Note from the future: This was a long project while at UChicago, using cell phone data to measure park use. It resulted in two papers. The first performed the measurement and showed the limitations of traditional measures of park access. The second, with colleagues from the psychology department, looked at the relationships between park use, street activity and crime.

Parks improve mental and physical health by nurturing social cohesion and enabling physical activity. City planners have a professional and ethical responsibility to provide public goods equitably. So who gets parks?

Park access has traditionally been evaluated using constructed variables of potential access: distance buffers or gravity models. These models have major limitations. They ignore commutes and other more intricate mobility behaviors. Their parameters are rarely if ever calibrated. Results are sensitive to these parameters, as well as to the choice of models, and assumptions made in defining distances, entry points, and land use. The (interactive) map below illustrates how different models lead to different conclusions about the relative accessibility of parks in the twenty largest American cities — and how those models differ from true visitation rates.

In this paper, I use a dataset of smartphone locations to identify actual park visits. Using these data, I calibrate existing models and evaluate their performance. Both potential, spatial acccess to parks and realized use are lower in Black or Hispanic than non-minority neighborhoods. However, even controlling for potential access, realized use in minority neighborhoods is lower than expected. In other words: the existing models are not just imprecise, they are systematically biased.

This study underscores the need to consider multiple barriers of access when evaluating public resource accessibility. It highlights the potential of new forms of “big data” in describing urban space and evaluating the provision and use of public resources.