Understanding How Individuals Make Travel and Location Decisions: Implications for Public Transportation (2008)

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.

19 This chapter discusses the literature, theories, and data concerning the factors that influence where people choose to live, work, and travel. The first section presents major trends in population, employment, and mode choice in the United States. It also includes a look at how age and life-cycle stage affect residential density of households. The second section presents research on the effect of land use on travel behavior. After a presentation of several com- prehensive reviews of the research, additional research is pre- sented that focuses on the question of whether living in higher density neighborhoods affects travel behavior. A third section looks at the question of whether trans- portation accessibility affects residential choice. Evidence from surveys of homebuyers and from residential choice models is included. Overall Trends Before discussing the variety of research on the association between land use and travel behavior, it is instructive to review trends in the United States over the past several decades. The common perception is that residences and jobs have been migrating to the suburbs from the central city, and that automobile travel has grown so that it dwarfs travel by transit. This perception is found in extensive literature on sprawl and on the consequences of automobile dominance (2, 3, 4). The statistics on trends in the United States confirm this general perception, with the caveat that recent decades are showing more stability in residences and jobs in the cen- tral city and that transit use appears to have stabilized. A well-known trend is the suburbanization of residences, which has been occurring since the time of the streetcar and which has accelerated beyond the growth of the overall pop- ulation. Another trend has been the decline in the popula- tion living outside metropolitan areas. The percentage of the population living in metropolitan areas has increased from 28.4% in 1910 to 80.3% in 2000. The percentage of the population living in the suburbs went from 7.1% in 1910 to 50% in 2000 (5). While the population of the United States tripled between 1910 and 2000, the population in center cities quadrupled, and the population in suburban areas in- creased by a factor of more than 21. Center city population has been approximately 30% of the total since around 1920. Figure 2-1 shows the number of people in the United States living outside metropolitan areas and in suburban and cen- ter city areas. Jobs have also moved to the suburbs, although not at the same rate as residences. For example, manufacturing jobs have declined from almost 70% in central cities around the time of World War II to 50% in 2000 (6). Total employment in the central city appears to have stabilized during the past decade, however. Journey-to-work data from the Census Bureau shows that between 1990 and 2000 the percentage of jobs in the center city actually increased slightly, from 40.8% to 41.6%. Jobs in the remainder of the metropolitan statistical area increased from 37.0% to 39.2%, whereas jobs outside the metropolitan statistical area declined from 22.1% to 19.1%. Figure 2-2 shows the number of workers by place of work for the United States. Public transportation declined in absolute terms during the last half of the 20th century, going from a high of 23.4 billion unlinked trips in 1946 to a low of 6.5 billion unlinked trips in 1972 (7). Transit use as a percentage of overall travel declined during the last half of the 20th century and remains only slightly more than 1% of all passenger miles. While there are many reasons for this trend, the suburbanization of housing and jobs is one key reason. Figure 2-3 shows the percentage of public transit passenger miles out of the universe of auto pas- senger miles and transit passenger miles since 1980 (8). Transit mode split is more significant for the journey to work. Figure 2-4 shows the percentage of trips by alternative modes for the work purpose. Alternative modes represent a little more than 10% of trips, and transit increased from 5.4% to 6.7% between 1985 and 2001 (9). C H A P T E R 2 The Relationship Between Residential Choice, Transportation, and Life-Cycle Stage

While these trends are not in dispute, there are alternative perceptions of what might happen in the future—whether better land use and transportation policies could promote better outcomes. There is also vast interest in the potential for land use development programs called nontraditional, transit-oriented design (TOD), which are referred to in this research as compact neighborhoods (CNs). The hope is that if more communities are formed that are higher density, with a fine-grain mix of land uses, there will be less use of auto- mobile trips and higher use of walking, biking, and transit trips. Such developments will, it is believed, promote more use of alternative modes (walking and transit), cause a de- crease in vehicle miles traveled, and provide a high quality of life. Census data can be used to examine basic lifestyle charac- teristics of those who might be more inclined to choose CNs. Because CNs are associated with higher than normal densities, and the detached single-family home plays a smaller 20 0 50 100 150 200 250 300 Year 1900 1910 1920 1930 1940 1950 1960 1970 1980 1990 Number outside metro areas Number in suburbs Number in central cities Figure 2-1. United States population, in millions, by residential location (5). 0 20 40 60 80 100 120 1990 2000 Work outside MSA Work in remainder of MSA Work in central city 0.00% 0.50% 1.00% 1.50% 2.00% 1980 1990 1994 1995 1996 1997 1998 1999 2000 Figure 2-2. Place of work in the United States (millions of workers). Figure 2-3. Transit passenger miles (percentage of total).

than normal role in these settlements, it can be observed that the selection of housing other than the single-family home varies over a family’s life cycle. Figures 2-5 and 2-6 document the choice of higher density housing as a function of age and as a function of stage in the life cycle of one particular group in the population—namely, family units of two parents with children. (A similar graph could be created, for example, for single-parent households.) Because of the similarity of pat- terns, the two graphs can be observed together. The graphs are based on the analysis of the results of the National Household Travel Survey (2001) undertaken by the Federal Highway Ad- ministration and the Bureau of Transportation Statistics (10); the graphs are based on a sample of the U.S. population living in urbanized areas. Figure 2-5 shows that more than 60% of single Americans with no children in urban areas live in multiple-unit housing. Figure 2-6 reflects this by showing that more than 50% of Americans between the ages of 21 and 25 in urban areas sim- ilarly live in multiple-unit housing. By the time the youngest child is over 5 years of age, the percentage of households living in multiple-unit housing declines to about 20%, as shown in Figure 2-5. The same phenomenon is shown in Fig- 21 0.0 2.0 4.0 6.0 8.0 10.0 12.0 1985 1989 1993 1997 1999 2001 Walk only Bicycle or motorcycle Public transportation Figure 2-4. Alternative modes for the journey to work (percentage of trips). "TOD" Type Housing by Life Cycle Stage 0 10 20 30 40 50 60 70 Single Couple Under 5 Under 15 Over 16 Couple Single No kids Couple With Kids No kids Phase in Life Cycle Pe rc en t i n M ul tip le U ni t H ou si ng Prime Market for Survey Process Se con dar y M ark et for Sur vey Pro ces s “TOD”= Transit Oriented Development Figure 2-5. Life-cycle stages and use of multiple-unit dwellings.

ure 2-6 as a function of age of the individual, with slightly more than 20% choosing higher density housing between ages 41 and 55. This rather basic tabulation from the National Household Travel Survey provides support for the concept that different stages of the life cycle (or age) involve different forces on the residential decision-making process. For young individuals who have not started the child rearing process, higher density living patterns are the accepted norm. By the time their chil- dren are old enough to be in school, however, the use of higher density residential patterns has fallen by about half. At some point in the aging process, there is a return to the use of multiple-unit housing patterns. Given these overall demographic trends and the percep- tion that better policies could produce better outcomes, what does the research tell us that will help policymakers un- derstand how and why people are making these choices and that will also provide some policy levers for influencing choice? The following section includes a discussion of the relevant research on the relationship between land use and transportation. Literature on the Effect of Land Use on Travel Behavior The evidence for the effect of land use on travel behavior is the subject of an extensive body of literature, and thus a number of authoritative critical reviews of this literature are available. In this chapter, the key results of reviews by Crane (11), Ewing and Cervero (12), Cervero et al. (13), Handy (14), and Kuzmyak et al. (15) are presented. Each of those reviews has a unique emphasis, but all share common themes specific to the subject, including methodological challenges, relevant theoretical frameworks, and range of travel behavior effects. Following the summary review papers, this chapter includes a discussion of several additional studies that provide a back- ground for this project. These papers provide information on the relationship between urban design, walking, and other transportation uses, as well as on the relationship between at- titudes or lifestyle and urban design. Summary Review Papers The Influence of Urban Form on Travel: An Interpretive Review—Randall Crane’s review is focused largely on the methodology limitations of past research and thus the collec- tive validity of findings (11). His review specifies three types of past research: hypothetical or simulation studies, descriptive studies, and multivariate statistical studies. Crane finds that hypothetical or simulation models provide little insight into the study of the effect of land use on travel behavior. These models can relate different scenarios “given certain behavioral assumptions,” but these assumptions are “too simplistic,” are “not intended to explain behavior,” and thus “cannot test hypotheses with regard to the effect of land use on travel behavior” (pp. 5–6). With respect to descriptive studies, Crane concludes that these studies have made some contribution to our understanding of the effect of land use on travel behavior (e.g., by providing “hard data on real behaviors,” p. 5), but that these studies have limited utility because they lack a the- oretical basis and cannot isolate the effect of land use variables from other competing explanatory variables (e.g., it is impos- sible to use such studies to identify “how much of the observed behavior is influenced by the street configuration or any spe- cific design feature alone” (p. 8). Crane identifies two cate- gories of multivariate statistical studies—ad hoc models and demand models. He finds that ad hoc models are of limited value because they lack a behavioral or theoretical foundation even though they “consider many measures of urban form 22 "TOD" Type Housing as a Function of Age 0 10 20 30 40 50 60 0- 5 6 to 1 5 16 -2 0 21 -2 5 26 -3 0 31 -3 5 36 -4 0 41 -4 5 46 -5 0 51 -5 5 56 -6 0 61 -6 5 66 -7 0 71 -7 5 76 -7 9 80 -8 4 85 + Age Pe rc en t M ul tif am ily R es id en ts Prime Market for Survey Process Sec ond ary Ma rke t for Sur vey Pro ces s “TOD”= Transit Oriented Development Figure 2-6. Age versus multifamily residence.

while attempting to control for differences among communi- ties, neighborhoods, and travelers” (p. 11). Demand models based on a microeconomic theoretical framework are deemed most promising, but unfortunately relatively few studies in the past decade have used this approach. Crane recommends that future “empirical work with strong behavioral foundations may be a useful and rigorous way to systematically link urban form to travel choices” (p. 4). Crane concludes that the group of relationships encom- passing urban form and travel behavior is “complex” and our knowledge regarding them is “tentative” (p. 3). He writes that “little verifiable evidence supports the contention that changes in urban form will affect travel as intended at the scale proposed” (p. 3). This, he continues, has been polarized into a black and white issue for many (“believers or skep- tics”), and as a result, many civil decision makers have been left to make their own conclusions on often limited and com- plicated results (p. 3). Travel and the Built Environment: Synthesis—In their review, Ewing and Cervero acknowledge the methodological limitation of available studies, but seek to summarize the collective weight of the evidence of the land use effects on a range of travel behaviors (12). They state that their synthesis focuses on “[examining] research designs without getting bogged down in details, and [generalizing] across studies without glossing over real differences” (p. 1). The empirical studies reviewed, most of which controlled for competing explanatory variables, explain the following four types of travel effects: “trip frequencies (rates of trip making), trip lengths (either in distance or time), mode choices or modal splits, and cumulative person miles traveled (PMT), vehicle miles traveled (VMT), or vehicle hours traveled (VHT)” (p. 2). Ewing and Cervero find that mode choice, of all the types of travel effects, has “received the most intensive study” (p. 13) and is “most affected by local land-use patterns” (p. 7). However, mode choice depends on both the built environ- ment and socioeconomic factors, “though probably more on socioeconomics” (p. 13). They also find differing influences of land use variables on transit and walking mode choice: transit use tends to depend primarily on “local densities, and secondarily on the degree of land-use mixing,” while walking tends to depend on both equally (p. 7). In addition, compos- ite measures of the quality of the transit and walking envi- ronment can also influence the choice to use transit or walk. Ewing and Cervero note that research on the effect of land use on trip lengths is less abundant than on mode choice. The results of these studies generally find that trips are shorter as accessibility or density increases, or when mixed uses are applied. This, they say, “holds for both the home end (i.e., residential neighborhoods) and non-home end (i.e., activity centers) of trips” (p. 6). Unlike mode choice, trip lengths appear to be a function of the built environment first and of socioeconomic characteristics second. Trip frequencies, Ewing and Cervero contend, are like mode choice in that they depend on socioeconomic charac- teristics first. In fact, trip frequencies are mostly dependent on socioeconomic characteristics, and “differ little, if at all, between built environments” (p. 4) and “appear largely inde- pendent of land-use variables, depending instead on house- hold socioeconomic characteristics” (p. 6). Similarly, they consider the issues of whether substitution or supplementation accounts for “the disproportionate numbers of walking and transit trips in traditional urban settings . . . [with regards to] longer automobile trips that otherwise would have been made out of the neighborhood or activity center” and find that the weight of the current evidence supports the substitu- tion effect (p. 4). With respect to the effect of land use on total travel (PMT, VMT, and/or VHT), the authors find that when the effects of regional accessibility are isolated, studies they review “differ on the effects of local density and mix on total vehicular travel” (p. 5). Thus, regional accessibility plays a greater role, and “total household vehicular travel, whether VMT or VHT, is primarily a function of regional accessibility” (p. 5). Ewing and Cervero suggest that future research should study “how much of [an] impact density on travel patterns is due to density itself as opposed to other variables with which density co-varies” (p. 8). Consideration should also be given to the standardization of such terms as “transit friendliness” and “walking quality,” because in current studies their defi- nitions across the board are “unclear” (p. 12). Such issues warrant “much more empirical testing and replication of results” (p. 12). Another interesting area of research that has received relatively little attention is the influence of land use on trip chaining. Transit-Oriented Development and Joint Development in the United States: A Literature Review—In their literature review, Cervero, Ferrell, and Murphy address the more specific relationship between TOD and/or transit joint devel- opment (TJD) on transit ridership (13). The literature review included “secondary sources—comprising reports, articles, and books assembled from libraries, personal collections, and various public agencies” of a relatively recent date (p. 9). In general, the authors find a positive relationship between TODs or TJDs and transit ridership. However, they identify self-selection as an important control variable in these stud- ies; for example, one study found the following: TOD residents patronized transit as their predominant com- mute mode more than five times as often as residents countywide; self-selection was evident in that 40 percent of the respondents who moved close to transit stops said they were influenced in their move by the presence of LRT [light rail transit]. (p. 41) They cite another empirical study that found a statistically interdependent relationship between office development and 23

ridership: “jointly developed office space atop or near a rail stop spurred ridership and ridership in turn spurred office development” (p. 42). Another benefit of TODs is “increased off-peak and reverse-flow patronage—i.e., mixed-use, all-day trip generators help fill up trains and buses at all hours of the day and in both directions” (p. 42). The authors conclude that the “research shows that living and working near transit stations correlates with higher rider- ship” (p. 40). However, the authors caution that current research does not allow for definitive conclusions on the rela- tionship between TODs or TJD and transit use: No empirical research has been produced to date that traces causal pathways between TODs or TJDs, resulting ridership gains, and eventual improvements in traffic or environmental condi- tions. Given the daunting methodological challenges of conduct- ing such a causal analysis, qualitative case studies have been largely relied upon in making the connections between TODs and broader transportation and environmental outcomes. (p. 43) Smart Growth and the Transportation–Land Use Connec- tion: What Does the Research Tell Us?—In her 2005 synthe- sis, Handy also reviews the influence of land use on travel be- havior (14). Like the other reviewers, she finds that research to date has not established a solid foundation to predict the travel behavior effects of smart growth policies and strategies. Handy contends that some have “argued that the connection between transportation and land use has weakened,” while others believe that it “still greatly matters” (p. 2). She believes that the results from empirical studies are “mixed” and focuses her review on a microeconomic theoretical framework, current studies, and comprehensive reviews (p. 2). Handy begins her review by outlining the microeconomic theoretical basis of the land use and travel behavior hypothe- ses. She states that “travel choices made, such as the choice of mode or destination, are determined by the characteristics of the choices available. Each possible choice offers a certain ‘utility’ or value to the individual, who seeks to maximize her utility” and “maximizing utility generally means minimizing travel time, but other factors can outweigh time” (p. 20). This, in turn, results in a “mixed [effect] on travel for new urban- ism strategies: these strategies may increase the utility of alternatives to driving, but they also tend to increase the util- ity of making trips, so that savings from a shift in travel modes may be offset by increases in the frequency of trips” (p. 20). In terms of mode choice, trip length, and trip frequencies, Handy references Ewing and Cervero (12) and states that the weight of the evidence suggests that mode choice depends on socioeconomic and built environment characteristics (though more so on socioeconomic characteristics); trip length is a function of the built environment first and of socioeconomic characteristics second; and trip frequencies are just the oppo- site, first a function of socioeconomic characteristics and second a function of the built environment. Finally, in regards to VMT, “characteristics of the built environment are much more significant predictors of VMT, which is the outcome of the combination of trip lengths, trip frequencies, and mode split” (p. 21). Handy also discusses attitudinal variables, which, according to one study, “had the greatest impact on travel behavior among all of the explanatory variables and . . . residential location type had little impact on travel behavior, suggesting that ‘the associ- ation commonly observed between land use configuration and travel patterns is not one of direct causality, but due primarily to correlations of each of those variables with others’” (p. 23). Like Cervero et al. (13), she proposes that the connec- tion between travel behavior and residential type is better explained through self-selection—i.e., “residents with certain attitudes . . . [select] certain kinds of neighborhoods” (p. 23). Handy concludes that “new urbanism strategies make it easier for those who want to drive less to do so” (p. 24) and that “the lack of reliable predictions does not necessarily mean that communities should not proceed with smart growth efforts” (p. 26). She argues that determining the role socioeconomic characteristics play in determining travel behavior, separate from the built environment, is a challenge. She continues, “It is safe to conclude that land use and design strategies such as those proposed by the new urbanists may reduce automobile use a small amount” (p. 23). Continued research, Handy writes, has shown “promising” use of geographic information systems (GIS), which she believes will lead to “more detailed measures of the built environment . . .” (p. 25). She also recommends “experimen- tal designs and longitudinal studies . . . and analysis tech- niques, including path analysis, structural equations modeling, and multi-level modeling” (p.25). A key question is whether “land use and design strategies can fundamentally change atti- tudes towards transportation and thereby change desired be- havior rather than simply enabling it” (pp. 23–24). Land Use and Site Design—In Chapter 15 of TCRP Report 95: Traveler Response to Transportation System Changes, Kuzmyak et al. provide another comprehensive summary of the known impacts of land use on travel demand (15). The report looks at the impact on travel of building codes and site- level zoning requirements, as well as traditional neighborhood and pedestrian-friendly development. The summary judg- ment from the report is that much is still unexplained in travel behavior, even after land use and urban form are taken into consideration. While this chapter draws from a broad range of research stud- ies that have attempted to identify, measure and explain the links between land use and travel demand, the level of confidence imparted by these studies is less than with most measure reported elsewhere in this Handbook . . . 24

The better assessments are often made through development of regression or logit models. The resulting statistics almost always show, excepting certain narrowly focused investigations, that significant sources of variation in travel behavior still remain unexplained after key variables—land use, urban form and transportation—are incorporated, to a degree the same may be said of most conventional travel demand models, but not quite to the same extent. (p. 15–6) Selected Additional Studies As the reviews cited above point out, our knowledge of the effect of land use on transportation is limited. Unexplained variation in models of travel behavior based on land use means that much is left unknown about the relationship. The variables describing transit and walking-friendly urban design have not been carefully measured. In addition, if people self-select into neighborhood types according to their travel inclinations, then those inclinations, rather than urban design, would be the explanation for their travel patterns. Urban design might merely be enabling some residents to travel the way they prefer. A recently published study on the effect of urban form on walking (commonly known as SMARTRAQ) (16) attempts to address one of the key methodological limitations of pre- vious studies. Many studies on walking behavior rely heavily on self-reported data (often in the form of travel diaries) that are subject to validity concerns. The SMARTRAQ study addressed this problem by using accelerometers that elec- tronically recorded walking activity. Many studies are also limited by “large-scale regionally averaged . . . measures of the built environment that do not provide the detailed infor- mation needed by policymakers” (p. 117). SMARTRAQ addressed this as “. . . each environmental variable was com- puted individually for each participant, using GIS to describe the ‘microenvironments’ that people experience regularly where they live” (p. 118). The results revealed that “measures of land-use mix, residential density, and intersection density were positively related with number of minutes of moderate physical activity per day” (p. 117). Moreover, the study states the following: This research supports the hypothesis that community design is significantly associated with moderate levels of physical activ- ity. These results support the rationale for the development of policy that promotes increased levels of land-use mix, street con- nectivity, and residential density as interventions that can have lasting public health benefits. (p. 117) In sum, their analysis is more conclusive on the specific characteristics of land use that affect travel behavior related to walking. However, their study does not deal with the issue of self-selection. As noted by Handy (14), some studies have shown that attitudes are more important than land use characteristics as determinants of travel behavior. Kitamura et al. (17) devel- oped models to predict travel behavior given salient charac- teristics of neighborhoods, including measures of residential density, public transit accessibility, mixed land use, and the presence of sidewalks. Additional data were then added to the analysis of attitudinal variables, which grouped attitudes into factors with such labels as pro-environment, pro-transit, sub- urbanite, automotive mobility, time pressure, urban villager, and workaholic. When all of the explanatory variables were examined together, the attitudinal variables explained the highest proportion of the variation in the data. This led the researchers to suggest that land use policies promoting higher densities may not alter travel demand unless residents’ atti- tudes also change. The paper by Kitamura (17) provides support for the concepts being examined in this project, which call for the integration of psychological (attitudinal) research techniques into the set of tools utilized by the transportation manager and planner. A later paper by Bagley and Mokhtarian (18) “empirically examines the relationship of residential neigh- borhood type to travel behavior, incorporating attitudinal, lifestyle, and demographic variables.” In terms of both direct and total effects, attitudinal and lifestyle variables had the greatest impact on travel demand among all the explanatory variables. By contrast, residential location type had little impact on travel behavior. This is perhaps the strongest evidence to date supporting the speculation that the association commonly observed between land use configuration and travel patterns is not one of direct causality, but due pri- marily to correlations of each of those variables with others. In particular, the results suggest that when attitudinal, lifestyle, and sociodemographic variables are accounted for, neighborhood type has little influence on travel behavior (p. 279). The authors acknowledge that a drawback to their analysis is the use of cross-sectional data rather than longitudinal data. That is, people might change their attitudes over time in response to their residential environment. Thus, people do change, both their attitudes and their behav- ior, in response to external stimuli—the questions are, how many people, which kinds, how much, and how long does it take? A process of attitudinal and behavior adjustment, whether due to physical constraints as described above or due to a more subtle alteration of attitudes over time, comes into play most forcefully when people’s predispositions and residential loca- tions are mismatched, and the extent to which that is the case is unknown. The current study not only found little effect of resi- dential location on (travel) behavior, it found no impact of resi- dential location on attitudes . . . Travel behavior, on the other hand, showed a tendency to reinforce related attitudes: vehicle miles positively affected the pro-driving attitude and negatively affected the pro-high-density attitude, and the converse was true for walk/bike miles . . . However, a major limitation of the cur- rent study is the inability of the available cross-sectional data to 25

capture dynamic changes . . . To conclude, evidence strongly suggests that land use characteristics have little independent impact on travel behavior. But a need still exists . . . through the use of more appropriate data and analysis techniques, to improve our understanding of the extent to which one’s residential envi- ronment influences the attitudes and lifestyle that do affect travel demand. (p. 295) Bhat and Guo (19) reported on research that attempted to sort out the impact of the built environment on travel, sepa- rately from the effect of auto ownership and demographics. They found that the attributes of the built environment do affect residential choice decisions, as well as car ownership decisions. They also found that the commonly used popula- tion and/or employment density measures are actually proxy variables for built environment measures, such as street block density and transit accessibility. Both household demograph- ics and the built environment affected car ownership, with demographics being the more important. Household income was the key variable influencing the choice of type of resi- dential neighborhood and the accessibility of the neigh- borhood. The researchers indicated that ignoring the effect of income on car ownership and the travel decisions related to car ownership could lead to an inflated effect of the built environment on travel behavior. Finally, the results did not support the notion that unobserved factors (like attitudes) predisposed people to select certain types of residential neigh- borhoods or to make car ownership decisions. This result implies that independent models of residential choice and car ownership choice (after accommodating the res- idential sorting effects of demographics) are adequate to exam- ine built environment effects on car ownership choice, in the current empirical context. But, in general, it is important to con- sider the methodology developed in this paper to control for the potential presence of self selection due to both observed and unobserved household factors. Only by estimating the joint model can one conclude about the potential presence or absence of self-selection effects due to unobserved factors. (p. 20) Research on Choice of Residential Location Another approach to examining the relationship between land use and transportation is to examine the reasons that people choose certain residential locations and determine whether transportation options have an impact on the choice of residence. Research into the choice of residential location is extensive since it is of interest to those in the business of developing homes, as well as to policymakers wishing to influence residential location. On a more theoretical level, the trade-off between residential location and travel time has been a subject of much research in the related fields of geog- raphy, regional planning, economics, and transportation. Key to this project is to learn what the research says about variables that would encourage living in areas that feature TOD. Two sets of studies follow: (a) a selection of surveys of homebuyers and (b) academic research into residential choice. Surveys of Homebuyers There are many examples of homebuyer surveys that examine the reasons a particular home is purchased. The results of these surveys vary according to slight variations in the questions asked, and so caution is required in the interpreta- tion of results. An important source of information concern- ing the reasons for American residential location decisions is the report Smart Growth: A Resource for Realtors, which was prepared by the Economics Research Group of the National Association of Realtors (6). The document includes a discus- sion of the “Top Reasons for Deciding Where to Live,” as determined by a survey of registered voters in February 2000. More than 30% of the survey population selected “safe area with little or no crime,” with the second highest (17%) consid- eration being good public schools. “Ability to afford to live in neighborhood of choice” was third (10%). By contrast, access to stores (a key element in some TODs) was chosen by only 3% of the sample. The minimization of traffic congestion was ranked most important by just 5% of the survey. “Close to work” was chosen by 8% of the sample. Thus transportation- related considerations were ranked lower than other attributes of the home and neighborhood. In addition to its survey of voters, the National Association of Realtors also regularly surveys recent homebuyers. The 1999 sur- vey found that 82% of homes purchased that year were single- family homes, 7% were townhouses, and 8% were condomini- ums or apartments. The city is chosen by 44% of first-time buyers, but by only 36% of repeat buyers. Nearly half of the buy- ers within a city neighborhood are first-time buyers. In response to questions about why homebuyers moved, the two most cited reasons were the desire to own a home (33%) and space considerations (25%). The survey responses indicate some of the reasons why homebuyers are choosing suburban locations (20). Over three-quarters of the homebuyers said that a key rea- son for their decision to purchase a specific home was the neighborhood. Other factors that influenced buyers included the following: • Proximity to place of business—34% • Location and quality of local schools—32% • Parks/recreational facilities—15% • Shopping centers—13% • Public transportation—5% Note that while there is agreement between the two surveys quoted above on the relative importance of schools, the results 26

differ on the relative importance of being close to work and of access to shopping. Other findings are found in several other surveys summarized in a review by Malizia and Exline (21). For example, the 1998 Vermonters’ Attitudes on Sprawl Sur- vey found that 74% of respondents preferred a home in an outlying area with a larger lot and a longer commute over a similarly priced home in an urban area close to transportation, work, and shopping. That same survey found that 65% of re- spondents considered lot size as somewhat or very important when choosing a home. However, 48% preferred communi- ties with houses, stores, and services within walking distance. The National Association of Realtors study (6) points out that changes in demographics over the next decade may cause an increase in demand for city living. Because there will be a decline in the absolute number of households headed by persons aged 25 to 35—the ages at which households tradi- tionally leave cities for the suburbs—growth of the suburbs relative to cities will decelerate. The expected increase in single-family households will also increase demand for city housing, as these households opt for city living at higher rates than other households. Myers and Gearin (4) describe the results of a variety of surveys on home and neighborhood preference. A consistent share of respondents preferred alternative residential styles to the single-family home. Those preferring townhouses ranged from 15% to 17%; for condominiums, the range was 9% to 14%. Some consumers also prefer higher density living, rang- ing from 37% in a 1998 Professional Builder survey to 57% in a 1996 National Association of Home Builders survey. The 1998 American Lives survey found 49% of respondents pre- fer a less auto-oriented street pattern, with narrow streets that encourage walking. The Seattle Planning Department conducted a residential preference study to determine whether TOD developments might have appeal (22). The study involved a telephone sur- vey of 600 residents in the area to determine the most impor- tant features of a home. That was followed by a series of focus groups to further explore the findings from the survey. The third phase was a telephone and mail survey using the con- joint measurement technique to measure the importance of features in choosing housing. The objective of the study was to determine those persons who would be most likely to choose a denser housing environment, as well as to determine the features that would make such housing more appealing. The initial survey responses to rating questions about hous- ing preferences were used to segment the market into three different market segments using cluster analysis. Mirroring the National Association of Realtors study, crime and school quality were found to be important factors, but much less important than the type of residence and the desire for home ownership. Affordability was found to be slightly more important than concern about crime and schools. One segment was found to be much more likely to be interested in residences with greater density. That segment, named “Urban Village,” represented 34% of the population. This segment tended to have lower incomes than other groups, to be more mobile, and to rent rather than own their homes. The group had the largest proportion of college-age individuals and also a large number of retirees. This segment ranked affordability and crime as most important, followed by travel time to work and school quality. Models of Residential Location Understanding how homebuyers rank factors in home purchase decisions does not necessarily help to forecast home purchase decisions. For this a model of the choice process, which shows the effects of different factors and which sorts out cause and effect, is needed. The traditional economic approach to understanding res- idential location has been relied upon for years. The Dutch geographer Petter Naess (23) summed up the traditional approach as follows: According to theories of transport geography and transport economy, the travel between different destinations is assumed to be influenced on the one hand by the reasons people may have for going to a place, and on the other hand by the discomfort involved when traveling to this location (Jones, 1978; Beimborn, 1979). Or, in other words, by the attractiveness of the locations and the friction of distance. (p. 1) In the classic view, transportation is the cost that must be borne to make possible those things valued most highly. Early models, such as the gravity model, expressed attraction in easily quantifiable terms, such as square feet of space (in the numerator) and travel discomfort (as travel time or distance in the denominator), raised to an empirically determined power. Transportation was viewed as a derived demand— as something to minimize as the required travel activity is accomplished. More sophisticated models, such as utility maximizing models, attempt to measure the utility of items that are cited as attractive to homebuyers and the disutility of travel. Early work in this area was by Weisbrod et al. (24) and Lerman (25). Consistent with the various opinion surveys, Weisbrod (24) found that the consumer tends to place a lower value on transportation attributes than those of other aspects of life. The empirical results suggest that households make significant tradeoffs between transportation services and other public ser- vice factors in evaluating potential residences, but that the role of both in determining where people choose to live is small com- pared with socioeconomic and demographic factors. (p. 1) The authors note that about 20% of the nation’s popula- tion changes its place of residence every year, and 42% move 27

within a 5-year period; about half of these relocations are within the same metropolitan area. There is considerable consistency in the literature concerning the important factors affecting residential choice. Factors beyond the scope of public policy, such as the desire for single-family, detached homes among families with children, and the reduced moving rates for older persons and families with several children, all affect mobility and location patterns more than other factors related to public expenditures. (p. 9) As part of this research effort, Lerman (25) developed a dis- crete choice model of residential location that identified some of the factors that influence residential choice and the relative importance of transportation accessibility. That work found that although transportation accessibility is a factor that households consider in residential location decisions, socio- economic and demographic factors (including the match between a neighborhood’s demographics and the individual’s demographics) were more important than transportation accessibility in determining residential locations. A more recent study by Srour et al. (26) tested various accessibility measures for their effect on residential choice and property values. Findings were that access to jobs, retail employment, and park space were statistically and practically significant in both choice models and models of property val- ues. That work suggests that consumers are willing to pay for location. “The access may be to jobs, retail centers, parks, good schools, views, or other amenities; it is all capitalized into rent through market bidding” (p. 32). Work by Waddell and Nourzad (27) incorporated neigh- borhood accessibility measures in an integrated land use and transportation model. Findings were that regional access to employment was positively related to choice of a residential neighborhood. There was a preference for residential loca- tions that had more walking access to retail shops. This effect was stronger for those households where there was less than one automobile per worker. Other findings were that there was an overall preference for lower density locations, and this was more pronounced for households with children. Younger households favored higher density residential locations, and households with fewer cars were more likely to favor higher density locations than households with more cans. Higher income households with children were very unlikely to choose the most urban sites, whereas lower income and childless households, particularly those in which no vehicle was owned, were more likely to choose the most urban sites. Krizek and Waddell (28) point out that life-cycle stage appears to affect the decision about where to live and the importance of accessibility. Through a combination of factor analysis of a lifestyle attributes (including travel characteris- tics, activity frequency, automobile ownership, and urban form) followed by cluster analysis of respondents by their lifestyle factor scores, the authors defined nine distinct sub- groups. The subgroups are similar in their travel patterns and the urban form of their neighborhoods, and thus illustrate the pairing of longer term decisions on residential choice with short-term decisions on travel. Findings were that five out of nine lifestyle groups, or 60% of the sample, rated highly on the accessibility of their residential location. Two groups were those typically expected to gravitate to new urbanist commu- nities: retirees and transit users, which together made up 18.4% of the sample. Other groups with high accessibility also are associated with high rates of travel. These included the single busy urbanists (7.8%), who took vehicle trips with complex tours, and the family and activity-oriented partici- pants (12.3%), who took lots of nonwork trips. The largest group was urbanists with higher incomes (21.3%), who were average as far as activity and travel dimensions. This group would seem to be attractive for new urbanist communities in that they appear in high accessibility locations and do not take lots of trips with complex tours. Lessons from the Literature on the Relationship Between Land Use and Transportation The results of literature reviews on the effect of land use on travel behavior indicate that studies on this subject to date are not conclusive because of inherent methodological and or theoretical challenges. However, the weight of the evidence suggests the following: • A relationship exists between mode choice and land use, but socioeconomic variables may be of greater significance. • Just the opposite is true for trip lengths: land use is of primary significance, and socioeconomic variables are of secondary significance. • Trip frequency is almost completely a function of socio- economic variables. • The more mix of land uses, density of housing, and streets with intersections, the more residents walked. • Since residents in more urban communities may be self- selected as desiring a neighborhood where they can drive less and can walk and take transit more, observed comparisons may exaggerate the impact of urban design on mode choice. However, evidence is mixed on the extent of this effect. In the research on transportation’s influences on choice of neighborhood, the findings are also mixed. While a distinct majority of Americans still favor the rural ideal, or at least a home with a large lot, there is a seemingly growing group interested in a home that is in closer proximity to stores and commercial areas. There is evidence that access to jobs, retail employment, and parking does positively affect the value of homes. 28