Purpose. Research on batterer intervention programs (BIPs) has yet to establish more than a small effect in controlled studies (Babcock, et al, 2004). Most research to date has been done with single programs. We will report program completion and re-arrest data for an urban batterer intervention system composed of 30 BIPs with a common set of state standards, common program completion criteria, and centralized criminal justice supervision.

Method. Subjects are 899 men age 18 or older who pled guilty or were found guilty of a domestic violence crime, and ordered by the court to attend a BIP. The men completed a 2-hour psychosocial assessment and were referred to one of 30 BIPs for a state-mandated 24-week program. At an average of 3.4 years after the assessment, we retrieved program completion and re-arrest data from state and county databases. Following logistic regression models for both program completion and re-arrest, we model the effect of program completion on re-arrest, controlling for all significant correlates of re-arrest.

Results. 73% of the men completed their program, substantially higher than the conventionally reported rate of 50%. Predictors of program completion are income, drug abuse, alcohol use, primitive psychological defenses, prior arrests, ethnicity, and employment. Domestic violence re-arrest rates for the 27% of men who drop out is 37% compared to15% for completers, representing a small effect size (h=.31). Completing a batterer intervention program in this county system reduces the odds of being re-arrested for domestic violence by 63 percent, after controlling for age, employment, prior arrests, and alcohol abuse. A general estimating equation model controlling for within-agency clustering did not yield substantially different results.

Implications. This study supports the effectiveness of a court-based multi-program intervention system in reducing re-arrest for domestic violence in a large, urban, ethnically diverse sample of batterers who complete the program. The hypothesis emerging from our study is that the observed effects of batterer programs are a function both of individual BIPs, which comprise the system, and of the system itself. Given the relationship between substance use on completion and recidivism, our findings also suggest the importance of addressing these issues in batterer programs. Our study was limited by the fact that we could not randomly assign to treatment condition. Nevertheless, we feel that by examining outcomes across multiple programs, this limitation was mitigated to an extent. Future studies should take into account additional factors that might affect outcomes such as law enforcement, other elements of the court, victim service agencies, and behavioral treatment programs.