UCLA

This study evaluates empirical evidence of a relationship between social characteristics of neighborhoods, disorder, and chronic disease outcomes in a representative, longitudinal sample of adults in Los Angeles County, CA (n=900).

Theory suggests that key population characteristics cause a breakdown of social control within neighborhoods, which leads to both social and physical disorder. When residents encounter disorder in neighborhoods, it can cause fear of victimization and social isolation, each sources of psychosocial stress. Disorder in neighborhoods is therefore a source of chronic stress and exposure to it over time can lead to chronic disease. In this study I look at three health outcomes among adults: hypertension, obesity, and depression. I also examine how the relationship between neighborhood characteristics and health outcomes varies by personal income.

I use data from two waves of the Los Angeles Family and Neighborhoods Survey, collected in 2000-2001 and 2006-2008 respectively. I use multilevel and standard OLS and logistic regression models to show the relationship between neighborhood characteristics at the first data wave and health outcomes at the second data wave, an average of 6.7 years later, net of individual-level controls. Models show the effect of neighborhood on change in health.

My findings suggest that neighborhood poverty and observed physical disorder have a measurable effect on increases in blood pressure only. This relationship is contingent on individual income, and the direction of the interaction is unexpected--people whose incomes are mismatched with the poverty status of the area in which they live have higher blood pressure than those whose income is similar to the area in which they live. The effects of neighborhood characteristics are not mediated by disorder, contrary to findings of other studies. I find no residual effect of neighborhood on change in BMI, possibly because the time window is not long enough to capture the important exposures. I also find no effects for depression, probably due to low prevalence of depression in my sample.

I also present a sensitivity analysis of the models to specification of neighborhood scale using four different geographic scales derived from census geography. Finally, I present results from a novel way of operationalizing "neighborhood" using time-weighted activity spaces. My main results are robust to these changes to the definition of neighborhood.