Abstract
This research used a couples analysis to test five models of partner abuse: (1) perpetrators' characteristics predict their abusive behavior, (2) victims' characteristics predict their victimization, (3) a dyadic process model in which both partners' characteristics independently and jointly predict couples' reciprocal abuse, (4) a traditional model in which men's characteristics alone predict both men's abuse and women's defensive reactions to it, and (5) an interactive model in which one partner's characteristics can moderate the other partner's abuse risk. We studied a representative sample of 360 young-adult couples comprising non-abusive couples, non-clinical abusive couples, and clinical abusive couples. As outcomes, we used measures of abuse perpetration, and as predictors we used partners' scores on negative emotionality, an indicator of violence-prone cognitions and emotions. Results were consistent with models 1, 2, and 3 for both males and females, but not with models 4 or 5. Findings applied to both non-clinical and clinical abusive couples. This study is the first to identify clinical abusive couples without selection bias (9% of the sample, having injury and/or official agency intervention), and to find that in such couples both sexes engage in abuse. Implications for primary prevention are that prevention programs should aim to: (a) target violence-prone cognitions and emotions, (b) reduce abusive behavior by both sexes, and (c) promote victim safety among both sexes. Implications for rehabilitation of already-abusive couples are: (d) treating only men will not reduce risk completely for most couples, and (e) more resources should be invested in researching couples treatments to reduce abuse. If replicated, the findings would suggest the need for policy that encourages development and evaluation of programs to reduce physical abuse by women.