, Volume 35, Issue 2, pp 179–199

Explaining obesity with urban form: a cautionary tale


  • Tudor D. Bodea
    • School of Civil and Environmental EngineeringGeorgia Institute of Technology
    • School of Civil and Environmental EngineeringGeorgia Institute of Technology
  • Michael D. Meyer
    • School of Civil and Environmental EngineeringGeorgia Institute of Technology
    • Georgia Transportation InstituteGeorgia Institute of Technology
  • Catherine L. Ross
    • City and Regional Planning ProgramGeorgia Institute of Technology
    • Center for Quality Growth and Regional DevelopmentGeorgia Institute of Technology

DOI: 10.1007/s11116-007-9148-2

Cite this article as:
Bodea, T.D., Garrow, L.A., Meyer, M.D. et al. Transportation (2008) 35: 179. doi:10.1007/s11116-007-9148-2


In recent years, there has been a dramatic increase in studies exploring associations between the built environment and obesity. Many studies have found that built environment characteristics, such as high-density land developments, mixed-land uses, and connected street networks, are associated with lower rates of obesity. However, depending on the research field and the researcher, how one specifies the experimental model and how sociodemographic characteristics of the population are defined and included in the model has led to different policy conclusions and implications. This is not a surprising observation; however, it is one that does seem to have been lost in current discussions. This article highlights several data-processing, model-specification, and model-estimation factors that should be comprehensively considered in studies of the built environment and obesity. Empirical results based on data from Atlanta, GA, USA, illustrate that the association between the built environment and obesity is sensitive to how age, income, and educational attainment are included in the model. Also, a detailed examination of land-use-mix measures shows that it is difficult to create this measure and that results are sensitive to the treatment of missing values. Models that distinguish between overweight and obese individuals are shown to provide richer insights into the associations among obesity, built environment, and sociodemographic characteristics for the Atlanta area. The article concludes by offering modeling recommendations for future studies.


Land use Obesity Specification bias Urban form

Copyright information

© Springer Science+Business Media, LLC. 2007