Predicting Abuse and Reassault Among Batterer Program Participants (From Violence Against Women and Family Violence: Developments in Research, Practice, and Policy, 2004, Bonnie Fisher, ed. -- See NCJ-199701)

• NCJ Number: 199730
• Author(s): D. Alex. Heckert, Edward W. Gondolf
• Date Published: July 2002
• Length: 12 pages
• Agencies: NIJ-Sponsored

Annotation

This study re-examined the prediction of abuse and recidivism for assault among batterer program participants by improving on previous research methodologies through the use of multiple outcomes, conditional variables, and a comprehensive multisite database.

Abstract

An extensive, multisite, longitudinal database of batterers and their female partners was used to test several possibilities for prediction, i.e., the utility of risk markers, conditional variables, risk instruments, and batterer types. The database also allowed for the examination of the dynamics of repeat assault and other alternative conceptions of violence. The database included intake interviews with 840 batterers and their partners, along with follow-up interviews every 3 months over 15 months. The follow-up response rate was 70 percent. The database encompassed four cities (Pittsburgh, Dallas, Houston, and Denver). The majority of the men (82 percent) were mandated to the programs by the courts, and the remainder voluntarily entered the program. The predictor variables were obtained from background questionnaires administered to the men and their partners at program intake. To assess the violent behavior of the men, the research team coded the women's descriptions of the violent incidents by using a sequential, situation conception of violence. The study found that the use of multiple outcomes apparently improved prediction through intake risk markers. The addition of conditional variables did not improve prediction, but it did identify important predictors. The items from the risk assessment instruments also modestly predicted the outcomes, but only Campbell's Danger Assessment Scale was more predictive than the women's perceptions by themselves. The strongest prediction occurred by entering risk markers as individual items and including women's perceptions, rather than combining them into a composite index. The study concluded that although more sophisticated models improve prediction, they still only modestly predict the outcomes and do not appear to be sufficient for clinical decisions by themselves. Implications of these findings are drawn for both researchers and for practitioners. 26 references