Do the Joneses Help You Keep Up? A Natural Experiment in Exposure to Nonpoor Neighbors

Abstract

This study capitalizes on a natural experiment in Montgomery County, Maryland, where low-income applicant families are randomly assigned to public housing that is either (a) clustered within seven public housing developments or (b) scattered into market-rate subdivisions via the county's inclusionary zoning policy. Through a survey of 453 public housing residents, we find that adults who lived in scattered public housing reported a lower proportion of low-socioeconomic status (SES) social ties generally and a lower proportion of low-SES neighbors specifically in their social networks. They also counted more high-SES individuals in their social networks, and this effect was related to the amount of time they lived in the neighborhood. Living in scattered public housing had no adverse effect on feelings of neighborhood belonging or satisfaction. The socioeconomic composition of respondents' social networks was associated with two health outcomes for respondents (smoking and depression) and modestly associated with respondents' household income.

Keywords::

• public housing
• social networks
• inclusionary zoning
• socioeconomic status
• health

Acknowledgments

We thank the MacArthur Foundation for funding this research. Craig Pollack receives salary support from the National Institutes of Health's National Cancer Institute and Office of Behavioral and Social Sciences Research (1K07CA151910-01A1). We are also grateful for Rachel Kleit's input on social network surveys and for the research assistance of Matthew Hoover. The opinions expressed herein are solely those of the authors and do not represent those of the MacArthur Foundation, the RAND Corporation, or Johns Hopkins University. Any errors are the responsibility of the authors.

Notes

^1. Social networks are distinct from the overlapping concept of social support; whereas social networks comprise social ties that are not implicitly positive or negative, social support generally refers to relationships that facilitate feelings of being cared for and valued as an individual or that transmit information, advice, or assistance (Gottlieb & Bergen, 2010). People may receive social support in greater or lesser degrees depending on the structural and compositional features of their social networks.

^2. The precise likelihood varies by household since households must be not only income-eligible but also size-eligible for the given home that has become vacant and is to be leased. For example, if a three-bedroom public housing unit becomes available, only a household with three to six persons on the waiting list is eligible, to prevent under- or overhousing stipulations. Further complicating the eligibility, the age and gender of children in the household determines whether a shared bedroom(s) for the children is acceptable per federal regulation and further restricts the likelihood of a household of three to six persons obtaining the three-bedroom apartment.

^3. HOC electronic and paper records held on-site do not date back far enough to trace the original date that the lottery started. However, one current employee and a retired employee have traced the lottery back to at least January 1, 1992.

^4. For example, based on a hand count of six months' worth of public housing offers made by the housing authority in 2008 (that the housing authority performed at our request), four out of 57 elderly or family households (7%) turned down either the first or second unit offered.

^5. On average, census block groups contain 600–3,000 people.

^6. Two comparisons were significant at the 10% level. Among respondents who have lived in their home for 7–11 years, 7% of scattered-site respondents have a spouse or partner compared with 23% of clustered respondents (p = .07). Within this same bracket of years, 12% of scattered-site respondents are Hispanic compared with 3% of clustered-site respondents (p = .10).

^7. On average, 62% of alters were female, 34% were family members, and 52% were friends. These social network characteristics did not significantly vary by whether the respondent lived in scattered-site or clustered public housing.

^8. While housing type could theoretically influence whether a respondent has a spouse or partner, we did not find a statistically significant relationship, and we include this variable to increase precision of the estimated association of interest.

Additional information

Funding

This work was supported by The John D. and Catherine T. MacArthur Foundation, grant number 96241-0.

Notes on contributors

H. Schwartz

Heather Schwartz is a full policy researcher at the RAND Corporation. Her work focuses on education and housing policies intended to reduce the negative effects of poverty on children and families. Her work falls in four policy areas intended to help close the income achievement gap: economically integrative housing and school programs, early childhood learning opportunities, expanded learning time, and school accountability measures. She is interested in experimental and nonexperimental methods to make causal inferences about the effects of public policies. Schwartz holds a PhD in education policy from Columbia University.

S. Burkhauser

Susan Burkhauser is a doctoral fellow at the Pardee RAND Graduate School with a specialization in education policy and a concentration in economics and quantitative methods. She is also an assistant policy analyst at the RAND Corporation. Her research focuses on K–12 education, postsecondary education, and school leadership. She holds an MPhil in public policy analysis from the Pardee RAND Graduate School and a BS in policy analysis and management from Cornell University.

B.A. Griffin

Beth Ann Griffin is a senior statistician at the RAND Corporation, where she codirects the RAND Center for Causal Inference. Her statistical research focuses on causal effects estimation when using observational data. Her substantive research has primarily fallen into three areas: substance abuse treatment for adolescents, neighborhood-level predictors of health, and military health. Griffin's research has appeared in leading journals such as Drug and Alcohol Dependence, Journal of the American Statistical Association, Annals of Applied Statistics, American Journal of Public Health, and Health and Place. She received her PhD in biostatistics from Harvard University.

D.P. Kennedy

David P. Kennedy is a behavioral/social scientist at RAND. He specializes in the effects of social relationships, especially romantic relationships, on health. Methodologically, he specializes in the combination of qualitative and quantitative research methodology and social network data collection and analysis, including mixed-methods studies of social networks. Kennedy is currently leading two studies using methods to elicit and analyze quantitative and qualitative data about the personal networks of homeless and formerly homeless people in Los Angeles to understand and reduce HIV risk behavior and substance use. He holds a PhD in anthropology from the University of Florida.

H.D. Green, Jr.

Harold D. Green, Jr. is a senior behavioral scientist at the RAND Corporation, a professor at the Pardee RAND Graduate School, and the director of the RAND Center for Applied Network Analysis and System Science. He uses network analyses to understand the social and cultural determinants of health. Green holds a PhD in anthropology from the University of Florida and is an alumnus of the University of Illinois training grant in quantitative psychology; the Center for Supercomputing Applications Center for the Humanities, Arts, and Social Sciences; and the Science of Networks in Communities Research Group.

A. Kennedy-Hendricks

Alene Kennedy-Hendricks is a doctoral student in the Department of Health Policy and Management at the Johns Hopkins School of Public Health with a concentration in social inequalities and health. She has been involved in research on the impacts of public housing policies on health behaviors and outcomes, the role of safety-net providers in facilitating access to health care among vulnerable populations, and policy responses to address the nonmedical use of prescription opioids. She holds a BA from the George Washington University.

C.E. Pollack

Craig Pollack is an assistant professor of general internal medicine at the Johns Hopkins School of Medicine, whose research focuses on how social factors—with an emphasis on housing and neighborhood environments—influence health and health care delivery. He has applied the methods of social network analysis to studying the structure of cancer care. Clinically, he practices as a general internist. Pollack earned a medical degree and completed his residency at the University of California, San Francisco, and also holds master's degrees in health sciences and health policy research.