The accuracy of risk prediction for intimate partner violence

Research summary
Vol. 13 No. 1
January 2008


How well can risk assessments predict spousal assault recidivism?


To effectively manage men who assault their intimate partners, it is important to accurately assess their risk for future spousal violence. Risk assessments aid in the identification of men who require increased levels of security, supervision and treatment. The accuracy of identifying higher risk individuals is crucial to enhanced public safety and the safety of their partners in particular.

Various approaches have been used to assess which offenders are likely to reoffend. The existing risk assessment instruments vary in terms of what they were designed to predict (i.e., general reoffending, violent recidivism, or strictly spousal assault). Differences also exist in how they were developed (i.e., items selected based on theory, prior research, or data analysis), and in how the final evaluation of risk is determined (i.e., unstructured judgement, structured clinical judgement, or the actuarial mechanical summation of items). All of these various approaches, including victim (partner) ratings of risk, have been used to predict spousal assault recidivism.

The goal of the present investigation is to examine the level of accuracy of these various approaches.


A quantitative review was conducted of studies that examined the ability of risk assessments to predict spousal violence, or any violence (including spousal) recidivism of male offenders who were released following an index offence for spousal violence. A total of 18 different studies with a median sample size of 188 yielded 94 effect sizes for spousal violence recidivism and 28 effect sizes for any violent recidivism. The most frequently reported measures were the Spousal Assault Risk Assessment (SARA), the Dangerous Assessment (DA), and victim (partner) judgement.


The various approaches to predicting spousal assault recidivism showed, on average, moderate predictive accuracy. The structured tools specifically designed to assess spousal assault risk showed similar levels of accuracy as tools designed to predict general or violent recidivism and global assessments of risk provided by the female partners. The most accurate tools were those where the items were selected based upon prior research evidence.

Further research is needed to determine the extent to which tools specific to spousal assault provide useful information not included in the already well-established risk tools designed for general or violent recidivism.

Policy implications

  1. The various approaches to assessing spousal assault recidivism showed moderate levels of predictive accuracy. As is commonly observed, the instruments in which items were selected empirically were the most accurate.
  2. The empirically derived risk assessment instruments designed to predict general or violent offending did as well as those designed to predict spousal assault. There was insufficient research, however, to identify a specific tool that is more accurate than the others.
  3. Decisions concerning which risk assessment approach to use should be based on the specific purpose for the assessment. Some assessments focus on a victim's need for safety and assistance, whereas others may focus on the offender's imminent or future risk to reoffend. Each of these purposes may benefit from differing approaches and/or tools to risk assessment.


For further information

Guy Bourgon, Ph.D.
Corrections Research
Public Safety Canada
340 Laurier Avenue West
Ottawa, Ontario K1A 0P8
Phone: 613-991-2033
Fax: 613-990-8295

Date modified: