Methods: To answer this question, I constructed a unique dataset on neighborhood risk factors, subject demographics and individual risk factors, and criminal justice outcomes for a cohort of probationers in San Francisco County (n=2,050). The dataset includes data from multiple sources, including administrative (probation and courts), extant (U.S. Census, California Department of Alcoholic Beverage Control, and San Francisco Police Incidents), and primary (systematic social observations of neighborhoods). Using these data, I first analyze the main effects of individual and neighborhood risk factors on time to recidivism using nested frailty models (i.e., multi-level survival analysis). Next, I examine the interaction between neighborhood risk factors and individual criminal risk scores.
Results: I find that the effects of neighborhood risk factors on recidivism varies by the individual criminal propensity of probationers, as demonstrated by statistically significant interactions between criminal propensity scores and neighborhood variables. Neighborhood disadvantage, disorder, residential mobility, and ethnoracial diversity have exponentially greater effects on the hazard of recidivating for probationers low in individual criminal propensity (β=.17-.27) than those high in criminal propensity (β=.01-.04).
Conclusions and Implications: Prior studies, largely informed by the psychology of criminal conduct, suggest that those highest in criminal propensity are most likely to recidivate because they face the greatest number of risk factors. Novel findings from this study suggest instead that those most vulnerable to risky neighborhood conditions are those lowest in individual criminal propensity. I argue that moving beyond the psychology of criminal conduct toward a deeper understanding of the ecology of criminal conduct not only adds to the scientific knowledge base for understanding criminal justice entry and re-entry, but also fits squarely into a social work research agenda. Implications for theory and practice are drawn.