Family violence risk assessment: A predictive cross-validation study of the Domestic Violence Screening Instrument-Revised (DVSI-R).

Abstract

This research was a cross-validation study of the Domestic Violence Screening Instrument-Revised (DVSI-R), using a diverse, statewide sample of 3,569 family violence perpetrators in Connecticut, assessed in February and March of 2007. It analyzed re-arrest data collected during an 18-month period post assessment. Three issues were central, which have been ignored in previous research on family violence risk assessment: (1) analyzing five refined measures of behavioral recidivism, (2) determining whether perpetrator characteristics and types of family and household relationships (beyond just heterosexual intimate partners) moderate the empirical relations between the DVSI-R and the behavioral recidivism measures, and (3) determining whether structured clinical judgment about the imminent risk of future violence to the victim or to others corresponds with recidivism predicted by the DVSI-R total numeric risk scores. The empirical findings showed that the DVSI-R had significant predictive accuracy across all five measures of recidivism. With one exception, these relations did not vary by gender, age, or ethnicity; and again with one exception, no significant evidence was found that types of family or household relationships moderated those empirical relations. In short, the evidence suggested that the DVSI-R was a robust risk assessment instrument, having applicability across different types of perpetrators and different types of family and household relationships. Finally, the empirical findings showed that structured clinical judgment about imminent risk-to-victim and risk-to-others corresponded with the prediction of recidivism by the DVSI-R total numeric risk scores, but the effects of those scores were significantly stronger than the perceived risk-to-victim or the perceived risk-to-others.