and were age 18 years or older were selected. The variables of interest in this application represent bicycle use in one 24-h period, including cycle trips (number of cycle trips in 24 h); trip types—work trip (any cycle trip to work, yes or no), commute trip (any cycle trip to work or school, yes or no), recreation or fitness trip (any cycle trip for recre- ation or fitness, yes or no)—and the distance cycled— total kilometers cycled in 24 h. 4. Authors’ calculation using weighted sample of NHTS respondents (not including add-on areas) aged 17 or more: 87,385,641 total trips and 3,904,365 total bicycle trips. 5. In the TBI sample cyclists are predominantly men (34% women versus 66% men) and nearly two-thirds (62%) are central city residents (Minneapolis or St. Paul) whereas 38% live in the suburbs. Specifically, Twin Cities women cyclists are less likely to have a college degree but more likely to be employed than are male cyclists. In the suburbs, women cyclists are more likely to have a college degree but less likely to be employed than their suburban male counterparts. 6. Two other criteria were applied to ensure applicability of these data. First, residents of the seven-county metro- politan area were selected to best comport with analysis from the other data sources, which were mostly from urbanized areas. Second, the data included individuals from all walks of life, many of whom never cycle. Uncovering why such individuals never cycle is impor- tant. However, the nature of the questions prompted the retention of individuals who indicated that they had used a bicycle in the past year (n = 146). Their responses were more in tune with the nature of the questions. 7. This analysis is performed by using a mixed-effects regression model, which provides a relative measure of attractiveness of the attributes of cycling facilities. 8. However, this travel time will still be higher than a travel time the subject has found acceptable in a previ- ous trade-off. 9. If the base facility had parking, individuals would be willing to add additional minutes to avoid that base facility. However, if the base facility has a bicycle lane, individuals are only willing to travel 1.08 additional minutes for the alternative off-road facility. Similarly, individuals are willing to bicycle an additional 3.24 min (95% CI = 1.61, 4.86) if an alternate route provides a bicycle lane as compared with a facility that has no parking and no bicycle lane. If the base facility has park- ing, the additional minutes they are willing to travel for the alternate bicycle lane facility increases by 4.44 min (95% CI = 3.41, 5.46). In addition, if the alternative also provides a parking improvement, they are willing to add another 2.25 min (95% CI = 5.85, 9.63). 10. For the ASP survey, there are multiple responses from each person, which requires an additional step to account for the within-person correlation. Thus, a lin- ear mixed-effects model was used, which allows for the specification of an additional variance component in the form of a random effect. The mixed-effects analysis was conducted with the NLME library in R statistical software (19). 11. The additional time that an individual is willing to travel also differs across demographic and economic variables. Household income and household size were also statistically significant. As income levels increase, individuals are willing to travel longer on the alternate facility. An increase in household size is associated with an unwillingness to trade time for alternate facilities. 12. Because the TBI and the Omnibus data sets were ran- dom samples obtained by means of complex sampling strategies designed to produce representative samples of the Twin Cities metropolitan area population, the sub- set of cyclists used in this study cannot be assumed to be representative. REFERENCES 1. Saelens, B. E., J. F. Sallis, and L. D. Frank. Environmental Correlates of Walking and Cycling: Findings from the Transportation, Urban Design and Planning Literatures. Annals of Behavioral Medicine, Vol. 25, No. 2, 2003, pp. 80–91. 2. Blumen, O. Gender Differences in the Journey to Work. Urban Geography, Vol. 15, No. 3, 1994, pp. 223–245. 3 9GENDER DIFFERENCES IN BICYCLING BEHAVIOR 98709mvpTxt 1_40 9/20/05 4:58 PM Page 39

3. McLafferty, S., and V. Preston. Gender, Race, and the Determinants of Commuting: New York in 1990. Urban Geography, Vol. 18, 1997, pp. 192–212. 4. Turner, T., and D. Neimeier. Travel to Work and Household Responsibility: New Evidence. Transportation, Vol. 24, 1997, pp. 397–419. 5. Sarmiento, S. Household, Gender and Travel. In Women’s Travel Issues: Proceedings from the Second National Conference, October 1996, FHWA, U.S. Department of Transportation, 2003. 6. Antonakos, C. L.. Environmental and Travel Preferences of Cyclists. In Transportation Research Record 1438, TRB, National Research Council, Washington, D.C., 1994, pp. 25–33. 7. Johnston-Anumonwo, I. The Influence of Household Type on Gender Differences in Work Trip Distance. Professional Geographer, Vol. 44, No. 2, 1992, pp. 161–169. 8. Collins, D., and C. Tisdell. Gender and Differences in Travel Life Cycles. Journal of Travel Research, Vol. 41, 2002, pp. 133–143. 9. Gordon, P., and H. W. Richardson. Bicycling in the United States: A Fringe Mode? Transportation Quarterly, Vol. 52, No. 1, 1998, pp. 9–11. 10. Handy, S.. In Women’s Travel Issues: Proceedings from the Second National Conference, October 1996, FHWA, U.S. Department of Transportation, 2003. 11. Noland, R. B. Perceived Risk and Modal Choice: Risk Compensation in Transportation Systems. Accident Analysis and Prevention, Vol. 27, No. 4, 1995, pp. 503–521. 12. Blais, A.-R., and E. U. Weber. Domain-Specificity and Gender Differences in Decision Making. Risk Decision and Policy, Vol. 6, 2001, pp. 47–69. 13. Rodriguez, D., and J. Joo. The Relationship Between Non-Motorized Mode Choice and the Local Physical Environment. Transportation Research, Vol. 9D, 2004, pp. 151–173. 14. Krizek, K. J., and P. J. Johnson. The Effect of Neighborhood Trails and Retail on Cycling and Walking in an Urban Environment. Active Communities/Trans- portation Research Group, University of Minnesota, 2004, 18 pp. 15. Cervero, R., and M. Duncan. Walking, Bicycling, and Urban Landscapes: Evidence from the San Francisco Bay Area. American Journal of Public Health, Vol. 93, No. 9, 2003, pp. 1478–1483. 16. Shafizadeh, K., and D. Niemeier. Bicycle Journey-to- Work: Travel Behavior Characteristics and Spatial Attributes. In Transportation Research Record 1578, TRB, National Research Council, Washington, D.C., 1997, pp. 84–90. 17. Brown, N. Why Aren’t Women on Wheels? San Francisco Bicycle Coalition, 2004. www.sfbike.org/ ?women_mb. Accessed June 8, 2005. 18. Tilahun, N., D. Levinson, and K. J. Krizek. Trails, Lanes, or Traffic: The Value of Different Bicycle Facilities Using an Adaptive Stated Preference Survey. Active/Communities Transportation Research Group, University of Minnesota, 2004, 21 pp. 19. R: A Language and Environment for Statistical Computing. R Development Core Team, Vienna, Austria, 2003. 20. Kwan, M.-P. Gender, the Home-Work Link, and Space- Time Patterns of Nonemployment Activities. Economic Geography, Vol. 75, No. 4, 1999, pp. 370–394. 21. Hanson, S., and I. Johnston. Gender Differences in Work-Trip Length: Explanations and Implications. Urban Geography, Vol. 6, No. 3, 1985, pp. 193–219. 22. Lu, X., and E. Pas. Socio-demographics, Activity Participation and Travel Behavior. Transportation Research, Vol. 33A, 1999, pp. 1–18. 23. Pucher, J., and L. Dijkstra. Making Walking and Cycling Safer: Lessons from Europe. Transportation Quarterly, Vol. 54, No. 3 (Summer), 2000, pp. 25–50. 4 0 RESEARCH ON WOMEN’S ISSUES IN TRANSPORTATION 98709mvpTxt 1_40 9/20/05 4:58 PM Page 40

4 1 Automobile ownership plays an important role in deter- mination of travel behavior. In contrast, gender differ- ence is shown to be a significant factor in automobile choice behavior. The primary objective of this study was to identify influential factors that affect gender differ- ences in automobile choice behavior. This study pre- sents the process of developing mixed logit discrete choice models that control for gender to predict auto- mobile type choice behavior. A variety of explanatory variables were used to provide a good model fit. Over- all, the models demonstrate that gender-specific auto- mobile type choice behavior is influenced by a wide variety of explanatory variables, including automobile attributes and household and individual characteristics. Automobile ownership plays a critical role indetermination of travel behavior. The decisionto buy a vehicle is one of the most important decisions made by a household. Automobile owner- ship has a dramatic effect on mobility and access to new opportunities such as employment and social services. During the past few years, increasing attention has been paid to the use of microsimulation modeling approaches to activity-based travel forecasting. Some of the earliest applications of microsimulation in the trans- portation field involved dynamic modeling of automo- bile ownership (Miller 1996), and the majority of new travel demand models include explicit modules that sim- ulate household automobile choice behavior, for exam- ple, those of Salvini and Miller (2003) and PB Consult (2002). Furthermore, it is clear that understanding the factors driving vehicle ownership and choice behavior is important in addressing a range of environmental issues. Analyses of household travel survey data have shown differences in travel demand across households that vary with family type and gender of the household members. It has been shown that women behave differently from men when it comes to their travel patterns. Women travel for shorter distances between work and home and make more trips because of their special role in the household (Wachs 1997). Wachs also reports that lower-income women are more likely to use public transit. Mokhtarian (1997) found that women are more likely than men to change their travel behavior as a result of congestion. Furthermore, it seems that women are more willing than men to reduce car use (Matthies et al. 2002). One can postulate the same tendency for gender differences in automobile ownership and utilization behavior. Gender difference is shown to be a significant factor in automo- bile ownership behavior, and this difference persists across all ages (Prskawetz et al. 2002). Automobile own- ership in female-headed households tends to be less than that in male-headed households, and this difference can increase dramatically for households with older mem- bers. One reason to explain this tendency can be the lower female labor-force participation rate and the higher rate of single-person, female-headed households. However, the overall percentage of female drivers is rising, whereas the overall percentage of male drivers is decreasing. According to the U.S. Department of Trans- Gender Differences in Automobile Choice Behavior Abolfazl (Kouros) Mohammadian, University of Illinois, Chicago 98709mvpTxt 41_88 9/20/05 5:05 PM Page 41

portation, the percentage of female drivers grew from 44% in 1972 to 49.7% in 2000 compared with the per- centage of male drivers, which fell from 56% in 1972 to 50.3% in 2000 (FHWA 2004). It has been also shown that women have influential buying power. Almost half of the new vehicles are purchased by women, and Ford Motor Company marketing research shows that women influence 80% of car-buying deci- sions and tend to have different preferences compared with their male counterparts. The way individuals and households make decisions with respect to automobile ownership has been the sub- ject of numerous studies across many disciplines. Trans- portation planners are interested to know how many and what type of automobiles are owned by households and how people utilize their vehicles. Several studies have been conducted in this area. One of the early disaggre- gate studies, by Manski and Sherman (1980), developed multinomial logit models for the number of vehicles owned and vehicle type choice. Separate models for automobile type choice were developed for households with one or two vehicles in their fleet. Mannering and Winston (1985) attempted to model number of vehicles, vehicle type, and vehicle usage in an interrelated frame- work. They estimated separate models for single- and two-vehicle households. Hensher and Le Plastrier (1985) used a nested logit structure to model a household’s auto- mobile holdings and composition choice. Brownstone et al. (1996) developed an automobile choice model using stated-preference data. In their study six hypothetical alternatives were presented to respondents with a num- ber of randomly distributed attributes within certain ranges. Yamamoto et al. (1999) developed a competing risk duration model of household vehicle transactions by estimating a hazard model for each type of transaction separately, assuming the independence of unobserved heterogeneity across the hazards. More recently Moham- madian and Miller (2002b, 2003a, 2003b) developed a series of dynamic automobile transaction and type choice models using nested logit, mixed logit, and machine learning methods. Some of the models reviewed here have dummy variables to specify gender in the model specification, but to the best of the author’s knowledge, no other study in the transportation litera- ture has modeled gender-segmented automobile choice behavior. The primary objective of this study was to develop econometric models of automobile type choice and to identify influential factors that affect gender differences in automobile choice behavior. Two different mixed logit models were developed for male and female decision makers while heterogeneity effects on choice behavior were accounted for. These models were then compared on the basis of the values of their fixed and random pa- rameters. This procedure provides an opportunity to investigate the effects of gender-specific factors that drive vehicle choice behavior and will offer scientific evidence to policy makers for further policy direction. In addition, the models developed here can be used to examine a wide range of scenarios and policy analysis interests. DATA The data set used in this study was obtained through the Toronto Area Car Ownership Study (TACOS), which was a retrospective survey (Roorda et al. 2000). The data set contained information on household vehicle transac- tions for up to 9 years from 1990 to 1998 in the Greater Toronto Area. Vehicle characteristics for each automo- bile in the sample were obtained from the Vehicle Speci- fications System of the Canadian Association of Technical Accident Investigators and Reconstructionists (CATAIR) (1999). The Fuel Economy Guide database of the Environmental Protection Agency (EPA) (2002) pro- vided fuel consumption information, and vehicle market values at time of purchase were gathered from Canadian Red Book, Inc. (1990–1998a, b). Thus, the prices that are used in the models are not the individual sale prices of the specific vehicles reported in the sample (which would be subject to potential self-selection biases or reporting errors because of the retrospective nature of the survey) but rather the average market values for each given make, model, or vintage of vehicle in the sample. Many variables are used in the literature to explain the difference in utility between different models. These vari- ables include the turning circle, braking distance, axle ratio, revolutions per minute, horsepower, luggage space, head room, leg room, length, width, fuel efficiency, engine size, and weight. In order to keep the model as simple as possible, only those characteristics were chosen that are judged to be the most important in representing variation in utility. Several vehicle characteristics such as weight, engine displacement, fuel intensity, luggage capacity, and wheelbase were chosen. These vehicle characteristics pre- sent a special difficulty when a model is estimated. For technological reasons, many of these variables are highly correlated. This high multicollinearity between variables might create problems of identification of the influence of car characteristics on vehicle utility. Principal components analysis was used to solve this problem (Mohammadian and Miller 2002a). As a result of the principal compo- nents analysis, two factors were identified that explain 89% of the total variance in the sample: vehicle perfor- mance (dominant variables include vehicle weight, engine displacement, and fuel intensity) and vehicle space (dom- inant variables include size, luggage capacity, and wheel- base). These two composite factors were used as two independent variables in the utility functions of the model. 4 2 RESEARCH ON WOMEN’S ISSUES IN TRANSPORTATION 98709mvpTxt 41_88 9/20/05 5:05 PM Page 42

The final sample used in the modeling effort of this study includes 597 automobiles for which all the required explanatory variables were available. Variables in the model were selected on the basis of prior experi- ence with this type of model (Mohammadian and Miller 2003a, 2003b). The sample means and standard devia- tions for the explanatory variables used in this study are presented in Table 1. MIXED LOGIT MODEL Random utility–based discrete choice models have found their way into many disciplines. The multinomial logit model is the most popular form of discrete choice model in practical applications. It is based on several simplifying assumptions such as independent and iden- tical Gumbel distribution (IID) of random components of the utilities and the absence of heteroscedasticity and autocorrelation in the model. It has been shown that these simplifying assumptions limit the ability of the model to represent the true structure of the choice process. Recent research has contributed to the develop- ment of closed-form models, which relax some of these assumptions to provide a more realistic representation of choice probabilities. Mixed logit (ML) and general- ized extreme value (GEV) models are examples of these alternative structures, a detailed discussion of which may be found elsewhere (Bhat 2002). The ML model was introduced by Ben-Akiva and Bolduc (1996) to bridge the gap between logit and probit models by combining the advantages of both techniques. A growing number of empirical studies implement the ML method, including those by Revelt and Train (1998), Bhat (1997, 2000), Brownstone et al. (2000), and Mohamma- dian and Miller (2003a). Consider the following utility function: (1) where ain is a constant term and captures an intrinsic preference of decision maker n for alternative i, giWn captures the systematic preference heterogeneity as a function of sociodemographic characteristics, and Xint is the vector of attributes describing alternative i for deci- sion maker n in the choice situation t. The vector of coef- ficients bin is assumed to vary in the population, with probability density given by f (b|q), where q is a vector of the true parameters of the taste distribution. If the e’s are IID Type I extreme value, the probability that decision maker n chooses alternative i in a choice situation t is given by (2) where Cnt is the choice set available to decision maker n in choice situation t, and j represents individual choices within choice set Cnt. The probability in Equation 2 is conditional on the distribution of bin. A subset of all of ain alternative-specific constants and the parameter bin vector can be randomly distributed across decision mak- ers. An important element of these random parameter models is the assumption regarding the distribution of each of the random coefficients. A more detailed treat- ment of preference heterogeneity may be found else- where (Bhat 2000). Since actual tastes are not observed, the probability of observing a certain choice is determined as an inte- gral of the appropriate probability formula over all pos- sible values of bn weighted by its density. Therefore, the unconditional probability of choosing alternative i for a randomly selected decision maker n is then the integral of the conditional multinomial choice probability over all possible values of bn: Pnt(i | q) = ∫ b Pnt (i | bn)f(b | q)db (3) In general, the integral cannot be analytically calcu- lated and must be simulated for estimation purposes. Since exact maximum likelihood estimation is not avail- able, simulated maximum likelihood is to be used by P i W X W X nt n in i n in int int j Cnt jn j n jn jnt jnt ( | ) exp( ) exp( ) b a g b e a g b e = + + + + + + Œ S U W Xint in i n in int int= + + +a g b e 4 3G E N D E R D I F F E R E N C E S I N A U T O M O B I L E C H O I C E B E H AV I O R TABLE 1 Sample Mean and Standard Deviations for Explanatory Variables Male Female Variable Mean Std. Dev. Mean Std. Dev. Vehicle performance factor 4.96 0.84 4.93 0.82 Vehicle space factor 0.44 0.30 0.46 0.28 Vehicle purchase price (¥$1000) 15.21 11.80 13.99 6.45 Vehicle ownership and operating cost (¥$1000) 7.74 2.32 7.07 2.06 Vehicle is a used car 0.52 0.50 0.50 0.50 Household income (¥$1000) 59.10 22.39 60.00 22.28 Driver age 43.30 13.94 40.55 13.69 Home owner 0.25 0.43 0.18 0.39 Driver is employed 0.73 0.44 0.67 0.47 Number of observations 350 247 98709mvpTxt 41_88 9/20/05 5:05 PM Page 43

drawing pseudorandom realizations from the underly- ing error process. A detailed discussion of this method may be found elsewhere (Louviere et al. 2000; Bhat 2000). Model Estimation Decision makers (defined by survey respondents) in this study are faced with the decision of what class of auto- mobile to purchase. The choice set contains six alterna- tives: subcompact, compact, midsize, full size, special purpose vehicle [sport utility vehicle (SUV) and pickup truck], and van (van and minivan). The data set extracted to develop this model contains 597 automo- bile class choice observations. It is assumed that all choices are available to all decision makers. Variables representing automobile attributes and individual and household characteristics are used in util- ity functions. In ML models, heterogeneity can be accounted for by letting certain parameters of the utility function differ across decision makers. It has been shown that this formulation can significantly improve both the explanatory power of models and the precision of parameter estimates. Ben-Akiva and Bolduc (1996) and Bhat (2000) provide detailed discussions of the ML models and their estimation method. In this study 1,000 repetitions are used to estimate the unconditional probability by simulation. This method will improve the accuracy of the simulation of individual log-likelihood functions and will reduce the simulation variance of the maximum simulated log- likelihood estimator. Two important aspects of model- ing strategy that need to be considered before estimation of an ML model are the identification of parameters with and without heterogeneity and the assumption regarding the distribution of each of the random coeffi- cients. These two aspects must be selected on the basis of prior information, theoretical considerations, or some other criteria. Random parameters in this study are estimated as normally distributed parameters in order to allow parameters to get both negative and pos- itive values. Both observed attributes of the decision makers and alternatives (explanatory variables) and their unobserved attributes (alternative-specific con- stants) were introduced as random parameters. Tables 2 and 3 show the results of the model for male and female decision makers in detail. These tables pre- sent statistically significant parameters and a good model fit—given the capability of the data set and model—espe- cially since the focus of the work is to define gender dif- ferences in automobile choice behavior and to identify factors that define that difference, not how each individ- ual selects the best alternative in a choice situation. In reviewing the results, one can observe consistency in the signs of coefficients across the models. Parameters of both models are statistically significant at the 95% con- fidence level or better. The signs of all utility parameters seem to be correct and unambiguous. Furthermore, esti- mated standard deviations of the random parameters of variables representing vehicle price, income, and alterna- tive-specific constants are statistically significant in the model. The significant t-statistics for these standard devi- ations indicate that they are statistically different from zero, confirming that parameters indeed vary in the pop- ulation. Results of the model estimation strongly imply that heterogeneity is a significant factor in the model developed here. Discussion of Modeling Results The vehicle performance factor has a significant positive coefficient in all utility functions of male decision mak- ers. The vehicle performance factor is a composite fac- tor representing physical and operational attributes of the vehicle, including weight, engine displacement, and fuel intensity. This finding conforms with the notion that male automobile owners prefer more power and performance in their vehicle. In the second model, the vehicle performance factor has a significant negative coefficient in all utilities, suggesting that female auto- mobile consumers probably tend to prefer practicality and safety of the vehicle over its performance. This find- ing is probably due to disparities in women’s injury and crash rates compared with those of men as well as their key responsibilities in the household and their concerns for children’s welfare. The vehicle space factor, another composite factor rep- resenting luggage capacity, wheelbase, size, and cargo fea- tures of the vehicle, has the expected significant negative coefficient for small-sized vehicles (subcompacts, com- pacts, and mid-sized cars) and a significant positive coeffi- cient in the utility function of large-sized vehicles (vans and minivans) in both models. Furthermore, the value of the coefficient of the space factor for sedan-class vehicles in the model developed for female decision makers is greater than the value of the coefficient in the male model, which suggests that female automobile customers are more sensitive to space-related attributes of sedan-class vehicles. They seem to prefer better safety features and more storage and other room in their vehicles. This find- ing confirms the notion that women prefer practicality and safety, as discussed earlier. This difference can be attributed to differences in activity needs and travel behav- ior between the two genders, which require further analy- sis and can lead to the need to design different vehicles for men and women. Modeling results suggest that both male and female decision makers are responsive to vehicle price. It is 4 4 RESEARCH ON WOMEN’S ISSUES IN TRANSPORTATION 98709mvpTxt 41_88 9/20/05 5:05 PM Page 44

expected that the purchase price would present negative effects on the utility of the alternatives. This assumption is supported by negative significant coefficients for sub- compact and compact vehicles in both models. How- ever, modeling results suggest that both female and male automobile customers tend to consider the purchase price as a positive factor when they choose full-size or special purpose vehicles. This finding is somewhat as expected given the luxury and exclusive features often found in full-size vehicles and SUVs that make them 4 5G E N D E R D I F F E R E N C E S I N A U T O M O B I L E C H O I C E B E H AV I O R TABLE 2 ML Model Estimation: Male Decision Makers Subcompact Compact Mid-Size Full-Size Special Purpose Van Variable Coeff. t-Stat Coeff. t-Stat Coeff. t-Stat Coeff. t-Stat Coeff. t-Stat Coeff. t-Stat Vehicle characteristics Performance factor 0.29 1.62 0.29 1.62 0.29 1.62 0.29 1.62 0.29 1.62 0.29 1.62 Space factor –4.97 –8.06 –4.97 –8.06 –4.97 –8.06 4.74 4.25 Log (purchase price) –0.85 –3.46 –0.75 –3.06 0.81 2.76 0.83 3.14 Std. dev. 0.33 0.91 0.37 1.42 0.20 1.59 0.17 1.14 Used car –1.73 –4.45 –1.73 –4.45 –1.19 –2.70 Household attributes Log (income) 0.76 1.77 0.84 1.69 Std. dev. 0.29 1.15 0.37 1.79 Home owner –0.68 –1.83 –0.82 –2.01 Individual attributes Log (driver age) 1.32 2.40 Driver employed 1.09 2.59 –0.92 –2.49 Constants 4.15 1.89 3.74 1.71 –5.52 –1.83 –1.99 –0.86 0.73 0.34 Std. dev. 2.53 1.08 0.95 1.24 1.12 1.62 0.62 1.67 0.22 0.99 Number of observations 350 Chi-squared 360.29 Log-likelihood at convergence –446.97 Log-likelihood at constants –623.62 Log-likelihood—no coefficients –627.12 Log-likelihood ratio 0.27 TABLE 3 ML Model Estimation: Female Decision Makers Subcompact Compact Mid-Size Full-Size Special Purpose Van Variable Coeff. t-Stat Coeff. t-Stat Coeff. t-Stat Coeff. t-Stat Coeff. t-Stat Coeff. t-Stat Vehicle characteristics Performance factor –1.41 –4.81 –1.41 –4.81 –1.41 –4.81 –1.41 –4.81 –1.41 –4.81 –1.41 –4.81 Space factor –6.95 –7.45 –6.95 –7.45 –6.95 –7.45 4.24 1.98 Log (purchase price) –2.44 –5.29 –0.13 –2.01 1.94 2.24 1.52 2.15 Std. dev. 1.01 1.33 0.09 1.26 1.79 2.32 1.16 1.60 Used car –0.97 –2.03 1.49 3.46 1.32 2.37 Household attributes Log (income) –0.49 –1.54 –1.07 –1.53 Std. dev. 0.47 1.51 0.66 1.43 Home owner –0.89 –1.72 –1.29 –1.59 Individual attributes Log (driver age) 1.54 2.80 Driver employed 0.80 1.44 Constants –1.80 –0.58 –4.19 –1.39 –14.37 –3.96 –12.94 –3.90 –9.22 –2.67 Std. dev. 0.90 1.44 1.36 1.18 2.83 2.12 2.14 1.48 1.62 1.08 Number of observations 247 Chi-squared 349.96 Log-likelihood at convergence –267.58 Log-likelihood at constants –408.59 Log-likelihood—no coefficients –442.56 Log-likelihood ratio 0.38 98709mvpTxt 41_88 9/20/05 5:05 PM Page 45

more fashionable. The coefficient of vehicle purchase price has a larger magnitude in the women’s model, sug- gesting that female decision makers are more sensitive to price than their male counterparts are. This finding confirms that disparities in the financial capacity of men and women could lead to different choices. This confir- mation was somehow expected since women have lower labor-force participation, which can affect their access to credit or may influence their buying power. In order to account for heterogeneity, parameters of vehicle price are entered into the utility functions as random parame- ters with normal distribution. Modeling results confirm that standard deviations of these parameters are statisti- cally different from zero, suggesting that the parameters vary in the population. Used cars purchased by women drivers are less likely to be subcompacts and have a higher probability of being a mid-size or special purpose vehicle. Results presented in Table 2 also suggest that male car buyers are less likely to purchase a used subcompact, compact, or van. With regard to individual-related characteristics, the variable for driver’s age has significant positive coeffi- cients in the utility functions of the mid-size alternative in both models. This finding suggests that older drivers are more likely to purchase mid-size automobiles. The magnitude and t-statistics of this parameter are more significant in the model developed for women drivers. Employment status was found to be important in both models. The utility of purchasing an SUV is higher for an employed woman. In contrast, male decision makers who are employed have a higher utility to purchase mid-size vehicles and are less likely to buy a full-size automobile. Two additional factors related to household charac- teristics were also explored. Male car buyers from households with higher incomes are more likely to pur- chase vans and full-size vehicles. Female members of high-income households are less likely to purchase vehi- cles that are either vans or compacts. The household income variable was also introduced to both models as a random parameter variable with normal distribution. The standard deviations of income parameters were found to be significantly different from zero, suggesting that these parameters are in fact random. Both male and female decision makers who are members of households that own their homes are less likely to purchase mid-size and special-purpose vehicles. Model results suggest that female car buyers are more sensitive to this variable than their male counterparts are. Other household-level vari- ables that may play an important role in automobile choice behavior include the presence of children and the household structure and lifestyle. Assessing the impacts of these variables on the structure of the model and explaining choice behavior remain as tasks for further research in this area. Alternative-specific constants that account for unob- served attributes of the decision maker and alternatives were also introduced into the models as random parame- ters with normal distribution. This method will allow accounting for heterogeneity of unobserved attributes. The results of the simulation with 1,000 draws confirm the importance of parameter heterogeneity in all alternative- specific constant terms introduced to utility functions of both models. CONCLUSIONS This study presents the process of developing ML mod- els to simulate vehicle class choice by male and female decision makers based on attributes of the alternatives and characteristics of the decision makers and their households. Models developed for male and female decision makers are compared on the basis of the sign and the value of their parameters as well as the value of mean and standard deviation of their random parame- ters. Interpretation of the effects of each explanatory variable in the model led to several interesting insights. These findings were consistent in models developed for both male and female decision makers, so it is clear that personality characteristics deserve further attention. It is shown that gender difference is a significant factor in automobile type choice behavior. Female automobile consumers are found to prefer practicality over perfor- mance. They usually tend to prefer better safety features and more storage and other room in their vehicles, prob- ably because of their responsibilities in the household and concerns for the children. At the same time, male decision makers tend to prefer more power and performance in their vehicles. It was also shown that female car buyers are more responsive to the price of automobiles than their male counterparts are, probably because of issues related to their buying power and access to credit. Several other factors explaining gender differences were also explored and discussed. A few other factors that were worth fur- ther examination were identified, including driver behav- ior, activity needs, presence of children, and household structure and lifestyle. Overall, what can be confidently concluded from this first attempt to explicitly account for gender-specific attributes is that such factors play an important role in the decision-making process and that these interactive effects are deserving of further attention in future analysis. The results of the study presented here can facilitate addressing a range of social and planning issues. It is hoped that these results provide scientific evidence to policy makers for further direction. These models—given the capability and limitations of the data set and the models—can be used to examine various scenarios of technology design and pol- icy analysis. Variations in vehicle technology, pricing, financ- 4 6 RESEARCH ON WOMEN’S ISSUES IN TRANSPORTATION 98709mvpTxt 41_88 9/20/05 5:05 PM Page 46

ing, and demographic assumptions are among the factors that can be explored. This study will lead to a better under- standing of what mechanisms and programs should be designed to facilitate meeting long-term goals of equitable and sustainable transportation systems more effectively. REFERENCES Ben-Akiva, M., and D. Bolduc. 1996. Multinomial Probit with a Logit Kernel and a General Parametric Specification of the Covariance Structure. Presented at Third International Choice Symposium, Columbia University. Bhat, C. 1997. An Endogenous Segmentation Mode Choice Model with an Application to Intercity Travel. Transportation Science, Vol. 31, No. 1, pp. 34–48. Bhat, C. 2000. Incorporating Observed and Unobserved Heterogeneity in Urban Work Travel Mode Choice Modeling. Transportation Science, Vol. 34, No. 2, pp. 228–238. Bhat, C. 2002. Recent Methodological Advances Relevant to Activity and Travel Behavior. In Perpetual Motion: Travel Behavior Research Opportunities and Application Challenges (H. S. Mahmassani, ed.), Elsevier Science, Oxford, England, pp. 381–414. Brownstone, D., D. S. Bunch, T. F. Golob, and W. Ren. 1996. A Transactions Choice Model for Forecasting Demand for Alternative-Fuel Vehicles. Working Paper UCI-ITS- WP-96-4. Institute of Transportation Studies, University of California, Irvine. Brownstone, D., D. Bunch, and K. Train. 2000. Joint Mixed Logit Models of Stated and Revealed Preferences for Alternative-Fuelled Vehicles. Transportation Research, Vol. 34B, pp. 315–338. Canadian Association of Technical Accident Investigators and Reconstructionists (CATAIR). 1999. Canadian Vehicle Specifications System: Version 1999.2. www.catair.net. Accessed Aug. 1, 2002. Canadian Red Book, Inc. 1990–1998a. Canadian Red Book: Official Used Car Valuations. Toronto, Canada. Canadian Red Book, Inc. 1990–1998b. Canadian Older Car/Truck Red Book. Toronto, Canada. Environmental Protection Agency. 2002. Fuel Economy Guide. Database files. Office of Mobile Sources, National Vehicle and Fuel Emissions Laboratory, Ann Arbor, Mich. www.fueleconomy.gov/feg/download. shtml. Accessed Aug. 1, 2002. FHWA. 2004. Highway Statistics. U.S. Department of Transportation. www.fhwa.dot.gov/policy/ohpi/hss/ index.htm. Accessed Aug. 7, 2004. Hensher, D. A., and V. Le Plastrier. 1985. Towards a Dynamic Discrete-Choice Model of Household Automobile Fleet Size and Composition. Transportation Research, Vol. 19B, No. 6, pp. 481–495. Louviere, J. J., D. A. Hensher, and J. D. Swait. 2000. Stated Choice Methods, Analysis and Applications. Cam- bridge University Press, United Kingdom. Mannering, F., and C. Winston. 1985. A Dynamic Empirical Analysis of Household Vehicle Ownership and Utilization. Rand Journal of Economics, Vol. 16, No. 2, pp. 215–236. Manski, C. F., and L. Sherman. 1980. An Empirical Analysis of Household Choice Among Motor Vehicles. Transportation Research, Vol. 14A, No. 5, 6, pp. 349–366. Matthies, E., S. Kuhn, and C. A. Klockner. 2002. Travel Mode Choice of Women: The Result of Limitation, Ecological Norm, or Weak Habit? Environment and Behavior, Vol. 34, No. 2, pp. 163–177. Miller, E. J. 1996. Microsimulation and Activity-Based Forecasting. In Activity-Based Travel Forecasting Conference, June 2–5, 1996: Summary, Recommendations, and Compendium of Papers, Travel Model Improvement Program, U. S. Department of Transportation and Environmental Protection Agency, pp. 151–172. Mohammadian, A., and E. J. Miller. 2002a. Estimating the Expected Price of Vehicles in a Transportation Microsimulation Modeling System. Journal of Transportation Engineering, ASCE, Vol. 128, No. 6, pp. 537–541. Mohammadian, A., and E. J. Miller. 2002b. Nested Logit Models and Artificial Neural Networks for Predicting Household Automobile Choices: Comparison of Performance. In Transportation Research Record: Journal of the Transportation Research Board, No. 1807, Transportation Research Board of the National Academies, Washington, D.C., pp. 92–100. Mohammadian, A., and E. J. Miller. 2003a. Dynamic Modeling of Household Automobile Transactions. In Transportation Research Record: Journal of the Transportation Research Board, No. 1831, Trans- portation Research Board of the National Academies, Washington, D.C., pp. 98–105. Mohammadian, A., and E. J. Miller. 2003b. Empirical Investigation of Household Vehicle Type Choice Decisions. In Transportation Research Record: Journal of the Transportation Research Board, No. 1854, Transportation Research Board of the National Academies, Washington, D.C., pp. 99–106. Mokhtarian, P. 1997. More Women Than Men Change Behavior to Avoid Congestion. ITS Review, Vol. 20, No. 2, p. 2. PB Consult. 2002. Household Car Ownership Model. Technical Memorandum. Mid-Ohio Regional Planning Commission Model Improvement Project. Prskawetz, A., J. Leiwen, and B. C. O’Neill. 2002. Demographic Composition and Projections of Car Use in Austria. Working Paper WP 2002–034. Max Planck Institute for Demographic Research, Rostock, Germany. 4 7G E N D E R D I F F E R E N C E S I N A U T O M O B I L E C H O I C E B E H AV I O R 98709mvpTxt 41_88 9/20/05 5:05 PM Page 47

Revelt, D., and K. Train. 1998. Incentives for Appliance Efficiency in a Competitive Energy Environment: Random Parameter Logit Models of Households’ Choices. Review of Economics and Statistics, Vol. 80, No. 4, pp. 647–57. Roorda, M. J., A. Mohammadian, and E. J. Miller. 2000. Toronto Area Car Ownership Study: A Retrospective Interview and Its Applications. In Transportation Research Record: Journal of the Transportation Research Board, No. 1719, TRB, National Research Council, Washington, D.C., pp. 69–76. Salvini, P., and E. J. Miller. 2003. ILUTE: An Operational Prototype of a Comprehensive Microsimulation Model of Urban Systems. Presented at the 10th International Conference on Travel Behaviour Research, Lucerne, Switzerland, Aug. 2003. Wachs, M. 1997. The Gender Gap: How Men and Women Developed Different Travel Patterns. ITS Review, Vol. 20, No. 2, pp. 1–2. Yamamoto, T., R. Kitamura, and S. Kimura. 1999. Competing-Risks-Duration Model of Household Vehicle Transactions with Indicators of Changes in Explanatory Variables. In Transportation Research Record: Journal of the Transportation Research Board, No. 1676, TRB, National Research Council, Washington, D.C., pp. 116–123. 4 8 RESEARCH ON WOMEN’S ISSUES IN TRANSPORTATION 98709mvpTxt 41_88 9/20/05 5:05 PM Page 48

4 9 Differences in Trip Chaining by Men and Women Nancy McGuckin, Travel Behavior Analyst, Washington, D.C. Yukiko Nakamoto, SAS Programmer, Columbia, Maryland Men’s and women’s commuting behavior continues to be distinctly different. The difference may be most apparent in the tendency to trip-chain—that is, to link short stops in the trip to or from work. As more women entered the workforce and went from higher education to profes- sional careers, it was widely assumed that aspects of women’s and men’s travel behavior would converge. However, research has found persistent gender differ- ences in distance to work, mode of travel, and automo- bile occupancy and in the propensity to trip-chain. This study examines whether trends in trip-chaining behavior show convergence or the continued persistence of gender differences. Trends show that trip chaining during the commute increased from 1995 to 2001, and men’s trip chaining increased nearly twice as much as women’s. The growth in men’s trip chaining is robust, but a large amount of that growth is for stops to get a meal or coffee on the way to work, called the Starbucks effect. Clarify- ing trends in the incidence of trip chaining and, more important, the details in terms of the direction, time of day, and purpose of the stops during commuting helps in the understanding of the persistence of gender roles in travel behavior. Such an understanding is vital to policy directives that aim to change travel behavior to ease congestion, reduce emissions, and save fuel. Women’s travel is distinctly different from thatof men. Overall, working-age women makemore trips but travel fewer miles and minutes than their male counterparts do. During their com- mutes, women make more short stops and stop for dif- ferent reasons than men. In addition, the occupations and job locations of working women are different from those of men—women work closer to home than men, even within the same general occupation categories. The focus of this study is to examine trends in trip chaining between men and women. Levinson and Kumar (1995) reported an increase in trip chaining. They related the increase to higher family incomes and less time as women entered the workforce and the fact that dual- career households buy services (such as daycare) that were formerly conducted in the home. McGuckin and Murakami (1999) determined that trip chaining was pre- dominantly the domain of women rather than men in the household, even as women entered the workforce. Bianco and Lawson (1997) found specifically that the work trip was becoming more complex as workers incor- porated personal, household, and child-care activities into their commutes. Likewise, Nishii et al. (1988) dis- covered that an important secondary role for the work trip was to provide an opportunity to link nonwork travel. Few researchers have examined trends in trip- chaining behavior. Definitions of what constitutes an incidental stop between destinations such as home and work can complicate the comparisons between years or areas, or both. In this study, trends are derived from the 1995 Nationwide Personal Transportation Survey (NPTS) and the 2001 National Household Travel Sur- vey (NHTS). Both data sets were processed by using the same definitions, which identify an incidental stop during the commute as one of 30 min or less. The sim- 98709mvpTxt 41_88 9/20/05 5:05 PM Page 49

ilarities in survey design and the definition of variables allow comparison between the two years, although the short time frame between the surveys gives an indica- tion of a direction rather than a trend. However, the findings suggest that there has been an increase in trip chaining during the weekday commute from 1995 to 2001. From 1995 to 2001, women made more short stops on the way to or from work than did men to perform household-sustaining activities, such as shopping and family errands, and working women in two-worker families were twice as likely as men to pick up and drop off school-age children at school during their commute. In the same period (1995 to 2001), men added more stops to their commutes for child care and household errands, especially men in families with young children. But a substantial part of the growth in men’s trip chain- ing was to make a short stop for a meal or coffee on the way to work. To the extent that trip chaining is a more efficient use of time and fuel, increasing such behavior is good. But what if the increase in trip chaining is for the kind of activities that were previously done at home, such as breakfast and coffee, but now are an added trip during the morning commute? If adding a trip changes the travel route or departure time, it complicates the fore- casting of travel demand. Since adding a trip may add an engine start (or, for drive-through windows, idling time), such stops may not bode well for air quality. Clarifying trends in the incidence of trip chaining and, more important, the details of trip-chaining behav- ior in terms of the direction, time of day, and purpose of the stops during commuting helps the understanding of the persistence of gender roles in travel behavior. Such an understanding is vital to policy directives that aim to change travel behavior to help ease congestion, reduce emissions, and save fuel. DEFINITION OF TRIP CHAINING The NHTS, like most household travel surveys, collects travel information about trips—movement from one address to another. In this way, every movement by any mode for any distance is reported for all respondents. A trip chain is a sequence of trips linked together between two anchor destinations, such as home and work. Economists, geographers, and transportation planners have recognized trip-chaining behavior since the 1960s, but even then the conceptualization of a trip chain was easier to agree on than the definition (Thill and Thomas 1987). Today there is still no formal agreement on the defin- ition of a chained trip, and little empirical research has been published on the incidence of or trends in trip- chaining behavior within comparable travel markets or with comparable data for the same market. Different terms and expectations exist as to what types of trips should be considered as part of a chain—only trips for certain purposes (e.g., dropping off a passenger) or only trips with certain dwell times (e.g., 15 min or less). Some of the earlier national research used no time or purpose constraints at all (Strathman and Dueker 1995; McGuckin and Murakami 1999), so direct comparison with that earlier work is difficult. Trip chaining may be difficult to define, but this com- mon behavior complicates the understanding of com- muting. As in the example in Figure 1, persons can make a total of four separate trips but two chained trips dur- ing their commute—from home to a coffee shop to work and then from work to a daycare center and then home. If these are considered separate trips, the trip from home to the coffee shop would not normally be seen as part of the commute. But linked together with the next trip, from the coffee shop to work, it can be seen as part of the chain of trips from home to work. Stops made during a commute may be regular daily activities, weekly scheduled activities, or infrequent and unscheduled. The stops may take the traveler well out of his or her way or be close to home or work. Since the NPTS-NHTS data series obtains travel for a sample day, the frequency of stops during an individual worker’s weekly commute is not known. Destinations in the national data sets are not geocoded to latitude and lon- gitude, so the proximity to home and work is also not known. However, using a common definition (trip chains include stops of 30 min or less) allows analysis of the change in incidence and purpose of short stops during the commute from 1995 to 2001. The analysis presented here describes the trends in trip chaining with this definition. 5 0 RESEARCH ON WOMEN’S ISSUES IN TRANSPORTATION Coffee Shop Home Trip 2 Trip 3 Trip 4 Work Daycare center Trip 1 FIGURE 1 Example of four trips and two trip chains. 98709mvpTxt 41_88 9/20/05 5:05 PM Page 50

DIFFERENCES IN WORK LOCATION AND OCCUPATION During the last several decades, women have increased their driving and have their own vehicles, better educa- tion, and professional careers, and as a result women’s travel has grown immensely and the nature of their travel has radically changed. These changes have had a perma- nent effect on travel behavior analysis and transportation planning and policy. Working women make more trips than working men (110 more trips per year), but the trips are shorter on average. The largest differences are found in households with small children, in which women make 5.2 trips to men’s 4.7 but travel on average 12 fewer miles a day (Table 1). One of the reasons for the difference in travel miles is that women have traditionally held jobs closer to home than men did. In 2001, women reported working about 11 mi from home compared with 15 mi for men. Wachs (1998) notes that although women have moved into the labor force, they are overrepresented in what is called the secondary work force, consisting of part-time and seasonal workers. The NHTS shows that women are twice as likely to be part-time workers (12% of women but only 6% of men work part time). Second, Wachs says that women are concentrated in occupations such as clerical and sales work. These posi- tions pay lower wages, and it is argued that women do not search for jobs farther away because they gain no wage advantage in these traditional occupations. Third, Wachs argues that women may work closer to home because the suburbanization of service and retail activity has resulted in more even distribution of women’s jobs than of the professional and technical jobs typically held by men. Spain (1999) asserts that the jobs in which women are concentrated (teaching, clerical work, and nursing) differ in that they are closer to home and require less travel. Figure 2 shows the distance to work measured as the crow flies [the great circle distance (GCD)]. These data demonstrate that even within the same general occupa- tional category, women choose jobs closer to home. Women in professional, managerial, and technical jobs work, on average, 2.6 mi closer to home than do men in the same occupations (9.9 mi versus 12.5 mi). Because these occupational categories are large and inclusive, it would be interesting to analyze differences in distance to work from an establishment or work- place survey to see if directly comparable occupations in the same location showed the same differences in trip length to work. Another explanation offered in the literature focuses on family commitments (child and elder care). For example, Gordon et al. (1989) postulate that women may work closer to home to minimize their work- related travel time and maximize the time they have for non-work-related trips and activities. Single women work closer to home than do single men, but the difference in distance to work becomes more pronounced in two-adult families with children. Some evidence of the amount of time women spend in household-sustaining activities comes from the recent American Time Use Survey (Bureau of Labor Statistics 2003). The 2003 results show that employed adult women (18 and over) spend about an hour more per day than employed adult men doing household activi- ties and caring for household members. Twenty percent of men reported doing housework on the survey day compared with 55% of women, and 35% of men did food preparation or cleanup compared with 66% of women. 5 1DIFFERENCES IN TRIP CHAINING BY MEN AND WOMEN TABLE 1 Miles and Minutes of Travel per Day for Men and Women Workers, 2001 Trips/Day Miles/Day Minutes/Day Men No child 4.6 62 103 Small child (<6 years) 4.7 58 104 Middle child (6–15 years) 4.8 63 105 Teen child (16–21 years) 4.8 52 98 Women No child 4.7 50 94 Small child (<6 years) 5.2 46 90 Middle child (6–15 years) 5.4 47 93 Teen child (16–21 years) 5.0 45 90 0.0 2.0 4.0 6.0 8.0 10.0 12.0 14.0 Sales/Serv Clerical/Admin Manufacturing Professional Miles Men Women FIGURE 2 Distance to work (GCD) by occupational category. 98709mvpTxt 41_88 9/20/05 5:05 PM Page 51

TRENDS IN MEN’S AND WOMEN’S TRIP CHAINING Almost 2 million more workers stopped during their com- mutes in 2001 compared with 1995—overall, 9% more workers trip-chained during their commute. This statistic compares with an 8% growth in civilian employment in the same time period, according to the U.S. census. Men’s and women’s stops both increased, especially in the home-to-work direction, but for men the increase was dramatic: 24% more men stopped during their home-to-work commute in 2001 than in 1995 (Table 2). The typical demographic variables used to forecast travel demand (e.g., age, income, and geography) did lit- tle to explain differences in trip-chaining behavior as found by Cao and Mokhtarian (2004) and by Li et al. (2004). Especially disappointing was the failure of the geographic variables available in the NHTS to show descriptive differences. Instead, the starkest differences in trip-chaining behavior, such as the number and purpose of stops, were related to life cycle, especially combined with sex and the presence of children. Twenty percent of men in families with two adults and small children trip-chained, more than any other group of men, and the change since 1995 is noteworthy (see Figure 3). Men in households with teenaged children or no children also showed increases in trip-chaining behavior. More than 40% of the women in two-adult house- holds with small children chained nonwork trips into their commutes, a percentage that grew little between 1995 and 2001 (see Figure 3). However, as with men, women who have teenaged children in the household or those with no children increased their trip chaining. The proportion of men and women stopping during their commutes to or from work varied within race and ethnicity, as shown in Figure 4. Hispanic men are the least likely to report stopping for any purpose during their com- mutes, whereas Hispanic women were almost as likely as whites and African-Americans to trip-chain. African- American workers, women in particular, are more likely to stop during their commutes than any other group. REASONS FOR STOPS DURING COMMUTE The most common purposes for stops in the commute to work are to drop off or pick up a passenger, to do family or personal errands, or to buy a meal or coffee. However, the most common stops after work are to shop, serve a passenger, or run family errands. 5 2 RESEARCH ON WOMEN’S ISSUES IN TRANSPORTATION TABLE 2 Percentage Change in Number of Workers Who Trip-Chained in 1995 and 2001 1995 2001 % Change 1995–2001 Chain home–work Men 4,378,082 5,441,096 24.3% Women 6,060,274 6,553,425 8.1% Chain work–home Men 5,942,466 6,076,712 2.3% Women 6,471,233 6,767,123 4.6% 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0.5 2+adlt, 0 child 2+adlt, child<5 2+adlt, child 6–15 2+adlt, child16–21 2+adlt, retired Mean Stops (All) 1995 Men 2001 Men 1995 Women 2001 Women FIGURE 3 Mean number of stops by life cycle. 98709mvpTxt 41_88 9/20/05 5:05 PM Page 52

Dropping off and picking up a passenger are com- mon stops in both directions. Of all the multioccupant vehicle trips to and from work, three-fourths were “fam-pools” (all occupants were from the same house- hold) and women drove 60% of those. Of the remaining fourth of carpools with nonhousehold members, men drove 64%. Many of the passengers in fam-pools are children being dropped at daycare or school or other activities by parents. This discovery is compatible with the finding by Lee and Hickman (2004) that the presence of chil- dren in households positively affects the duration of out- of-home activities. The recently released American Time Use Survey (Bureau of Labor Statistics 2003) indicates that the average woman in the United States spends 84 h a year picking up and dropping off household children (in the American Time Use Survey all women are aver- aged, not just families with children). To understand this common type of stop during the commute, serve-passenger trips to drop off or pick up chil- dren (less than 14 years old) were examined further. According to the 2001 NHTS, more than 7 million families with two working parents drop off or pick up their chil- dren (less than 14 years old) during a weekday commute. In total, 2.7 million men and 4.3 million women insert a drop-off or pickup trip (or both) into their work trip. When two working parents commute to work, twice as many trips to drop off or pick up a child are made by women (66% compared with 34% for men). Eighty per- cent of the drop-off trips occur before 9:00 a.m. on weekdays, but perhaps because of “after-care” and after- school activities, the pickup trips are not so clustered. Another common reason to stop during a commute is to shop and conduct family errands, especially on the way home from work, and Figure 5 shows that women make more of these stops than men. There is dramatic variation by race or ethnicity in the trends for stopping to shop during the work-to-home commute, as shown in Figure 6. Between 1995 and 2001, the number of shopping stops by Hispanic com- muters increased more than 20%, whereas the number of stops to shop by African-American commuters declined by 12%. As the research presented here has shown, men have increased their incidence of trip chaining, and the types of trips that women traditionally insert into their com- mutes, such as serving a passenger, running errands, and shopping, have increased modestly. One trend is intriguing and accounts for a surprising amount of the growth in men’s trip chaining: the increase in the number of stops to get a meal or coffee on the way to work, as shown in Figure 7. In just a 6- year span, more than 1.5 million more stops were added to get a meal or coffee (1995 to 2001). There was a large increase in the number of such trips by both women and men workers, but especially by men. This effect is called the “Starbucks effect.” CONCLUSIONS AND FURTHER RESEARCH The analysis presented here relied on a definition of a trip chain as a sequence of trips bounded by stops of 30 min or less. This operational definition facilitates a rich 5 3DIFFERENCES IN TRIP CHAINING BY MEN AND WOMEN 0 5 10 15 20 25 30 35 White Afr-American Hispanic Asian Percent Men Women FIGURE 4 Percentage of workers who made stops during commuting. 98709mvpTxt 41_88 9/20/05 5:05 PM Page 53

analysis of trip-chaining behavior, and thus the authors invite other travel researchers to use it. The chained files for the 1995 NPTS and the 2001 NHTS are publicly available for researchers and analysts on the NHTS website (nhts.ornl.gov/2001/index.shtml). An important finding is that the presence of children continues to affect women’s travel patterns more than men’s. Overall, women work closer to home than men, even within the same occupational categories. This char- acteristic is especially marked when young children are present in the household. Women are twice as likely as men to drop off or pick up children in two-worker households. Further, such trips are highly constrained within the morning and evening peaks. This finding suggests that women may have less flexibility in departure time than men since school and daycare start and end times may influence the commute times of women workers more than those of men. 5 4 RESEARCH ON WOMEN’S ISSUES IN TRANSPORTATION 0 5 10 15 20 25 30 35 Shopping and Errands Serve Pass. Meals/Coffee Shopping and Errands Serve Pass. Meals/Coffee H om e- to -W or k W or k- to -H om e Percent of Stops Men Women FIGURE 5 Percentage of stops by men and women by selected purpose. –15 –10 –5 0 5 10 15 20 25 Afr-Am Hisp White W or k- H om e Percent Change 1995–2001 FIGURE 6 Percentage change in stops for shopping by race and ethnicity, 1995–2001. 98709mvpTxt 41_88 9/20/05 5:05 PM Page 54

The evidence continues that travel is a gender-related activity. The household and child-care responsibilities of women make it likely that women will chain some of those tasks into the commute. One surprising finding from the trends shown here is that in-home activities, such as cooking meals, are being replaced with activities requiring travel—picking up a meal. This finding is true even for a cup of coffee in the morning, no longer brewed at home but purchased at the local coffee shop on the way to work. Many researchers expected an effect on travel linked to the growth of the Internet, and congestion relief from telecommuting continues to be sought, but the apparent substitution of travel for what was traditionally an in- home activity (breakfast and coffee) needs further study. One of the biggest questions for the future is how household dynamics, social roles and expectations, and perhaps market and lifestyle changes will affect the travel behavior of both men and women. Although women have made great strides and accom- plishments in the last quarter-century, change in societal expectations is slow. Differences in travel related to gen- der roles persisted over the short time frame studied, although there are indications that men in households with small children have increased their trip chaining for household- and child-related purposes. Perhaps there is a cohort effect in the coming generation. More research is needed. Especially interesting ques- tions have been raised on the effect of geographic factors and commuting distance on the probability of trip chain- ing. Nishii et al. (1988) also raised the importance of incorporating more information about the travel envi- ronment into the analysis of trip chaining. The clear next step is analysis with a geographic component. In addition to these questions about the conditions of the travel environment that encourage or discourage trip chaining, further research into the conditions of the traveler, specifically looking at demographic factors as this study looked at gender and life cycle, would be useful. DATA USED IN THIS RESEARCH An NHTS has been conducted by the U.S. Department of Transportation periodically since 1969 to obtain an inventory of daily travel for the nation. Details about the survey methods, questions, and weighting can be found at nhts.ornl.gov/2001/index.shtml. Most important for trend analysis, the 2001 NHTS and the 1995 NPTS were processed simultaneously with the same rules and logic streams to develop the trip chains analyzed in this research. Changes in trip-chaining behav- ior found during comparison of the 1995 NPTS and the 2001 NHTS, when statistically significant, are not arti- facts of differences in scope, methodology, or question wording. REFERENCES Bianco, M., and C. Lawson. 1997. Trip Chaining, Childcare, and Personal Safety. In Women’s Travel Issues: Proceedings from the Second National Conference, Report FHWA-PL-97-024, FHWA, U.S. Department of Transportation. Bureau of Labor Statistics. 2003. American Time Use Survey. U.S. Department of Labor. www.bls.gov/tus/home.htm. 5 5DIFFERENCES IN TRIP CHAINING BY MEN AND WOMEN 0 200,000 400,000 600,000 800,000 1,000,000 1,200,000 Women Men Women Men Home–Work Work–Home 1995 2001 FIGURE 7 Number of stops to get meals or coffee, 1995–2001. 98709mvpTxt 41_88 9/20/05 5:05 PM Page 55

Cao, X., and P. Mokhtarian. 2004. How Do Individuals Manage Their Personal Travel? Objective and Subjective Influences on the Consideration of Travel-Related Strategies. Presented at 83rd Annual Meeting of the Transportation Research Board, Washington, D.C. Gordon, P., A. Kumar, and H. Richardson. 1989. Gender Differences in Metropolitan Travel Behavior. Regional Studies, Vol. 23, No. 6, pp. 499–510. Lee, Y., and M. Hickman. 2004. Household Type and Structure, Time Use Pattern, and Trip Chaining Behavior. Presented at 83rd Annual Meeting of the Transportation Research Board, Washington, D.C. Levinson, D., and A. Kumar. 1995. Activity, Travel, and the Allocation of Time. APA Journal. Li, H., R. Guensler, J. Ogle, and J. Wang. 2004. Using Global Positioning System Data to Understand Day-to-Day Dynamics of Morning Commute Behavior. In Transporta- tion Research Record: Journal of the Transportation Research Board, No. 1895, Transportation Research Board of the National Academies, Washington, D.C., pp. 78–84. McGuckin, N., and E. Murakami. 1999. Examining Trip- Chaining Behavior: Comparison of Travel by Men and Women. In Transportation Research Record: Journal of the Transportation Research Board, No. 1693, TRB, National Research Council, Washington, D.C., pp. 79–85. Nishii, K., K. Kondo, and R. Kitamura. 1988. Empirical Analysis of Trip Chaining Behavior. In Transportation Research Record 1203, TRB, National Research Council, Washington, D.C., pp. 48–59. Spain, D. 1999. Societal Trends: The Aging Baby Boom and Women’s Increased Independence. In Searching for Solutions, Report FHWA-PL-99-003, FHWA, U.S. Department of Transportation. Strathman, J. G., and K. J. Dueker. 1995. Understanding Trip Chaining. In 1990 NPTS Special Reports on Trip and Vehicle Attributes, FHWA, U.S. Department of Transportation, pp. 1–27. Thill, J.-C., and I. Thomas. 1987. Toward Conceptualizing Trip-Chaining Behavior: A Review. Geographical Analysis, Vol. 19, No. 1. Wachs, M. 1998. The Automobile and Gender. In Women’s Travel Issues: Proceedings from the Second National Conference, Report FHWA-PL-97-024, FHWA, U.S. Department of Transportation. 5 6 RESEARCH ON WOMEN’S ISSUES IN TRANSPORTATION 98709mvpTxt 41_88 9/20/05 5:05 PM Page 56

Next: TRANSPORTATION, ACCESS, AND COMMUNITY DESIGN »
Research on Women's Issues in Transportation - Volume 2: Technical Papers Get This Book
×
MyNAP members save 10% online.
Login or Register to save!
Download Free PDF

TRB’s Conference Proceedings 35: Research on Women’s Issues in Transportation – Volume 2: Technical Papers contains peer-reviewed breakout and poster papers and several abstracts of papers presented at the November 18–20, 2004, conference in Chicago, Illinois. The conference was designed to identify and explore additional research and data needed to inform transportation policy decisions that address women’s mobility, safety, and security needs and to encourage research by young researchers. Volume 1, which will be released this winter, will include the conference summary, the four peer-reviewed overview papers presented by the topic leaders, and a list of conference participants.

  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!