Residential Instability, Homelessness, and Criminal Recidivism

Background: The ability of social workers and probation officers to reduce the rate at which probationers recidivate (i.e., are rearrested or re-imprisoned) depends on their ability to identify and intervene upon modifiable recidivism risk factors. Historically, research on recidivism has focused on psychological (e.g., criminal thinking) and social (e.g., criminal associates) factors, finding socioeconomic factors (e.g., income) less relevant. This study challenges the premise that socioeconomic factors are unrelated to recidivism. Focusing on the neglected role of housing in shaping recidivism outcomes, this study asks: (1) is housing instability a risk factor for recidivism, above and beyond other criminal risk factors and demographic characteristics? And, (2) is homelessness a risk factor for recidivism, above and beyond other criminal risk factors and demographic characteristics?

Methods: To answer this question, we use a longitudinal study design and draw on a unique dataset containing data on criminal risk factors, demographics, housing characteristics (number of moves prior to probation start, homelessness at probation start), and criminal justice outcomes (revocation or re-arrest) for a cohort of probationers in San Francisco, California (n=4,341).  First, we used two Kaplan Meier curves to examine differences in the probability of recidivating between (1) frequent (≥ 3 moves) and infrequent movers (≤ 2 moves), and (2) homeless and housed probationers.  Next, we use two multivariate Cox Proportional Hazard models to assess the degree to which residential instability and homelessness predict time to recidivism, above and beyond other risk factors.  These models adjust for recidivism risk (using a validated risk prediction instrument containing a variety of psychological and social characteristics; COMPAS), gender, race, and age.

Results: At one year, frequent movers had an unadjusted probability of recidivating of 52% (CI = .50-.54), while infrequent movers had an unadjusted probability of recidivating of only 32% (CI = .28-.36). In the same period, homeless probationers had an unadjusted probability of recidivating of 55% (CI = .53-.57), while housed probationers had an unadjusted probability recidivating of only 31% (CI = .29-.34). Multivariate models indicate that, even after adjusting for other established risk factors, frequent movers are 37% (HR = 1.37, p < .001) more likely to recidivate at any given time than infrequent movers, and homeless probationers are 52% (HR = 1.52, p < .001) more likely to recidivate at any given time than housed probationers. Post-hoc analyses revealed that the effect of housing status at probation start was robust to controlling for prior housing instability.

Conclusion and Implications:  Results indicate that housing instability and homelessness are risk factors for recidivism among this cohort of probationers. For practitioners, these findings suggest efforts to reduce recidivism should include increasing access to stable housing for individuals under community supervision.  For scholars, these findings suggest it is premature to dismiss socioeconomic factors broadly as irrelevant to recidivism and instead indicate that housing may deserve a place among other commonly accepted recidivism risk factors.