The Domestic Violence Safety Assessment Tool (DVSAT) and intimate partner repeat victimisation

Family violence Violence against women Victims of family violence Recidivism Intimate Partner Violence (IPV) New South Wales
Attachment Size
apo-nid143601.pdf 853.17 KB


Aim: To examine the predictive ability of the Domestic Violence Safety Assessment Tool (DVSAT) in determining a victim’s risk of intimate partner repeat victimisation.

Method: The study sample was 24,462 victims of intimate partner violence who were administered the DVSAT and recorded in the Central Referral Point (CRP) database between 1 January 2016 and 30 June 2016. Repeat victimisation was defined as experiencing a new domestic violence incident within 12 months of the index incident. The incident had to involve an intimate partner and result in a subsequent incident recorded in the CRP database. Two measures of repeat victimisation were defined: one included intimate partner related incidents overall (regardless of the type of incident); the other was limited to physical incidents of intimate partner violence, involving homicide, assault, sexual assault or robbery. Various predictors were examined, including the number of ‘yes’ responses to DVSAT items, with particular focus on ‘yes’ responses to 12 or more items, whether the victim had been involved in 2 or more DV incidents in the 6 months prior (a proxy for the repeat victim trigger), and being classified as ‘at serious threat’ (based on 12 or more ‘yes’ responses, the repeat victim trigger and/ or an officer’s professional judgement). Responses to individual DVSAT items were also examined. Predictive accuracy was assessed separately for female and male victims.

Results: Responding ‘yes’ to 12 or more items was associated with repeat victimisation but was a poor indicator in terms of discriminating those who experienced repeat victimisation from those who did not. The classification of ‘at serious threat’ was a better indicator of repeat victimisation than was responding ‘yes’ to 12 or more items, however, predictive accuracy was still poor. While some individual DVSAT items were predictors of repeat victimisation, many were weak predictors, and some, intended as indicators of increased risk of repeat victimisation, actually signalled a lower risk of this outcome.

Conclusion: This study highlights the importance of empirical validation when developing a risk assessment tool and provides evidence in support of further evaluation and review of the DVSAT.

Related Information

Assessing the impact of NSW’s Safer Pathway Program on recorded crime outcome…

Publication Details


Contemporary Issues in Crime and Justice no. 213