Travel and the Built Environment

Abstract

Problem: Localities and states are turning to land planning and urban design for help in reducing automobile use and related social and environmental costs. The effects of such strategies on travel demand have not been generalized in recent years from the multitude of available studies.

Purpose: We conducted a meta-analysis of the built environment-travel literature existing at the end of 2009 in order to draw generalizable conclusions for practice. We aimed to quantify effect sizes, update earlier work, include additional outcome measures, and address the methodological issue of self-selection.

Methods: We computed elasticities for individual studies and pooled them to produce weighted averages.

Results and conclusions: Travel variables are generally inelastic with respect to change in measures of the built environment. Of the environmental variables considered here, none has a weighted average travel elasticity of absolute magnitude greater than 0.39, and most are much less. Still, the combined effect of several such variables on travel could be quite large. Consistent with prior work, we find that vehicle miles traveled (VMT) is most strongly related to measures of accessibility to destinations and secondarily to street network design variables. Walking is most strongly related to measures of land use diversity, intersection density, and the number of destinations within walking distance. Bus and train use are equally related to proximity to transit and street network design variables, with land use diversity a secondary factor. Surprisingly, we find population and job densities to be only weakly associated with travel behavior once these other variables are controlled.

Takeaway for practice: The elasticities we derived in this meta-analysis may be used to adjust outputs of travel or activity models that are otherwise insensitive to variation in the built environment, or be used in sketch planning applications ranging from climate action plans to health impact assessments. However, because sample sizes are small, and very few studies control for residential preferences and attitudes, we cannot say that planners should generalize broadly from our results. While these elasticities are as accurate as currently possible, they should be understood to contain unknown error and have unknown confidence intervals. They provide a base, and as more built-environment/travel studies appear in the planning literature, these elasticities should be updated and refined.

Research support: U.S. Environmental Protection Agency.

Keywords:

• vehicle miles traveled (VMT)
• walking
• transit
• built environment
• effect sizes

Acknowledgments

The authors wish to acknowledge funding for this study from the Development, Community, and Environment Division of the U.S. Environmental Protection Agency. We also wish to acknowledge data and other assistance from the following individuals, hoping we didn't miss anyone: Chandra Bhat, Marlon Boarnet, Rob Boer, Mark Bradley, Jason Cao, Dan Chatman, Cynthia Chen, Mike Duncan, Yingling Fan, Ann Forsyth, Larry Frank, Jessica Greene, Mike Greenwald, Daniel Hess, Ken Joh, Kara Kockelman, Rich Kuzmyak, Chanam Lee, Tracy McMillan, Petter Naess, Mike Reilly, Daniel Rodriguez, Elizabeth Shay, C. Scott Smith, Qing Shen, Xiaoduan Sun, Chris Zegras, Ming Zhang, and Brenda Zhou.

Some limitations of this meta-analytic study should be mentioned. Although the minimum number of studies to permit a metaanalysis is only three studies (Treadwell, Tregear, Reston, & Turkelson, 2006) and many published meta-analyses contain nine or fewer studies (Lau, Ioannidis, Terrin, Schmid, & Olkin, 2006), the small number of seven studies included in this meta-analytic review limits the generalizability of our findings and the possibilities of examining and adjusting for publication bias by means of more complex analytic methods (Macaskill, Walter, & Irwig, 2001). (p. 605)

Notes

^a. Cao, Mokhtarian, et al. (2009a) notes nine different approaches used to control for residential self-selection. The least rigorous incorporates attitudinal measures in multivariate regression models, while the most rigorous jointly estimates models of residential choice and travel behavior, treating residential choice as an endogenous variable.

^a is the mean estimated probability of occurrence.

^b. Applied only to positive values of the Tobit distribution (i.e., where y> 0).

Ψp < .10

*p < .05

**p < .01

Ψp < .10

*p < .05

**p < .01

^a. Sign reversed.

Ψp < .10

*p < .05

**p < .01

^a. Sign reversed.

*p < .05

**p < .01

^a. Sign reversed.

Ψp < .10

*p < .05

**p < .01.

^a. Sign reversed.

^b. Sign reversed and multiplied by 2 to make x variable equivalent to others.

Ψp < .10

*p < .05

**p < .01

^a. Proportional reduction relative to conventional suburban neighborhood.

Ψp < .10

*p < .05

**p < .01

^a. Computed at median cutpoint by Jason Cao.

^b. Significance level indeterminate.

Ψp < .10

*p < .05

**p < .01

^a. Sign reversed.

^b. Significance level indeterminate.

Ψp < .10

*p < .05

**p < .01

^a. Sign reversed.

^b. Significance level indeterminate.

^c. Because either the elasticity or significance level must be misreported in the published article we dropped this observation from the meta-analysis.

^d. Significance level indeterminate.

Ψp < .10

*p < .05

**p < .01

^a. Sign reversed.

*p < .05

**p < .01

^a. Sign reversed.

*p < .05

**p < .01

^a. Proportional increase relative to conventional neighborhood.

*p < .05

**p < .01

Ψp < .10

*p < .05

**p < .01

^a. Sign reversed.

Ψp < .10

*p < .05

**p < .01

^a. Sign reversed.

*p < .05

**p < .01

a. Sign reversed.

*p < .05

**p < .01

^a. Sign reversed.

^b. Sign reversed and multiplied by 2 to make x variable equivalent to others.

^a. Proportional increase relative to conventional neighborhood.

1. A full list of studies is available from the corresponding author.

2. Vehicle trips (VT) is not studied as widely as these other outcome measures and is not related to as many important outcomes. However, it is a critical determinant of regulated vehicle emissions, which was the focus of our 2001 literature review.

3. The percentage varied depending on which locations were paired and compared, whether urban and suburban locations, urban and exurban, etc.

4. Transit route density is measured by miles of transit routes per square mile of land area.

5. Linear regression is used where the travel variable in continuous, Poisson regression where the travel variable is a count, logistic regression where the dependent variable is a probability, and so forth.

6. Several studies applied ordered probit regression to data on counts of walk and transit trips. We excluded all but one of these studies from the meta-analysis because the breakpoint parameters (μ) for the ordered categories were unavailable, which meant we could not calculate marginal effects. These parameters were available for one ordered probit study (Greenwald & Boarnet, 2001), and Jason Cao computed elasticities for us. We used elasticities for the median ordered category.

7. Due to a dearth of solid research, we could not study certain important travel outcomes with meta-analysis. Most notably, this article is silent regarding the effects of the built environment on trip chaining in multipurpose tours, internal capture of trips within mixed-use developments, and the choice of bicycling as a travel mode.

8. The following quotation from Rodenburg, Benjamin, de Roos, Meijer, and Stams (2009) explains that a meta-analysis in another field settled on seven studies as a minimum sample size: