EITM: Neighborhood Choice and Neighborhood Change: Evaluating Dynamic Models of Residential Segregation

University Of California-Los Angeles, Los Angeles CA

Investigators

Abstract

This research will (1) construct models for explaining patterns of racial and economic segregation in large American cities, and (2) develop innovative methods for evaluating these models. The models connect the decisions of individuals who belong to various race-ethnic and economic groups about where to live to aggregate patterns of neighborhood change. The project is multi-disciplinary and will be conducted by two teams of investigators. There are two major components. The first part is to analyze the effects of neighborhood characteristics - especially their race/ethnic and economic makeup - on individuals' decisions about where to move. This research is based on unique data from the Los Angeles Survey of Families and Neighborhoods on residential mobility and neighborhood characteristics. It will examine the residential mobility of individuals using statistical models for discrete choice. Using realistic computational models, it will also examine the implications of individual decisions for changes in patterns of racial and economic segregation. The second part of the research will evaluate these models by adapting novel statistical methods for the evaluation of computer models. These methods have been applied to problems in engineering, atmospheric science, and transportation, but have yet to be applied in the social sciences, will be used to assess how well the proposed models explain and predict historical changes in the distribution of populations across neighborhoods, as shown in data from the Decennial Censuses. This research will make a number of scientific contributions: (1) It will be a major advance over past studies of residential segregation, which have tended to rely on Census data, vignette data on survey respondents' stated preferences for neighborhoods, or oversimplified artificial models of residential mobility. The proposed research combines data with an explicit model of the relationship between individual mobility and aggregate neighborhood change. (2) It will serve as a demonstration of how to analyze the interdependence of micro level phenomena (residential moves) and macro level phenomena (residential segregation) at both the theoretical and empirical level. (3) It incorporates state of the art methods of evaluation of computational models into social science research for the first time. (4) It will serve as a prototype for further contemporary and historical analyses of residential segregation throughout the United States. This research will also have an impact beyond its specific scientific objectives: (1) In focusing on racial and economic residential segregation, it confronts one of the most enduring social problems in the United States. Segregation is a key source of persistent inequality, social disadvantage, and conflict. Understanding the processes that maintain and change it is a vital social issue. (2) The proposed research will serve as a paradigm for interdisciplinary collaboration between statisticians, who have a deep understanding of general methodological issues and the applications of methodology in numerous disciplines, and social scientists, which have a deep knowledge of subject matter in their own fields but limited familiarity with the latest methodological developments. (3) It will develop methods of analysis and empirical examples that can be used to bring rigorous simulation methods into social science curricula. (4) It will serve as a model for the dissemination of scientific results. In addition to the usual venues for the publication of research results, it will provide a web site that makes publicly available all documentation, intermediate data, software, and working papers.