Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 15, Land Use and Site Design (2003)

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.

15-1 15 — Land Use and Site Design OVERVIEW AND SUMMARY Transportation, acting through enhanced accessibility, is a long acknowledged influence in the shaping of cities and the determination of land development potential. The reverse, however, the impact of land use decisions on transportation outcomes, has only gradually achieved recognition. It is these reverse impacts — of interest in the treatment of land use and site design options as “transportation” strategies, a facet of “smart growth” — that provides the impetus for this chapter. Presented here is information on what is known or surmised about the relationships between land use/site design and travel behavior. Included within this “Overview and Summary” section are the following: • “Objectives of Land Use and Site Design Strategies,” summarizing key reasons why planners and decisionmakers view the land use-transportation connection as important. • “Types of Land Use and Site Design Strategies,” characterizing the types of strategies of concern to transportation analysts, and relating them to the elements of land use and site design around which this chapter is structured. • “Analytical Considerations,” identifying analytic approaches that have been used to examine the transportation-land use link, and offering guidance as to their reliability. • “Traveler Response Summary,” providing a condensation of key travel behavior findings. Before the reader opts to use any of the factors or relationships presented in this summary, it is recommended that the initial introductory sections be read as background on context and research caveats, and that relevant detail in the balance of the chapter be consulted. Following the four-part “Overview and Summary,” greater depth and detail are provided in the following sections: • “Response by Type of Strategy” provides information on what has been determined about the response of travelers to land use density, mix and site design. • “Underlying Traveler Response Factors” offers insights on aspects of travel demand important in understanding the link between land use/site design and travel. • “Related Information and Impacts” presents information on related areas of interest ranging from example residential densities to land use and site design effects on transit service feasibility, automotive travel trends, cost-effectiveness issues and environmental impacts. • “Case Studies” are presented to illustrate specific examples of the land use forms or strategies discussed in this chapter.

15-2 To facilitate expeditious use of this lengthy chapter, the user is first of all urged to take advantage of the “Use of The Handbook” suggestions offered in Chapter 1, “Introduction.” Second, the user should be aware that Chapter 15 has a second cut on the travel demand impact findings first presented at length in the “Response by Type of Strategy” section. This second cut is in the “Related Information and Impacts” section under “Trip Making and VMT,” and offers context — especially in the “Trip Making and VMT Differentials” subsection — that may be especially instructive as an overview. This material is suggested for up-front reading as a supplement to the “Traveler Response Summary.” Finally, as indicated below under “Types of Land Use and Site Design Strategies,” note that the “Response by Type of Strategy” section of this chapter is organized around the effects on travel demand of Density, Diversity (Land Use Mix), and Site Design, in that order. A research results introduction to the “Response by Type of Strategy” section may be obtained by reading through the research findings summary tables provided within each of the three subtopics. These tables are: • Under “Density” — Tables 15-3 (density as prime indicator), 15-7 (density along with other indicators), 15-9 (density guidelines for transit service) and 15-10 (density and transit use). • Under “Diversity (Land Use Mix)” — Tables 15-14 (diversity: jobs/housing balance), 15-16 (diversity: accessibility, entropy, etc.) and 15-22 (land use mix and transit use). • Under “Site Design” — Tables 15-23 (site design of suburban activity centers), 15-30 (design of transportation networks), 15-32 (design of neighborhoods) and 15-41 (transit supportive design). Each of these tables directs the reader to where within the “Response by Type of Strategy” section, or elsewhere in the chapter, more detail is provided. Available detail should definitely be consulted before applying any of the summarized findings, especially to gain appreciation of pertinent research limitations. The subject matter of Chapter 15, “Land Use and Site Design,” though largely self-contained, does overlap with Chapter 16, “Pedestrian and Bicycle Facilities,” in the area of site design impacts on non-motorized travel. In addition, Chapter 17, “Transit Oriented Development,” goes further in examining that particular application of land use and site design. Chapter 13, “Parking Pricing and Fees,” and Chapter 18, “Parking Management and Supply,“ may also be of special interest in view of the close linkage between land use density and parking supply/pricing. Impacts of transportation actions on development patterns are addressed in the “Related Information and Impacts” sections of the chapters on relevant transportation facilities, such as in Chapter 4, “Busways, BRT, and Express Bus”; Chapter 8, “Commuter Rail”; and Chapter 9, “Light Rail Transit.” Objectives of Land Use and Site Design Strategies The architecture of our land use patterns and streetscapes reflects a melding of numerous economic, social and other influences, of which transportation is but one. Similarly, the non- transportation objectives of seeking particular land use and site design approaches are many, including providing desirable and affordable housing, enhancing quality of life, supporting economic development, preserving agricultural and environmentally sensitive lands, and

15-3 minimizing dollar costs. Cost items include new infrastructure which, in addition to transportation facilities, includes sewers, water and schools. Costs also include the social, economic and lost resource costs of losing productive open space, and of having previously viable urban neighborhoods left behind in outward growth. Not to be overlooked, however, are worthy transportation objectives for shaping land use patterns and site design features in the interests of transportation efficacy and impact mitigation. These objectives include: • Reductions in vehicle miles of travel (VMT), pollutant emissions, and energy consumption. Concentrated, contiguous development and balanced land use provide opportunity for households to meet daily needs with shorter automobile trips or by walking, bicycling, or taking transit, thus contributing to reduction in overall VMT and efforts to manage congestion, reduce energy vulnerability, and achieve air quality health standards. • Increased transit use and productivity. Clustering and intensification of residential and commercial development along transit lines and around transit facilities increases the number of opportunities that can conveniently be reached by transit, which in turn leads to higher levels of ridership, correspondingly increased service productivity and cost effectiveness, and potential for even higher transit service levels. • Pedestrianization of activity centers. Concentrated, mixed land uses coupled with pedestrian friendly site design not only facilitate non-motorized and other non-auto- driver travel by residents, but also by commuters and commercial visitors. Knowledge that most activities within a center can be reached on foot or via local transit once there diminishes perceived need to drive to a center, enhancing choice of transit and carpooling. Types of Land Use and Site Design Strategies There are a number of specific actions that governments or planning agencies have taken to try to manage or influence land use or site design in relation to transportation or other public policy concerns. Examples include: • Growth Boundaries or Regulatory Controls: A number of states and metropolitan areas have enacted legislation or imposed regulatory controls on growth in the interests of curbing sprawl and associated deleterious effects. The state of Oregon has metro area Urban Growth Boundaries to constrain urban growth within established limits. Portland, Oregon, established its Urban Growth Boundary in 1980. Minneapolis/St. Paul has a similar boundary, largely to protect commercial agricultural lands and northern lakes wilderness areas, and to control regional water and sewer requirements. More recently, states like Maryland, New Jersey, and Massachusetts have enacted “Smart Growth” types of laws with comparable objectives. • Planning and Zoning: Planning and zoning are among the oldest tools used to guide growth at the local level. An area’s comprehensive land use plan and zoning designates the location, mix, and intensity of uses that are desired for development in the community. At a macro scale, Master Plans may be developed for cities, counties, or regions to establish intended uses in terms of intensities, location and supporting transportation facilities. Sometimes addressed in these plans is the jobs-housing ratio, a measure of the balance among land uses, particularly in relation to work travel. A major

15-4 planning consideration is highway, street, and pedestrian facility layout, typically enforced at the local level through design standards and land subdivision controls. • Growth Moratoriums or Traffic Ordinances: Some jurisdictions have adopted ordinances that regulate the pace of new development to ensure adequate capacity and performance of existing and new public facilities. Some limit development at a site if its addition would increase traffic congestion beyond a specified threshold. “Adequate Public Facility” and similar “Concurrency” ordinances fall in this category. They must be carefully structured to avoid inadvertent discouragement of desired construction such as higher-density, compact development supportive of transportation alternatives. • Building Codes and Site-Level Zoning Requirements: At a site level, building codes and site-level requirements of zoning may have provisions that can have important effects on transportation options and travel behavior. Some areas, like Bellevue, Washington, and Montgomery County, Maryland, limit or seek to discourage on-site parking by placing maximums on spaces per 1,000 square feet or offering density incentives for building less parking (see Chapter 18, “Parking Management and Supply”). Other tactics include reduced building setbacks to improve access for walk, bike and transit users, and suburban office park requirements for supply of a mix of pedestrian-accessible services on site, to reduce need for auto commuting. • Transit Oriented Development (TOD): Development earns the TOD designation when growth is focused or intensified in the immediate vicinity of a transit route, station or other service node. Along with the higher densities, TODs need pedestrian and transit friendly design. San Diego, the San Francisco Bay Area, Portland, Oregon, and Arlington, Virginia, are examples of areas that have actively pursued this type of land use arrangement. • Traditional Neighborhood and Pedestrian Friendly Development: A movement has emerged to build new or redeveloped areas which look and behave more like traditional towns. Structuring an activity center or community so that it has key traditional town characteristics of mixed uses, walkable distances, sidewalks, and other design features conducive to walking, biking or transit use is often termed Traditional Neighborhood Development (TND) or, less frequently, Traditional Neighborhood Design. If such developments reflect past practices in extensive detail, such as accompaniment by streets and alleys laid out in a full conventional grid pattern, they may be classified as Neo- Traditional Development. Conversely, Pedestrian or Transit Friendly Development may provide close-at-hand retail and services, walkability, easy bicycling and transit supportive infrastructure without Neo-Traditional design constraints. Unlike TOD, neither TND nor Pedestrian Friendly Development necessarily requires high densities. • Infill and Brownfields Development: Efforts to strengthen central places, make better use of existing infrastructure, and reuse semi-abandoned urban lands, all in preference to equivalent outward expansion, have led to use of infill and brownfields development. Infill refers to building on vacant parcels within otherwise developed urban landscapes, while brownfields development pertains to redeveloping sometimes large urban tracts often saddled with industrial contaminants that must be remediated. Infill and especially brownfields development often require incentives and other seed money. • Incentives and Fees: Pricing mechanisms may be applied to alter existing conditions in the market place that act as development signals. These may directly or indirectly affect

15-5 land use or transportation. Governments are experimenting with location efficient mortgages or job creation incentives to attract development to desired locations. Government investment in infrastructure or programs can also entice development into particular areas. While the actions or strategies listed above are typical of the approaches that can or have been taken to influence land use, a lack of pertinent studies greatly limits the ability to relate traveler responses directly to these particular actions. The direction of land use research has been more along the lines of trying to establish whether a link exists between a particular type of land use pattern and resulting travel. In other words, contemporary research may be described as “investigative” — trying to prove that there is or is not a land use/travel link, the determinants of the link, and the strength of the link — rather than tracking results of particular strategies. The predominant land use characteristics usually studied by researchers are the “3 Ds” — Density, Diversity, and Design (Cervero and Kockelman, 1997; NTI, 2000), as follows: • Density, which relates to concentration or compactness of development, measured by the number of opportunities (activities, jobs, places to live, or combinations) located within a given geographic space. • Diversity or “Land Use Mix,” which relates to the extent and nature of the mix of uses, and the balance, or compatibility, of the uses with each other. • Design, which refers to the way in which the various uses are combined, linked and presented on a site, and the results in terms of ease of access, use, and attractiveness. As a partial guide to how the eight actions or land use strategies listed at the outset may be aligned with these three more elemental characteristics of land use, Table 15-1 provides a cross-referencing between the various strategies and the aspects of land use they are most likely to influence. Fullness of the circles indicates strength of the connection: Table 15-1 Land Use Strategies and the Characteristics They May Influence Land Use Characteristic Land Use Strategies Density Diversity Design Growth Boundaries/Regulatory Controls
Planning and Zoning


Growth Moratoriums/Traffic Ordinances

Building Codes/Site Level Zoning
Transit Oriented Development


Traditional/Pedestrian Friendly Design
Infill and Brownfields Development


Incentives and Fees

15-6 This chapter is organized around the three land use characteristics of Density, Diversity, and Design. One of the land use strategies is addressed directly, however, in its own chapter. That is “Transit Oriented Development,” the previously mentioned Chapter 17. Analytical Considerations While this chapter draws from a broad range of research studies 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 measures reported elsewhere in this Handbook. Studies of the connection between land use-travel and behavior may be grouped according to three basic analytic approaches (Handy, 1997): • Simulation Studies, which use traditional “4-step” transportation models to compare alternatives. These studies do not provide behavioral relationships between land use and travel patterns, but utilize established relationships to test the potential of different land use arrangements and street network designs to alter travel parameters and volumes. • Aggregate Studies, which look at differences in travel patterns between different forms of communities. These aggregate comparisons of different types of neighborhoods normally focus on differences in average travel characteristics among a typically small number of neighborhoods. • Disaggregate Studies, which utilize cross-sectional, micro-level, observed data to explore how differences in urban form influence individual travel choices. These studies delve into the underlying mechanisms that explain why people make particular choices about travel, and specifically how land use or urban design influences those choices. Most research findings presented in this chapter are from aggregate or disaggregate studies attempting to interpret what role land use appears to have in travel behavior based on cross- sectional comparison of different areas exhibiting different land use patterns, at everything from region to community to traffic analysis zone level of detail. In some such studies, attention to non-land use variables of dominant importance — such as family size and household income — is weak or lacking, leaving substantial doubt as to what really caused the observed differences. In the more trustworthy studies, various techniques are applied to control for confounding variables, such as stratification by socio-demographics, or inclusion of non-land use variables within the specification of explanatory models. 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. A related concern is that practically none of the studies in the land use and site design arena have been advanced to the point of fully employing detailed transportation system and travel characteristics descriptions equivalent to the use made by regional transportation network models. There are some exceptions, road network simulations and certain calculations of accessibility among them, but current land use and transportation research model development typically lacks the benefit of travel parameter estimates specific to origin-

15-7 destination pairings. This adds concern that some of what is attributed to land use differences or even study population attitudes may in part actually relate to unidentified transportation service differences, or conversely, that some land use effects may be lost in too- generalized descriptions of the transportation environment. Additional concerns include the following issues (Deakin, 2002): • The reliance of aggregate studies on a small number of cases, effectively placing reliance on a very small sample. • The existence of problems of scale and aggregation level, to the point where averages mask variations in characteristics. • Other measurement problems, including use of rather gross and clumsy indicators in lieu of in-depth measures. • Wide differences in definitions of key variables, making comparison and transferability of findings across studies — and between research and application — quite tenuous. • Confusion of correlation and causality in interpretation of results, overlooking the fact that just because two or more parameters move in parallel, the one is not necessarily causing the other(s). It is well to remember that correlation analysis, while a relatively sophisticated technique for identifying relationships, neither proves causality nor identifies direction of causality. Indeed, a crucial question regarding any hypothesized relationship between land use and travel behavior is the nature and direction of causality. Is it the built environment itself that is precipitating the travel behavior? Is it external factors that are causing or influencing the behavior? Might the built environment be drawing particular types of individuals who bring with them travel and lifestyle needs and preferences that are resulting in the observed behavior? For example, while available studies usually show auto use to be lower in “traditional” neighborhoods, they also raise questions as to “why” behavior is different. Some aspects of this issue are addressed further in the “Underlying Traveler Response Factors” section under “Attitudes and Predispositions.” Few if any before-and-after type studies of impacts of land use on travel behavior have been performed. This deficiency may be partly due to the lack of foresight to set up a measurement framework and take “before” surveys, but more likely, it is because of the long time frames associated with emergence of impacts from land use changes. Moreover, given the long time spans involved, it can be presumed that many factors are at work in influencing the subsequent changes: market trends, prices, demographic shifts, income changes, technology, and even shifts in basic tastes and preferences. Thus, researchers are left with only inferential statistical analysis options, aggregate or cross-sectional, rather than closely- monitored cause-and-effect studies. A major issue in research on the land use-transportation connection has been the confounding role played by density. Most early land use studies relied strongly on density as the chief measure of land use and urban form, and while they found significant correlation between density and travel, they also discovered that density alone was not sufficient to explain all of the variation in observed travel behavior. Moreover, because higher density is a close proxy for other characteristics of urban form, like centrality of location, greater mix of

15-8 uses, better pedestrian friendliness, higher levels of transit service, higher accessibility to activity opportunities, lower availability of parking, and various household socioeconomic differences, it often masks the effects of these other characteristics when included in models. A commonly applicable criticism of earlier studies (and certain newer ones) is the lack of attention to the association of generally lower incomes with historically high density areas and also with a preponderance of older traditional neighborhoods. Similar concerns revolve around associations between density and household size and composition. Increasingly, land use research has come to appreciate these dilemmas. Considerable effort and creativity has been devoted to specifying more meaningful measures of urban form, and to research designs that better control for intervening factors, such as socio-demographics and transportation supply. Commentary within this chapter attempts to alert the reader as to where in this continuum of research advancement a particular study seems to fall. In interpreting findings relative to density, care must be taken to ensure that the units of measure being employed are understood. A brief discussion of density units of measure, with selected conversion factors, is provided under “Examples of Residential Densities,” the first subsection of the “Related Information and Impacts” section. Additional evaluation and measurement issues to be alert to in any synthesis of findings are covered in Chapter 1, “Introduction,” under “Use of The Handbook” — “Degree of Confidence Issues,” “Impact Assessment Considerations,” “Demographic Considerations,” and “Concept of Elasticity.” Traveler Response Summary This “Traveler Response Summary” is organized by type of behavioral impact. Summaries organized by land use or site design factor, and identifying individual research efforts, are found within Tables 15-3, 15-7, 15-9, 15-10, 15-14, 15-16, 15-22, 15-23, 15-30, 15-32, and 15-41, as identified in more detail at the outset of this “Overview and Summary” section. The research findings suggest that land use and urban form exert an important cumulative influence over most aspects of travel demand. An exception is frequency of person trips by all modes combined, where any effect there may be is not well understood and normally quite small. Where development is compact, land uses are compatible and intermingled, and there is good transit access and pedestrian interconnection, it appears that average trip lengths are shorter, greater use is made of transit and non-motorized travel modes, and household vehicle trip generation and particularly household VMT are less. These effects are further enhanced by centrality of urban location. The huge sunk investment in urban development and infrastructure dictates, however, that travel behavior shifts in response to changes in land and site development approaches will necessarily be localized near term, extremely gradual overall, and more oriented to dampening of adverse trends than reversals. Auto ownership is both a household characteristic and a behavioral response to available income and one’s living and transportation environment. Lower auto ownership, once established, is generally found to be associated with reduced likelihood of choosing to drive, and fewer VMT. Nationwide, auto ownership declines from almost 1.2 vehicles per adult at 0 to 99 persons per square mile densities, to 0.7 vehicles at 50,000 persons or more per square mile. Although individual researchers demur, most have found that a small causal relationship between higher densities and lower auto ownership evidences itself after controlling for confounding income effects. This relationship is presumably connected to a

15-9 greater ability to meet daily needs without an auto at higher densities, paired with higher costs and inconvenience of garaging and using an auto. Most researchers have not isolated effects of land use mix or site design on auto ownership. However, analysis of detailed San Francisco Bay Area data obtained a -0.07 elasticity for auto ownership relative to density, another -0.07 with respect to accessibility, -0.03 relative to local land use balance, and -0.01 relative to dissimilarity, a measure of fine-grained mix.1 For estimates of elasticities of VMT to density in the range of -0.04 (San Francisco data) to -0.07 (national data), calculations are that all to two-thirds of the effect is channeled through vehicle ownership, respectively.2 Portland, Oregon, studies found good pedestrian environment to be related to lower auto ownership. Person trip generation encompasses the making of trips by all modes, including walking. On this basis, national data shows only minor variation: 15 percent difference in household rates between the very highest and lowest densities. There is a preponderance of agreement that there is no causal relationship between population or employment densities and person trip making. However, at least four studies have identified somewhat elevated person trip generation where population density, employment density and mixing of land uses are greater. This trip rate elevation seems to typically take the form of additional walk trips and/or trip chaining, and is generally coupled with VMT rates that are lower. Trip length is the outcome of location of opportunities to meet travel needs. One San Francisco Bay Area analysis found it took a suburbanite 4 to 8 times as many vehicle miles to accomplish as much as a resident of denser urban areas did with a mile of transit travel. Assessments of home-to-work travel in U.S. West Coast cities suggest good jobs-housing balance may be associated with 7 to almost 30 percent shorter commute trips, but these studies may have allowed balance to act as a proxy for other characteristics, overstating the effect of density per se. Washington State studies indicate that normally 20 percent or more of residents in cities and other Census-designated places with good jobs-housing balance live in the same city, although some San Francisco Bay Area cities with theoretically good balance exhibit percentages only in the teens of employees living in town. Addition of downtown housing is believed to have stabilized congestion in the face of doubling office space in downtown Toronto, by reducing associated trip lengths to walkable distances or otherwise short commutes. Seattle area mixed use communities, as compared to surrounding areas, have almost 4 times as many opportunity sites for meeting trip needs within one mile or less of home, and 2 times as many within 2 miles. The travel distance differentials for these communities are summarized under “Vehicle Miles of Travel.” Paired community analysis in the San Francisco East Bay found only 3 percent more non-work trips under 2 miles in length in the pedestrian oriented neighborhood, a streetcar-era TND community, compared to the auto- 1 An auto ownership elasticity of -0.07 relative to density indicates a 0.07 percent decrease (increase) in auto ownership in response to each 1 percent increase (decrease) in density, calculated in infinitesimally small increments. The negative sign indicates that the effect operates in the opposite direction from the cause. An elastic value is -1.0 or +1.0, or beyond, and indicates a demand response that is more than proportionate to the change in the impetus. (See “Concept of Elasticity” in Chapter 1, “Introduction,” and Appendix A, “Elasticity Discussion and Formulae.”) 2 Where density type is unspecified in this “Traveler Response Summary,” it generally refers to population density at the home end of a trip. However, the “Response by Type of Strategy” presentations that follow should be referred to for the fullest available detail.

15-10 oriented area of conventional suburban design (CSD). However, average non-work trip lengths were 6.8 miles for TND community residents compared to 11.2 miles for CSD area residents, a difference at least partially attributable to neighborhood design and with substantial implications for VMT generation. Transit mode choice and ridership are highly related to density if one includes the second- order effect of the better transit service higher density typically brings into play. The effect of density in contributing to sheer volume of riders is illustrated by Arlington, Virginia’s focusing of dense development on Metro stations. There key-station ridership, the second decade after opening, rose 121 to 164 percent in 9 years. Elasticities approaching +0.4 (Chicago) to +0.6 (multi-system) have been estimated for heavy and light rail station volumes relative to residential density, with lesser effects for commuter rail. Multi-system elasticities for light and commuter rail station volumes relative to central business district (CBD) employment density are +0.4 and +0.7, respectively. These elasticity examples assign to density, rather than to the full array of urban form and transportation characteristics, all effects for which density serves as a marker. In terms of mode share, apportioning effect among the full array of urban form elements, 10 persons or employees more per acre at origin and destination has been estimated to equate (in greater Seattle) to at least a 1 to 2 percent higher transit share. A good jobs-housing balance has been estimated to be linked (in Florida) to a transit mode share 2 percentage points over that for a single use area.3 Research to date on land use mix and site design appears as likely to find no effect on transit choice as to find positive effect, but some studies have estimated substantial impact. Suburban activity center (SAC) transit share elasticity to mix has been estimated at nearly +0.3. Transit commute shares at worksites with Travel Demand Management (TDM) were found to be half again to over twice as high with land use mix. Studies in California suburbs suggest community design effects on choice of transit mode may be small on average, but in no case was TND design found other than neutral or positive. Urban environment measures from “aesthetics” to pedestrian friendliness to urban vitality have been found linked with elevated mode shares for transit. Non-motorized travel (NMT) choice, primarily walking and biking, reaches 7 percent of daily trips nationwide at population densities of 2,000 to 5,000 persons per square mile, climbing to 46 percent at over 50,000 persons. These differentials are, however, in response to density plus all the urban characteristics usually accompanying it, including greater land use mixing, shorter distances between attractions, and more sidewalks. Evaluation of NMT choice relative to “pure” density found no significant linkage on the basis of detailed San Francisco Bay Area data, while estimated walk/bike elasticities were a shade above +0.2 for accessibility and also for land use balance. On the other hand, Seattle area evaluations found more significant effects on NMT choice for density than for land use mix. One San Francisco Bay Area research effort concluded that attitudes were more important to NMT and other travel choices than either household or urban form characteristics. Research on the role of land use mix has generally but not always found mix and design to have their strongest influence on walking and bicycling. Study of suburban employment centers (SECs) identified mixed use as having a small but positive effect on incidence of walking trips. Surveys within Houston’s SECs showed 20 percent of all trips to be made on foot, despite impediments, with a heavy concentration at midday. Income stratified comparisons found walk shares in Southern California TND neighborhoods ranging from 3 “Percentage points” refers to an absolute difference in percentages, rather than a relative difference.

15-11 17 percent less to 53 percent more than in late 20th Century planned unit developments (PUDs). Walk mode shares were double in mixed use locales of Seattle and environs relative to comparison areas. San Francisco East Bay paired community analysis showed non-work trips under 2 miles in length to have a 52 percent NMT share in the TND environment versus 17 percent for CSD. Walk shares for trips of all lengths were 7 versus 1 percent for non-work (TND against CSD), and 10 versus 2 percent for work travel. Good pedestrian environment has been found positively related to higher NMT shares in Portland, Oregon, in San Francisco proper, and at worksites having TDM programs with financial incentives. Choice of the auto mode versus alternatives is such that nationwide, the proportion of trips by auto is close to 88 to 90 percent of all travel for densities below 5,000 persons per square mile. The auto share declines as density increases, to as low as 22 percent at over 50,000 persons per square mile, with taxi trips included. Part of density’s dampening effect is channeled through higher parking costs and congestion. Having 10 persons more per acre at both origin and destination has been estimated in Seattle area studies to be linked with auto shares lower by 7 percent, or 2 to 3 percent in the case of employees per acre. The same studies found land use mix less negatively correlated with single occupant vehicle (SOV) use than density, and less positively correlated with bus use and walking than density, but to be the only urban form variable significantly correlated (positively) with carpool use. Research models based on comprehensive San Francisco Bay Area survey data showed auto choice to be primarily sensitive to socio-economic variables, including auto ownership (elasticity of +0.12), but also population density (-0.01) and accessibility (-0.03). Differentiating by trip purpose, and working with 50 contrasting San Francisco area neighborhoods, transit, bike and walk (non-auto) travel for work trips exhibited no sensitivity to greater density, elasticities of +0.05 to +0.34 for more diversity, and +0.03 to +0.12 for better design. The corresponding non-work, non-auto elasticities were: +0.06 to +0.11 for density, +0.11 to +0.14 for diversity, and +0.08 to +0.18 for design. Suburban activity centers nationwide with some on-site housing had 3 to 5 percent more non-auto commute trips, and every 10 percent more commercial and retail space was apparently related to 3 percentage points higher non-auto share. Centers with greater mix had lesser midday auto shares to major retail. U.S. national household data exhibits a 2 to 5 percentage point dampening of auto choice in the presence of retail. Efforts to develop quantitative links between individual site design characteristics and travel demand are in their infancy. Further evaluation of the San Francisco East Bay paired communities failed to isolate most individual effects, but identified about a 10 percentage point higher non-work, non-auto share composite effect. Study of individual office buildings in edge cities found scale and mix of uses to be positively related to carpooling, with 0.8 more passengers per work trip in million square foot office buildings as compared to buildings half that size. Parking supplies less by half were associated with 0.5 more auto passengers per work trip. Choice of mode for accessing transit, like choice for short distance local area trips, is very sensitive to urban form. Walking normally predominates for only up to 1/2 to 3/4 of a mile for access to rail service. Higher densities place more riders within the walking radius. Population density higher by 1 percent is associated with 1 to 2 percentage points higher choice of walking to rail transit in Chicago and to Bay Area Rapid Transit (BART) in the San Francisco area, and about 2 percentage points lower auto use for the more suburban systems. Also, bus use for rapid transit access is lower, by about 1 percentage point, as walking

15-12 increases. These relationships would translate into elasticities near or well into the “elastic” range. Higher residential area employment density, actually a measure of land use mix, was also shown to enhance walking to the urban systems — by the same order of magnitude. Elasticities for access/egress to BART stations relative to an index of mix were +1.1 for walk and -1.3 for auto. San Francisco East Bay paired communities analysis showed the TND neighborhood to engender a 31 percent walk share for transit station access, compared to 13 percent for the station with a CSD environment (and a substantial commuter parking lot). Vehicle trip generation, the auto driver share of person trip generation, follows a pattern much the same as auto share, but with carpooling effects imposed. Estimates of household vehicle trip generation elasticity to population density that distinguish density from other elements of urban form are in the “no significant effect” to -0.08 elasticity range. Analysis of San Francisco Bay Area data produced a density elasticity of -0.014 and an accessibility elasticity of -0.034. Meta-analysis of a number of studies obtained an average elasticity for vehicle trips relative to local population density of -0.05, along with -0.03 for local diversity (mix), and -0.05 for local design. Corresponding sensitivity to regional accessibility has been identified as either nil or -0.05 depending on selection and interpretation of the data. Suburban activity center research found employee vehicle trip rates to exhibit an elasticity of -0.06 to land use mix. One study has estimated vehicle trip generation reduction in the range of 1 to 3 percent for improved pedestrian access to large-scale regional shopping centers. Office employee vehicle trip rates 6 to 8 percent lower have been reported for edge city office buildings with retail (versus office buildings with none). Rates on the order of 10 to 15 percent less have been found for worksites with TDM programs and good availability of on-site services (versus other worksites with TDM programs), Portland areas with fairly good pedestrian environment (as compared to pedestrian-hostile), Southern California TND communities (versus conventional PUDs — with reductions up to 23 percent),4 and a TND community in the San Francisco East Bay (relative to its paired CSD area). Living in the East Bay TND community had about the same effect on lowering auto use as having one less auto in the household. Vehicle Miles of Travel is almost universally agreed to exhibit an inverse relationship with density. There are two schools of thought, however, on the strength of that relationship and its transferability to new development. Studies that allow density to stand as a surrogate for commonly associated urban form and household characteristics, and the second-order effects historically accompanying density, have estimated “double the density” to be associated with 15 to 30 percent less VMT per household. To actually approach such a reduction with “new” density, the metropolitan area centrality, transportation alternatives, and demographics historically linked with higher density would need to be present. Typical research elasticities for VMT as a function of more narrowly defined density are fairly modest: in the -0.05 to -0.1 range. The previously noted meta-analysis provides an elasticity of VMT to local density of -0.05, along with -0.20 for regional accessibility, -0.5 for local diversity (mix), and -0.3 for local design. These four elasticities are deemed to be additive in application, thus the cumulative 4 The Southern California reduction of 10 to 23 percent is for both auto drivers and auto passengers, for individual TNDs. Other multiple-site observations in the 10 to 15 percent reduction range are placed there on the basis of study averages for vehicle trip reductions. All observations are per unit of activity measure (person, employee, household, etc.).