Beyond the Sidewalk: Pedestrian Risk and Material Mismatch in the American Suburbs

Abstract

American suburbs designed in the twentieth century to house a growing middle class are experiencing escalating poverty. Utilizing a mixed-method social epidemiology, this study finds high pedestrian risk in these communities, where a high poverty rate coincides with high levels of suburban sprawl. In declining suburbs, those without access to an automobile navigate a landscape centered upon the private automobile. Suggesting the term ‘material mismatch’ to describe cases where the configuration of the built environment is incongruent with the needs of a growing population of residents, the study makes a contribution to mobility studies, environmental justice research, and urban sociology.

Key Words:

• Transportation
• Mobility
• Pedestrian risk
• Sustainability
• Walkability
• Environmental justice
• Suburbs
• Urban sociology

Acknowledgements

I am indebted to Paul DiMaggio, Harvey Molotch, Hana Shepherd, Doug Massey, Katherine Newman, Alex Murphy, Katherine Slevin and Eric Klinenberg for offering helpful responses to earlier drafts, presentations, or discussions pertaining to this work. All errors are mine. The research presented in this paper was partially funded by the Princeton University Center for Health and Wellbeing.

Notes

1. FARS offers a national, annual census of pedestrian deaths, as all traffic fatalities are required by law to be reported by local officials to the federal government. Using the FARS data-set, all pedestrian fatalities (person code = 5) occurring during the period of study were identified and annual fatality counts were generated for each county and each year covered by the analysis. The log pedestrian mortality rate (per 100,000 persons per year) was derived by calculating the mortality rates for each of the years in the period of analysis using county population estimates from the Census Bureau’s annual population estimates program.

2. The author would also like to thank an anonymous reviewer for suggesting the inclusion of this variable.

3. Poisson, zero-inflated Poisson, and negative binomial regression models were fit as well, using counts of fatalities rather than population-adjusted rates as the outcome variable, and including county population as a control variable measuring exposure. Results (available upon request) were identical in terms of the direction and significance of the key variables of interest. The log mortality rate analysis is presented here because this model does the best job predicting risk among outliers that produced overdispersion and zero inflation in count-based models.

4. When two continuous variables are used to construct an interaction term, the effects of constituent terms become virtually meaningless when the interaction is included. For this reason, the lower order effect of sprawl is only of interest when the interaction term is omitted, and even then, its effect is only of passing interest, given the clear significance of the interaction term. Put differently, the interaction term suggests that it is only possible to accurately interpret the effect of sprawl by interacting sprawl with carless households. (See Brambor, Clark, and Golder (2006) for an insightful discussion of the interpretation of interactions effects.)

5. Although freely available from public sources, the names of victims have been changed out of respect for the privacy of their families.