Impact of Risk Assessment on Recidivism Rates in Intimate Partner Violence - CONCEPT Professional Training

Kirk R. Williams, University of California, Irvine
Richard Stansfield, Rutgers University—Camden

To manage intimate partner violence (IPV), the criminal justice system has turned to risk assessment instruments to predict if a perpetrator will reoffend. Empirically determining whether offenders assessed as high risk are those who recidivate is critical for establishing the predictive validity of IPV risk assessment instruments and for guiding the supervision of perpetrators. But by focusing solely on the relation between calculated risk scores and subsequent IPV recidivism, previous studies of the predictive validity of risk assessment instruments omitted mediating factors intended to mitigate the risk of this behavioral recidivism. The purpose of this study was to examine the mediating effects of such factors and the moderating effects of risk assessment on the relation between assessed risk (using the Domestic Violence Screening Instrument-Revised [DVSI-R]) and recidivistic IPV. Using a sample of 2,520 perpetrators of IPV, results revealed that time sentenced to jail and time sentenced to probation each significantly mediated the relation between DVSI-R risk level and frequency of reoffending. The results also revealed that assessed risk moderated the relation between these mediating factors and IPV recidivism, with reduced recidivism (negative estimated effects) for high-risk perpetrators but increased recidivism (positive estimate effects) for low-risk perpetrators. The implication is to assign interventions to the level of risk so that no harm is done.

intimate partner violence, risk assessment, risk management, punishment, recidivism

“The present research … analyz[es] the interrelations between risk assessment (in this case using the DVSI-R), risk management strategies related to court orders for incarceration and/or supervision, and IPV recidivism during an average 18-month follow-up period postassessment. The present research seeks to replicate some aspects of prior work, but it focuses on the impact of risk management decisions on recidivism by exploring how sentences imposed for jail or probation mediate the risk-IPV recidivism relation” (p. 345).

“The first objective was to determine how much of the estimated effect of risk assessment on recidivism was mediated by risk management strategies … The second objective of the present research was to assess the extent to which the level of assessed risk moderates the relation between risk management strategies and recidivism” (p. 349).

“‘[H]igher-risk offenders need more intensive and extensive services if we are to hope for a significant reduction in recidivism. For the low-risk offenders, minimal or even no intervention is sufficient.’ This assumption implies that the relation between risk assessments and risk management strategies is a monotonic, linear function: the greater the assessed risk, the greater the intensity of interventions … more intense interventions were associated with an increase in recidivism for low-risk offenders, but they were associated with a decrease in recidivism for high-risk offenders” (p. 346).

“Higher risk perpetrators were likely to be sentenced to more days on probation or in jail, in addition to being more likely to recidivate. In short, the greater the risk level, the greater was the seriousness of sentencing (probation or jail time), and the greater was the frequency and likelihood of IPV recidivism … That is to say, greater time sentenced to probation or jail was more likely to be associated with IPV recidivism among low-risk perpetrators than high-risk perpetrators … the relation for low-risk perpetrators is positive, but it is negative for high-risk perpetrators, meaning that as the likelihood of any sentence for jail time increased, the probability of rearrest for IPV recidivism increased for low-risk perpetrators but decreased for high-risk perpetrators” (pp. 349-351).

“[R]isk management strategies are typically recommended for IPV perpetrators, based on the assessment of risk, with the intent of altering their behavior … the overall total effect of risk assessment on recidivism is a combination of the direct effect of risk assessment on recidivism and the indirect effects of risk assessment on recidivism as mediated by risk management strategies. The interrelations between risk assessment, risk management, and recidivism … will influence the nature of the bias in estimating the total effect. Hence, formally incorporating risk management strategies as mediators of the relation between assessed risk and actual subsequent recidivism becomes vital to delineate empirically the overall relation between risk and recidivism and to account for any confounding influences” (p. 346).

“Virtually all previous studies of IPV risk assessment, particularly those designed to determine the predictive validity of risk assessment instruments, focused on the relation between calculated risk scores and subsequent reoffending. As discussed earlier, by focusing solely on this relation, previous studies of the predictive validity of risk assessment instruments, including the DVSI-R, omitted an integral component of the RNR model—risk management. This omission increases the chances that the estimated relation between those scores and recidivism will be contaminated by what happened to supervised offenders as a result of risk management efforts” (p. 348).

“The results also revealed that assessed risk moderated the relation between risk management strategies and IPV recidivism, with low-risk perpetrators worse off following a recommendation of probation time or jail time” (p. 351).

“The “risk principle” assumes that criminal and violent behavior can be predicted. More importantly, this principle further assumes the intensity of interventions should be aligned with the risk level of perpetrators. Their greater need means that programs serving a greater percentage of high-risk offenders may be more effective than programs focused on individuals with a lower risk for reoffending” (p. 344).

“The core of the need principle is to identify criminogenic needs and align treatment to those needs. The principle of responsivity requires that programs also match their services to the general and specific ways that offenders respond best to services: cognitive-behavioral and social learning approaches, in general, and individual characteristics and preferences, in particular” (p. 344).

“[S]entencing outcomes, such as probation and jail, may place low-risk and high-risk perpetrators in close contact with each other, thus facilitating the transmission of proviolence normative beliefs. Further, such outcomes may undermine many of the characteristics rendering low-risk perpetrators as low-risk—especially the prevalence of prosocial activities and individuals in their lives … imposing interventions on low-risk perpetrators that are better suited for high-risk perpetrators violates a principle of the RNR model: ‘match intensity of services to the risk level of cases’” (p. 351).

As always, please join the discussion below if you have thoughts or comments to add!