Background
Efforts to understand how urban form influences physical activity confront a difficult methodological problem – persons likely choose where to live in part based on their desired physical activity. The existing cross-sectional correlations between dense, urban-oriented environments and physical activity do not establish that urban environment causes an individual to change their activity. To address this issue, we take advantage of an opportunity to implement an experimental study design. The California Safe Routes to School (SR2S) program allocated funds for construction projects (e.g. sidewalks, bicycle paths, speed humps) near schools, with the intention of making the environment safer and more conducive to walking and bicycling. We report on a study of children’s travel to school before and after
the SR2S construction program.

Objectives
The primary objective was to observe changes in the physical environment, and to infer whether those changes altered travel behavior in ways that could be associated with increased physical activity. A secondary objective is to get insights into the feasibility of quasi-experimental studies in this problem domain.

Methods
We surveyed parents of third through fifth graders at nine schools before and after the SR2S program construction. Survey response rates varied by school, but were typically above 50%, yielding 1,562 returned “before construction” surveys and 1,111 returned “after construction” surveys.

The research design provides three methods to infer causal influences of the SR2S program:

(1) Observing changes in reported walking or bicycling travel of children after the SR2S project is constructed,
(2) observing how those changes differ across students for whom the SR2S project is along their path to school, versus children who would not pass that project
going to or from school, and
(3) observing how different types of construction projects yield different changes in child walking or bicycling. Considered together, this provides three methods to isolate the causal influence of urban form on child walking or bicycling travel to or from school.

Results
The results suggest that SR2S projects intended to increase walking or bicycling have had the desired effect. The results also illustrate possible difficulties with quasi-experimental research designs. For example, a widely publicized child abduction case occurred between some “before” and “after construction” observations, and there is some evidence that walking and bicycle travel to schools dropped throughout the sample after that abduction. Against that backdrop, the SR2S programs appeared to at least mitigate the reduction in walking for children for whom the project was along their route to school.

Conclusions
We conclude that urban form modification can have an independent and causal impact on non-motorized travel behavior among our study population. This is encouraging news for efforts to use urban form to alter physical activity. Yet this study also illustrates difficulties in applying the quasi-experimental method, and provides evidence that a broad range of factors other than urban form are influential in children’s travel behavior.