The impact on subsequent violence of returning to an abusive partner

This paper examines the consequences, measured by subsequent violence, of returning to a previously abusive relationship. Two stage least squares estimation is used to control for the effect of prior violence on the probability of leaving, thus isolating the independent effect of leaving on later violence. The theoretically expected result is ambiguous: On the one hand, if aggressors view the attempt to leave as disobedience, then violence should increase upon return. If, instead, aggressors believe the temporary leave indicates that the victim will leave permanently (because she is unwilling to tolerate further abuse), then violence should decrease. Using data from the 1985 Physical Violence in American Families survey, the results show that victims who temporarily leave their abusers suffer increased violence relative to those who never leave. This paper contributes to the literature on domestic violence in several ways. First, the analysis explicitly models the endogeneity of the decision to leave. Second, the sample of abuse victims is nationally representative and includes both women who temporarily leave and women who never leave their abusers. Finally, the analyses test the prediction that the mere presence of shelters will decrease violence within intact relationships. While the results do not support this hypothesis, this may be due to the imperfect (state-level) measure of support services used.