The average predictive validity of intimate partner violence risk assessment instruments

Area of Science:

• Forensic Psychology
• Criminology
• Public Health

Background:

• Intimate partner violence (IPV) risk assessment is a rapidly developing field.
• Most research on the predictive validity of IPV risk assessment tools has emerged in the last decade.
• Several instruments exist to predict IPV recidivism and lethality.

Purpose of the Study:

• To examine the average predictive validity, weighted by sample size, of five standalone IPV risk assessment instruments.
• To compare the predictive accuracy of the Ontario Domestic Assault Risk Assessment (ODARA), Spousal Assault Risk Assessment (SARA), Danger Assessment (DA), Domestic Violence Screening Inventory (DVSI), and Kingston Screening Instrument for Domestic Violence (K-SID).

Main Methods:

• A meta-analysis approach was used, calculating the Receiver Operating Characteristic Area Under the Curve (AUC) for each instrument.
• Predictive validity was weighted by sample size across multiple validation studies.
• The analysis included 20 measures of predictive validity.

Main Results:

• The ODARA demonstrated the highest average weighted AUC (.666, k=5), followed by SARA (.628, k=6), DA (.618, k=4), DVSI (.582, k=3), and K-SID (.537, k=2).
• The overall effect size for average AUCs was small, with ODARA showing a medium effect size.
• Risk assessment instruments were administered correctly in only 45% of the included measures.

Conclusions:

• While ODARA shows the strongest predictive validity among the assessed tools, the overall effect size is small.
• The widespread use of proxy instruments and application in non-intended settings highlights current limitations.
• Factors beyond predictive validity, such as assessor skills and setting, are crucial when selecting an IPV risk assessment instrument.