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.

15-90 The modelers conclude that the significance in many models of neighborhood vitality, readily evident in Table 15-44, indicates a strong relationship between non-auto mode usage and urban form in San Francisco. Notably, not-good vitality was negatively related to choice of the mode in walk, bicycle, and transit mode models. Adverse topology, often encountered, was also significant as a negative in a number of the models (Cambridge Systematics et al., 2002). The lesser significance overall of pedestrian network connectivity and ease of street crossing may not be a broadly transferable finding, as San Francisco proper has relatively little in the way of suburban style street networks or broad intersections to offer contrast. Urban Design Overall. A recent synthesis addressing “Travel and the Built Environment” characterizes urban design impacts on travel as “the newest frontier in travel research” with “few studies to draw on.” The authors hypothesize that the largest impact may be on trips within an activity center; secondary trips that may not even be recorded in many survey returns dependent on respondent recall. Based on their meta-analysis of the available quantitative research, the authors offer a Vehicle Trip elasticity relative to local design, represented by sidewalk completeness, route directness and street network density, of -0.05. Their corresponding elasticity estimate for VMT is -0.03. The authors highlight automobile parking as a particularly important design feature largely neglected in travel studies and research. The expanses of parking found in many cities and most suburbs displace active land uses, create dead spaces, make adjacent sidewalks less attractive by diminishing human interaction, and create access problems between sidewalks and buildings (Ewing and Cervero, 2001). (For additional information from this study see “Related Information and Impacts” — “Trip Making and VMT” — “Consolidated Vehicle Trip and VMT Elasticities.”) UNDERLYING TRAVELER RESPONSE FACTORS While there are clearly strong associations between land use, urban form, and travel behavior, questions abound as to the precise nature or direction of the relationship: Is it compact land use and transit/pedestrian friendliness that produces the measured differences in travel behavior, or are there intervening and other confounding factors that lead to these eventual differences? The following are key issues that underlie the complexity of land use/travel behavior causality questions. Density Versus Accessibility Density has been used, or misused, as a key measure of alternative land use structures — such as compact development versus urban sprawl — perhaps longer than any other descriptor. The “Related Information and Impacts” section provides an encapsulated history of the evolution of density assessment. Look there under “Trip Making and VMT” — “Trip Making and VMT Differentials” — “Vehicle Miles of Travel.” Typical density measures include persons or employees per square mile, dwelling units per acre, and so forth. Much existing high density development is characterized by centrality of place, mixed land uses, above average transit service, higher parking costs, diminished auto driving convenience, smaller household sizes, and historically, lower incomes. The close tie between dense development and these characteristics, all of which are generally associated with reduced auto use in trip making, introduces substantial potential for miscalculating the actual role of density itself in shaping urban trip making.

15-91 Density, mixed land uses, centrality of place and being at a focal point of transportation services all increase the “opportunity” of taking care of one’s daily activities with a minimum expenditure of time and resources for travel. Increased study of relationships between land use and travel suggests that it is this “opportunity” provided by having everything closer together — along with the associated transportation infrastructure that makes interaction among “opportunity sites” faster — that most influences travel behavior, rather than density per se. This opportunity is labeled accessibility: the “ease with which activities can be reached from any location” (NTI, 2000). Accessibility as commonly defined consists of an activity element providing the motivation or needs satisfaction of being in or reaching a place, and a transportation element providing the means of getting there. The one part is measured with an activity descriptor, such as one or another type of employment, and the other by friction, which is best when small and is typically measured in units of time, cost, distance, or combinations thereof (Handy, 1993). Accessibility may be quantified for analysis with simple measures such as the amount of activity within a 1/3-mile walk or a 30 minute drive, or with more complex measures based on travel demand models. As intimated above, accessibility is not only enhanced by compacting land use, but also by placing complementary land uses near to each other. While in theory it might be possible to quantify this benefit of land use mix entirely with measures of accessibility, separate measures of activity diversity have proved useful. One form of quantitative land use mix measure employed by several researchers is an Entropy Index (Cervero, 1988; Frank, 1994; Messenger and Ewing, 1996; Sun, Wilmot and Kasturi, 1998). A comprehensive sorting out of density, accessibility, and land use mix was accomplished by Kockelman in the course of University of California thesis development. This research into the significance of urban form to travel behavior specifically questioned the validity of past findings that relied on simplistic urban form measures and analytic methods. In particular, the work took issue with the way density has been used, given its confounding role as a proxy for other key variables. As an alternative to primary reliance on density, Kockelman’s framework employed discrete measures of three key dimensions of urban form deemed likely to have a causal relationship with travel behavior (Kockelman, 1996): • Intensity, represented by various Accessibility Indices, describing number of activity opportunities available weighted by the ease of getting to them (inverse of friction — favorable travel time). • Land Use Balance, represented by an Entropy Index, reflecting the variability of and relative balance among local area land uses. • Land Use Integration, represented by a Dissimilarity Index, reflecting the degree of fine grained mix in virtually adjacent land uses (mix at a pedestrian scale). Expressions showing the generalized formulation for each of these urban form indices are portrayed in Figure 15-7. Also displayed in Figure 15-7 is a diagram depicting the geography of the Dissimilarity Index calculation.

15-92 Figure 15-7 Formulae for Kockelman Urban Form Variables Where K = Number of actively developed hectares in tract, and Xik = 1 if central active hectare’s use type differs from that of neighboring hectare (Xik = 0 otherwise) C R R C R R Example, where: C = Commercial I = Industrial R = Residential Middle hectare has a Dissimilarity Index value of 4/8, since 4 of the 8 adjacent zones have a different land use. I I R Notes: Updated per Kockelman (1997). In these formulations of the indices: • Accessibility resembles the format of the gravity model, the index value increasing in proportion to the number of opportunities (numerator), and decreasing in proportion to the travel time (or cost) of accessing the opportunities (denominator). In application a 30 minute maximum was imposed for certain models. • Entropy, representing land use balance, is formed by describing the land use makeup of each tract in terms of six primary land use categories: residential, commercial, public, office/research, industrial, and parks/entertainment. The value of the measure ranges from 0 to 1, with 1 = “perfect balance of uses.” In application, a “mean entropy” was constructed to take into account the neighborhood, avoiding bias against smaller tracts. • Dissimilarity, which represents fine grained land use mix, is calculated as the ratio of the number of adjoining hectares with different use to the total number of adjacent hectares being compared. (See the example following the formula). The average of the point accumulations across all active hectares in the tract is the mix index for the tract. Source: Kockelman (1996 and 1997). Accessibilityi =
j Aj / f(tij), Where Aj = attractiveness of zone j and tij = travel time zone i to j Entropy = –
j (Pj x ln(Pj)) / ln(J), Where Pj = proportion of land in the jth use type Dissimilarity Index = Mix Index = Σ
k 1/K Σ
i8 Xik / 8, Σ Σ

15-93 The urban form measures were tested for significance and causality in regression models of total household VMT, non-work home-based VMT, auto ownership, personal vehicle choice, and walk/bike choice. Demographic variables tested included household size, per-member income, and auto ownership, which itself was estimated making use of the three urban form indices. It is important to note that both the income and auto ownership variables are household totals divided by the number of household members 5 years of age and older. Land use variables tested in addition to the three indices included jobs density and population density, which contributed to the explanatory power of individual models along with the indices. Model results supported the use of more precise urban form measures and the hypothesis that land use integration and compact development can reduce auto reliance and use (Kockelman, 1996). The numerical results, and discussion of them, were presented in the “Response by Type of Strategy” section of this chapter within two separate subsections: “Density” — “Density and Other Indicators at the Behavioral Level” — “Density, Accessibility, and Community Type,” and also “Diversity (Land Use Mix)” — “Accessibility, Entropy, and Other Measures,” including Table 15-18. Although Kockelman’s research specifically addressed that second order effect of density which is expressed through auto ownership, it did not become involved in other related issues such as the chicken or the egg question of transit ridership: Is transit ridership inherently “there” because of density and associated travel demand, causing transit officials to respond with appropriate service levels, or is transit ridership an outcome of specific levels of service, such that transit service level effects should be factored out when examining the role of density? Although that question is not resolved here in Chapter 15 either, the following subsections touch upon related issues along with revisiting the auto ownership effect. Spatial Separation Spatial separation is a key factor in travel choices and patterns, playing a direct role in measures such as accessibility, and an implied role in measures of mix. Distance is, of course, a major determinant of travel time. When an opportunity to satisfy a practical travel need is closer at hand, it will be more attractive than one farther away. This is reflected in traditional travel demand estimation techniques, most notably the “gravity model” of trip distribution. If a destination is close enough, walking or bicycling in lieu of motorized travel becomes a viable option. The sensitivity of would-be pedestrians to even small differences in walking distance is covered in Chapter 16, “Pedestrian and Bicycle Facilities.” Even if driving is the selected travel mode, the closer a destination is the less VMT will be produced getting there. If use of transit is an option, spatial separation from the transit stop will markedly affect both whether the option is selected and the decision of whether to walk or drive to the transit service. Empirical evidence of one or the other of these phenomena is presented in Chapter 3, “Park and Ride/Pool;” Chapter 7, “Light Rail Transit;” and in Chapter 10, “Bus Routing and Coverage” under “Underlying Traveler Response Factors” — “Transit Accessibility.” To fully understand the underlying effects of land use density, diversity, and design on travel demand, their effects on spatial separation must be dissected. Density reduces spatial separation by simply putting everything closer together. Increasing density is, among other things, a brute force way to increase accessibility. Accessibility is, after all, a measure of separation from where one is likely to want to go, with good accessibility reflecting closeness.

15-94 The “Transportation Service Levels” discussion to follow highlights how density increases the number of people available to use a transit service (the transit market) and may thereby synergistically lead to better service. The implications of spatial separation for transit use and choice of transit access mode suggest that non-uniform densities may be manipulated to the benefit of transit service feasibility and ridership, by placing the higher densities closest to transit stops and stations, as in TOD. A large-scale example is found in the case study “Arlington County, Virginia, Transit Oriented Development Densities.” Diversity, or land use mix, narrows in on the reduction of spatial separation by looking at which land uses are complementary in terms of meeting travel needs, and working to place those needs closer together. Thus land use balance reduces travel-causing spatial separation by putting a matching number of jobs closer at hand to housing. All members of the resident population may not choose to work at the close-in jobs, but the opportunity will nevertheless serve to reduce commute distances overall and better balance traffic flows. Placement of shopping and services amongst both residences and workplaces is a matter of both land use mix and design at the community and site levels. It places opportunities for non-work purpose travel and work-related travel closer at hand, with corresponding effects on travel choices, especially enhancing choice of non-motorized travel modes. It is this sort of practical spatial closeness that Kockelman has attempted to address with indices of Entropy, and especially Dissimilarity (Kockelman, 1996), although neither one quite gets down to the fine grain of service types relevant in detailed site planning. Trip Chaining Trip chaining has been identified as being linked to accessibility in two different studies, but in one case it has was posited as a trip-maker solution to poor accessibility, and in the other as a response to opportunities presented by excellent accessibility. These apparently conflicting conclusions clearly identify this is an area requiring further study before even a basic understanding is available. A study of six communities in Palm Beach County, Florida, surmised that trip chaining was the way that residents of sprawling suburbs compensate for limited accessibility to nearby activities and services. While households in sprawled communities generated up to 2/3 more vehicle hours of travel per person than comparable households in a traditional city, differences in accessibility were estimated to be almost ten-fold. Residents of these communities were found to compensate by linking trips of household members into multipurpose tours, constituting what was labeled “linked accessibility.” Average trip time for residents of the most sprawled community dropped off dramatically as trip chains increased in length, leading the researchers to conclude that communities should internalize as many services as possible, and land uses should be arranged to facilitate efficient auto trips and tours (Ewing, Haliyur and Page, 1994). Research on differences in trip generation, trip length, and VMT between mixed-use and predominantly conventional neighborhoods in the Seattle area reached something of the opposite conclusion regarding travel behavior. In the three mixed-use neighborhoods (Queen Anne, Wallingford, and Kirkland), about 60 percent of weekday trips were single- purpose/single destination tours, mainly trips connecting home and work. In contrast, households in the control portion of the sample, located in North Seattle and inner and outer King County suburbs, had about 70 percent of their weekday trips in single-purpose/single- destination tours, suggesting that these residents had a lower rate of multi-purpose trips than

15-95 those in the mixed-use neighborhoods. The average length of trip tours was shorter in the mixed use communities, for example, 7.1 versus 11.4 miles for home-to-work tours, 6.1 versus 6.9 miles for home-to-other tours, and 8.3 versus 13.0 miles for home-to-home tours (no stop over 90 minutes). These differences, despite a fractionally smaller number of trips per tour in the mixed-use neighborhoods, translated into trip mileage totals in the mixed use relative to control neighborhoods that were lower by 10 to 24 percent (Rutherford, McCormack and Wilkinson, 1997). It could be said that the Seattle area mixed-use neighborhoods have a built-in internalization of services of the sort recommended in the Florida study. Thus, although trip chaining effects may not be well understood, both studies support internalization of services as a means to reduce travel. Auto Captivity The earlier discussion of spatial separation notes that if opportunities to meet practical needs are close enough, walking in lieu of auto travel becomes a viable option. This outcome may have the effect of reducing need of an auto at one’s workplace when services are close at hand, which in turn may enhance the viability of choosing commute modes that involve leaving the auto at home. Such commute modes include not only transit riding, but also the auto passenger mode (carpooling), vanpooling, bicycling, and walking to work. Need for an auto at work or en route to work are the primary determinants of what has been termed “auto captivity” (Pratt, 1970). These needs are amenable to remediation with worksite and/or residential land use mix. Needs that have to be met at work, such as eating lunch out, may be resolved by placement of appropriate services within walking reach of the workplace. Needs that don’t necessarily have to be met at work, such as convenience shopping, may be satisfied by locating appropriate land uses either within walking distance of the work place or within a convenient distance — by whatever mode — of home. Thus a theoretical underpinning is provided for the modestly higher non-SOV shares in the presence of land use mix that have been isolated in some studies (See Table 15-18 for example). Auto captivity has been successfully related to accessibility in an application for purposes of Washington, DC, regional mode choice modeling, utilized in the 1970s. Accessibility, as indicated in prior discussion, is closely related to and enhanced by both mix and density. The mode choice models in question, for both work and non-work purpose trips, explicitly divided binary choice between auto and transit into three realms: a percentage of presumed “transit captivity” where there was no evidence of auto use, a percentage of presumed “auto captivity” with no evidence of transit use, and a percentage of “free choice” space — the remainder — where electing use of auto versus transit was based on the time and cost characteristics of the alternative modes for the trip in question. “Captivity” rates were estimated — using travel survey data — on the basis of three income levels, four levels of accessibility at the home end of the trip, and four levels of accessibility at the work end of the trip. The phenomenon of presumed “auto captivity” probably reflected needs for an auto imposed, in some instances, by not only urban form but also total lack of any meaningful transit service. Taking middle-income commuters as an example, commuters who both lived and worked in high accessibility areas were estimated to have an auto captivity rate of 12 percent, versus 34 percent if they lived in high accessibility areas and worked in low accessibility areas, and 38 percent if they lived in low and worked in high accessibility areas. Estimated auto

15-96 captivity was 85 percent for middle-income commuters who both lived and worked in low accessibility areas. Although this example and results for lower income commuters suggest that accessibility at the residence was found to be somewhat more important than at the workplace, for higher income commuters a slight reversal was estimated, with auto captivity rates of 41 percent if they lived in high and worked in low accessibility areas, and 38 percent if they lived in low and worked in high accessibility areas. For all income levels and purposes of travel, less evidence of auto captivity was found wherever accessibility was higher (R. H. Pratt Associates, 1973). The results of this modeling approach tend to support the underlying concept that the nature of opportunities in one’s residential area and at the workplace or destination affect auto dependency, with a corresponding second-order effect on choice of mode for the travel in-between. Auto Ownership Household vehicle ownership appears to be a key intermediate outcome through which impacts of density and other urban form factors are channeled on their way to possibly effecting differences in travel behavior. Some researchers have, therefore, examined whether and to what degree urban form and neighborhood design have a primary effect on auto ownership. The task is made more difficult by the historical relationship of higher densities with both lower auto ownership and lower income, producing uncertainties as to whether it was the higher densities or limited means that reduced auto owning. A number of the research findings have been reported on already, and will be compared and referred back to here. All cross-references are to this chapter’s “Response by Type of Strategy” section. Findings from an examination by Dunphy and Fisher of the 1990 NPTS (National Personal Transportation Survey) data were presented under “Density” — “Density at the Behavioral Level.” Regional level evaluations suggested an inverse relationship between higher densities on the one hand and lower auto ownership and VMT — along with higher transit use — on the other. At the neighborhood level, analysis of differences among households at different density levels found average auto ownership per adult to decline with increasing population densities from almost 1.2 vehicles per adult at 0 to 99 persons per square mile, to 0.7 vehicles at 50,000 persons per square mile or more. A causal relationship operating through density was not established. The authors did postulate that, among other things, density reduces auto travel by influencing the characteristics of households, leading to higher dependency on public transit (Dunphy and Fisher, 1996). Another analysis of 1990 NPTS data, by Schimek, was introduced under “Density” — “Density and Other Indicators at the Behavioral Level” — “Density, Spatial Relationships, Demographics, and Transit.” That research developed models to explore the importance of density and other household and location factors on vehicle ownership, vehicle trips, and VMT, controlling for demographic and geographic factors. A 1 percent higher gross density was found to be associated with 0.11 percent fewer vehicles per household. Household income, household size, and number of workers were determined to be more important determinants of the numbers of vehicles per household. An elasticity for the effect of density on annual household VMT was estimated at -0.069, with one-third identified as a direct effect, and the other two-thirds attributed to an indirect effect operating through auto ownership. Data accuracy issues were discussed in the “Density” section (Schimek, 1996).

15-97 Research by Kockelman, set forth under “Density” — “Density and Other Indicators at the Behavioral Level” — “Density as a Proxy,” arrived at not too dissimilar conclusions with San Francisco Bay Area travel data and much more detailed geographic information. In Kockelman’s work, density was found to be insignificant in direct estimation of VMT per household. Population density and accessibility were, however, the two more significant non-household variables for estimating vehicle ownership. The higher these two variables were, the lower estimated vehicle ownership was, with lower VMT in turn. An elasticity for household vehicle ownership with respect to density of -0.07 was estimated (somewhat less than the -0.11 elasticity implied by Schimek’s work). The estimated elasticity of VMT with respect to vehicle ownership was +0.56 (Kockelman, 1996). As previously noted, these elasticities suggest that the indirect causal effect of population density on household VMT, channeled through vehicle ownership, could be expressed by an elasticity of -0.04. Other researchers also covered earlier in the “Response by Type of Strategy” section have identified a significant although not necessarily large impact of density operating through auto ownership to affect travel demand (Sun, Wilmot and Kasturi, 1998; Messenger and Ewing, 1996; Bradley et al., 2001). Most of those who have identified an impact of density on travel have found it to be primarily a second order, or secondary effect, with auto ownership a major player through which the effect is channeled. In addition, accessibility and other land use mix and design parameters or indicators, including pedestrian environment, have also been estimated in some studies to have significant second order effects channeled through auto ownership (Kockelman, 1996; Cambridge Systematics, Putman Associates and Calthorpe Associates, 1992; Parsons Brinckerhoff, 1993). Transportation Service Levels Practically all of the research findings assembled in this particular chapter, as discussed more explicitly under “Overview and Summary” — “Analytical Considerations,” have been derived without fully detailed quantification of trip-specific transportation service levels. Yet, as demonstrated in other chapters of this Handbook, travelers do respond to different levels and costs of transit service, parking, and highway operation. The typical land use and transportation research model simply leaves certain land use characteristics to stand as surrogates or markers for good or bad transit or highway service. This approach means that transportation service levels are implicit in many of the land use and site design findings, particularly with respect to land use density. On the highway and parking side, density is typically associated with the higher activity levels that equate to more competition for scarce highway space (congestion), and with higher land costs that lead to tighter parking supplies and higher prices for parking. These relationships in turn make driving less attractive where densities are high. On the transit side the reverse occurs: density generally promotes better transit service, as well as improving conditions for walking, a support mode for transit as well as a means of travel on its own. The second-order effects on travel demand of density operating through transit service levels may be the most important of all; the very same effects that have mostly not been explicitly quantified. The relationships between use of transit and land use density are complex and synergistic. As identified in previous discussions, available analyses — while imperfect — tend to suggest that density per se introduces only very limited first order effects on choice of the transit mode, though those effects that do occur are generally positive. Density from a transit perspective, however, is at its heart a measure of how many people are available to make use

15-98 of transit service within a defined geographic area. All other things being equal, the greater the number of people, the greater will be transit ridership, in direct proportion. Transit service is almost never equal, however, among broad areas of substantially different densities, and the differences leverage transit use upward at the higher densities. The relationships involved are not a continuum; there are several thresholds encountered in relating transit use to density. These thresholds are caused by the realities of determining what transit service levels to provide. Transit agencies generally offer no service at all below certain minimum thresholds; therefore, there can be no ridership. Where transit service reasonably can be provided, two measures are used to determine how much service should be offered on a route — a policy standard that defines the minimum level of service (typically called a “policy headway”9), and a capacity standard that defines when additional trips should be added. When density and, hence, ridership are low, a policy headway usually applies. Designed to assure availability of a basic albeit limited service, policy headways may specify bus operating frequencies of one per hour or even less. Such frequencies are adequate and appropriate only when ridership is meager. When density is such that policy headways pertain, each bus typically has capacity to accommodate additional riders. Thus the added demand arising from incremental increases in density can be accommodated without adding service, and service quality remains essentially constant. As density and ridership increase beyond the next threshold, however, the capacity standards come into play. Serving more riders will require adding more service; initially more frequent trips on existing routes and, as ridership grows, new routes. With additional service the two most onerous components of travel by transit, accessing a route and waiting for a vehicle, are made shorter. The incremental service improvement beyond this threshold can be quite large, especially when the policy headway involved a long interval between vehicles. A change from one bus per hour to two buses per hour significantly enhances transit service in the eyes of the potential rider, leading to greater ridership. Each incremental increase in density now not only expands the number of potential riders, but also the transit mode share of the potential riders. Further increases in service frequency lead to further increases in ridership, but at a lesser rate since the reductions in waiting time are smaller. (See Chapter 9, “Transit Scheduling and Frequency,” and Chapter 10, “Bus Routing and Coverage,” for further explanation.) At some point, ridership will reach still another set of thresholds where capital investment in rail or bus rapid transit becomes appropriate. Such investment in turn further enhances transit service and use, offering more competitive speeds in addition to high frequency operation. (See Chapter 7, “Light Rail Transit,” Chapter 8, “Commuter Rail,” and Chapter 4, “Busways, BRT, and Express Bus.”) It is thus crucial to recognize that while first order effects of density on travel demand may be modest, the second order effect deriving from higher sheer volume of potential transit riders — and the higher service levels this can lead to — may have major synergistic consequences. 9 Headway is the scheduled interval between buses or trains operating on a transit route. A frequency of one bus per hour provides an hourly headway, two per hour provide a 30-minute headway, etc.

15-99 This particular effect then has potential for combining with the lower attractiveness of auto travel in high density areas to induce substantially different mode shares where densities are high enough. Estimates of threshold densities to support various types of transit service are presented in the “Related Information and Impacts” section of this chapter under “Transit Service Feasibility Guidelines.” Other Effects and Complexities of Density In research discussed earlier with respect to auto ownership, the 1990 NPTS was used to examine other differences among households at different density levels as well. While the study found that household size varied very little through most density ranges, the highest density places had slightly smaller households. Household income reached a peak at population densities between 1,000 and 5,000 persons per square mile, typical U.S. suburban densities, but then declined. The authors postulate that “travel would decrease in an area with households that tend toward small size, low automobile ownership, and good transit service [as occurs with higher densities] regardless of the area’s population density” (Dunphy and Fisher, 1996). Rather than being of interest as factors through which second order effects of urban form may be channeled, such parameters as household size and income are more in the category of confounding factors that must be statistically controlled for, in order to better understand land use and site design impacts. On the other hand, there are other impacts through which substantial second order effects may be channeled — besides auto ownership and transit service levels — that are of policy interest. Density is associated with relative difficulty of driving and with limited parking, both of which may pertain at either the residence or the workplace or other opportunity site. It is also strongly related to higher parking costs, with the higher land costs in higher density areas a major underlying factor (Kockelman, 1996). Some considerable research has been done on the impacts of parking supply and price on travel demand, as covered in Chapter 18, “Parking Management and Supply,” and Chapter 13, “Parking Pricing and Fees.” Numerous relationships have also been developed, in the course of applied travel demand modeling, relating density to likely parking prices. Nevertheless, as in the case of transit service levels, the ties between density and travel demand operating via roadway and parking conditions and price have not been quantified explicitly and in isolation. Attitudes and Predispositions An issue that arises with respect to traveler response to almost any form of transportation system or land use is the role of attitudes. Some attribute much to attitudes. Conversely, other transportation researchers and practitioners conclude that if the parameters of the system and its travel options are modeled in depth, along with socio-economic factors, that attitudes tend to fade into the background relative to the rational, self-serving reactions of travelers to options presented. An explicit assessment that arrived at a conclusion supporting importance of attitudes was research of Kitamura and others introduced under “Response by Type of Strategy” — “Site Design” — “Community Design and Travel Behavior.” This research examined travel behavior by developing incrementally expanded regression models, utilizing travel survey, attitudinal survey, and physical data obtained in five diverse San Francisco Bay Area neighborhoods. The attitudes of residents were categorized on the basis of responses to 39 statements intended to establish the respondent’s values with respect to the environment,

15-100 transit, suburban life, automotive mobility, time pressure, urban villages, highway construction and management, and work style (workaholic). When scores on these characteristics were introduced into the models of travel behavior based first on household and neighborhood characteristics, the researchers found the attitudinal variables explained a higher proportion of variation in the travel data than either socio-economic/demographic variables or the urban form variables. They concluded that while all three blocks of data contributed significant explanatory power, the attitudinal variables contributed the most. Table 15-45 lists those attitudinal variables that showed up as significant negative or positive explanatory factors in the individual models for different measures of travel behavior (Kitamura, Mokhtarian and Laidet, 1994). Table 15-45 Significant Associations of Attitudes with Trip Rates and Modal Shares in Five-Neighborhood San Francisco Bay Area Study Travel Demand Variable Significant Negative Factors Significant Positive Factors Number Person Trips (person trip generation) • Pro-Environment a • Pro-Transit/Ridesharing • Automotive Mobility Type b Number Transit Trips (transit trip generation) • Automotive Mobility Type • Pro-Transit/Ridesharing Number Bike/Walk Trips (non-motorized trip generation) • Automotive Mobility Type • Pro-Environment • Pro-Transit/Ridesharing • Urban Villager Type Fraction Trips by Auto (auto mode share) • Pro-Environment • Pro-Transit/Ridesharing • Under Time Pressure • Urban Villager Type • Automotive Mobility Type • Workaholic Fraction Trips by Transit (transit mode share) • Automotive Mobility Type • Pro-Transit/Ridesharing Fraction Trips by Bike/ Walk (non-motorized mode share) • Automotive Mobility Type • Pro-Environment • Pro-Transit/Ridesharing • Urban Villager Type Notes: a Variables in italics, although significant, were less so than other attitudinal variables for the travel demand independent variable in question. Two attitudinal variables were never significant: “Suburbanite” typecasting and “TCM” (persons willing to pay tolls, and pro- High Occupancy Vehicle lanes). b Attitudinal variables in plain type are the most significant for the parameter in question. Source: Kitamura, Mokhtarian and Laidet (1994). The results of testing attitudinal variables led the researchers to conclude that transit- and pedestrian-friendly land use policies may not produce the desired travel changes unless attitudes are also changed (Kitamura, Mokhtarian and Laidet, 1994). Like most studies of urban form vis-à-vis travel behavior, this one did not model the parameters of the transportation network and its travel options in depth, as would be done in a full-scale travel demand model such as the San Francisco Travel Model introduced under “Response by Type