|
Abstract
|
Research Question
Does the effect of arrest for misdemeanor domestic assault on suspect recidivism vary by characteristics of the neighborhood in which the individual suspect resides?
Data
This analysis combined 1990 Census data for Milwaukee with data on 1133 domestic assault suspects randomly assigned in 1987-1988 to arrest vs. warning in the Milwaukee Domestic Violence Experiment (Sherman 1992). Outcomes were measured by the prevalence of one or more new arrest for misdemeanor domestic assault in the 6 months after the randomly assigned response to the presenting incident.
Methods
The analysis compared models using only individual-level suspect characteristics to models that combined individual data with Census tract characteristics relevant to community stakes in conformity and social disorganization.
Findings
The models show substantial effects of structural characteristics of community context that moderate arrest effects on recidivism. Arrested suspects with medium individual stakes in conformity show an 18% reduction in the prevalence of repeat domestic assault, but arrestees who live in Census tracts with medium levels of community stakes in conformity show a one-third (32%) greater likelihood to repeat domestic assault than suspects in those medium stakes Census tracts who were not arrested. For all suspects with high levels of individual stakes, arrest decreased recidivism within 6 months by 39.7% compared to not making an arrest. In contrast, for individuals who live in communities with high stakes, arrest decreased recidivism within 6 months by 50%.
Conclusion
The effects of arrests on offending in this experiment were highly dependent on the social strength of community characteristics where suspects resided. Police strategies for safeguarding victims may be vulnerable to backfiring in areas of greatest social vulnerability. Only more randomized experiments can show whether mandatory arrest policies help or hurt domestic abuse victims in different kinds of communities.
Language: en |