The Causal Effects of Active Living-Oriented Zoning on Adult Leisure Time Activities
thesisposted on 08.02.2018, 00:00 by Xuan Yin
This is the first study on the causal effects of zoning on active living outcomes, and the first study to quantitatively verify the mediation of the built environment for the effect of zoning on active living outcomes. I introduce the causal inference to the literature of the associations among zoning, built environment, and active living outcomes. To solve the endogeneity problem of zoning and built environment, I build two instrumental variables (IVs), which are manufacturing establishment density in 1900 and farmland proportion in 1900. The two IVs represent the conflict between the rapidly-developing manufacturing industry and the traditional agricultural economy in the late 19th century. I show that the conflict gave birth to the American zoning and shaped the built environment by exploring the history and the institution of American zoning. I argue for the validity of the two IVs by investigating the literature and the history of American manufacturing industry and performing statistical analysis. By utilizing IV Probit model and the general method of moments (GMM), I test the endogeneity of zoning and estimate the causal effect of zoning on active living outcomes. In addition, I conduct mediational analysis to verify the mediation of the built environment for the effect of zoning on active living outcomes. The results show that the active living-oriented zoning promotes adult leisure-time physical activity, such as walking, running, and bicycling, and discourages adult leisure-time sedentary behaviors at home, such as TV watching, radio listening, and music listening, partly through shaping the built environment. The data come from five sources. Zoning data originate from the research team at the Institute for Health Research and Policy at the University of Illinois at Chicago and are used to construct the independent variable—active living-oriented zoning. NAVTEQ 2011 third quarter GIS data and the American Community Survey (ACS) 2011 1-year estimates were combined to build the county-level independent variable—walkability—which is used to measure built environment. The American Time Use Survey 2010-2015 is used to create the outcome variables of time usage and individual-level control variables. The ACS 2011–2015 5-year estimates are used to construct the county-level control variables. The 1900 census is used to create the two instrumental variables.