Differential Neighborhood Organization: A Multi-Level Theory of Crime

University Of Washington, Seattle WA

Investigators

Abstract

SES-096662 Ross Matsueda University of Washington This project addresses a crucial social problem facing contemporary America: criminal violence in local communities. An innovation of the project is the specification of a multi-level model, in which individual choices produce community social capital, which affects crime rates. The direction of this effect depends on how social capital is used--either to facilitate or control crime. Crime rates are hypothesized to be high when organization in favor of crime is strong and low when organization against crime is strong. The project studies organization against crime by examining how neighborhood social capital and collective efficacy are produced. Residents are hypothesized to act rationally: they exchange favors and information when they have an incentive to do so (e.g., needing help fixing a car) and are free of constraints (e.g., they have the time to reciprocate). In turn, these incentives and constraints are hypothesized to be a function of broader community structures and resources. Exchange among residents has a positive externality--the creation of neighborhood social capital, including trust, social ties, obligations, and norms. The resulting social capital provides resources for collective efficacy, the ability of a neighborhood to solve local problems, such as crime. Collective efficacy is, in turn, hypothesized to reduce neighborhood criminal violence. A final hypothesis to be tested is that the correlation between neighborhood disorder and violence, based on the broken windows hypothesis, is actually spurious because collective efficacy explains both disorder and violence. The project also investigates organization in favor of crime by testing the hypothesis that a violent street culture based on the code of the street develops in structurally disadvantaged neighborhoods. Residents, particularly young black males, are hypothesized to have less human capital, poor labor market prospects, and few resources, leading to alienation from society. This alienation is hypothesized to be exacerbated by observations of racial bias and profiling by police, which leads to distrust, and in turn, increases the likelihood of turning to the streets to resolve disputes and gain respect, status, and credibility. It is hypothesized that when such alienated residents reach a critical mass in the presence of social capital, their frustration is translated into a neighborhood system of achieving status through a code of the street (e.g., never back down from a fight, always retaliate when a peer is disrespected, project a tough image on the street). Finally, collective efficacy and neighborhood street codes are hypothesized to interact in their effects on violence. To test the hypotheses, the project will use survey data collected on nearly 6,000 households across 123 census tracts in Seattle, along with neighborhood census data and police crime data. The nested design will be used to estimate multilevel generalized linear models, including individual-level rational choice models of reciprocated exchange and distrust of police and neighborhood-level models of collective efficacy and codes of the street. To explore further the possible mechanisms by which individual action translates into macro-level outcomes, the project will sample neighborhoods on and off the regression lines, and conduct in-depth qualitative interviews of residents. This exploratory analysis will help identify causal mechanisms specified in the model, and by examining outlier neighborhoods, provide insights into mechanisms not measured in the survey. Broader Impacts: The project is a key feature of the Racial Democracy, Crime, and Justice Network, an NSF-supported research network for fostering the study of race, crime, and justice and increasing diversity in the academy by mentoring junior faculty of color. The PI and several members of the project team (faculty of color) are core members of RDCJN. The project incorporates a gender and racially diverse set of graduate and undergraduate students into the research process. The PI will incorporate his research findings Into both undergraduate and graduate teaching, and make data available to the public.