Aim: To estimate an unbiased effect of commencing the EQUIPS Domestic Abuse Program (DVEQUIPS) on general re-offending and domestic violence (DV)-related re-offending.

Method: To deal with omitted variable bias, instrumental variables approaches are used to examine general and DV re-offending within 12 months free time after program referral for DVEQUIPS starters and offenders who were referred but did not start (non-starters). The program was delivered as a closed group which permitted the use of the peer non-commencement rate as an instrumental variable (IV), to exploit differences in participation based on the number of other offenders available to start the program at the time of an offender’s referral.

Results: Instrumental variables analyses indicated that selection bias was not a major problem, supporting the use of logistic regression to compare re-offending outcomes for starters and non-starters. The treatment effects estimated for offenders who started the program were not significantly different from zero for either general or DV-related re-offending within 12 months of referral.

Conclusion: There is no evidence of a treatment effect for those who start DVEQUIPS within 12 months of referral when compared with those offenders who were referred but did not start.