This project will test, with a randomized controlled trial, an intensive community supervision strategy for probationers and parolees who are at extremely high risk of committing homicidal acts.

The proposed project will test, with a randomized controlled trial, an intensive community supervision strategy for probationers and parolees in Philadelphia who are at extremely high risk of committing homicidal acts. The project is jointly funded by NIJ and the Smith Richardson Foundation with NIJ's contribution going primarily for project management costs involved in testing the forecasting of homicide risk.

Since 2005, the University of Pennsylvania and the Adult Probation and Parole Department (APPD) of the First Judicial District (covering Philadelphia County) have developed highly reliable ways of forecasting which among the APPD's 52,000 clients are most likely to kill someone or attempt to. With present forecasting tools, the highest-risk 1 to 2% of APPD clients have been observed to be 38 times more likely to be charged with homicide or attempted homicide within two years of beginning community supervision than the lowest-risk 60% of clients. Under community supervision by standard models, almost 12% of the highest-risk 2% of cases was charged with homicide or attempted homicide, while 32% of them were charged with serious violent offenses (attempted or completed) homicide, rape, robbery or aggravated assault.

The applicant states that randomly assigning 400 of these high-risk clients to experimental or control treatment groups (N = 200 each) will create an expected outcome of 24 cases in each group being charged with murderous conduct, and 64 cases charged with all serious violence. The sample is thus large enough to have an 80% probability of detecting as statistically significant (P = .05) a 12% or larger absolute difference (62% relative difference) between groups in serious violence charges, and a 7.6% or larger absolute difference (63% relative difference) in charges of murderous conduct.

The experimental program, the Strategic Anti-Violence Unit (SAV-U), has caseloads under 20 per officer, compared to 150 or more with standard supervision. SAV-U conducts diagnostic assessments of offenders' psychiatric needs, provides risk-based rehabilitation services, conducts weekly home visits to the offenders' homes, holds weekly Cognitive Behavioral Therapy (CBT) sessions in groups of 4 to 6, and provides intensive assistance and encouragement to offenders to complete schooling and obtain jobs.

While only CBT offers prior evidence of success in reducing recidivism, no one has ever tested any prevention strategy on such a high-risk group of highly harmful behavior. The proposed study is a test of a strategy that may not be optimal, but which is well within the capacity of most urban American community corrections agencies. The findings would be broadly applicable to some five million offenders under community supervision in the US of whom one percent lives in Philadelphia.

ca/ncf