An ecological model of intimate partner violence perpetration at different levels of severity.

Intimate partner violence (IPV) is a significant public health concern. This study proposed and tested an ecological model of both general and clinically significant (i.e., injurious or fear-evoking) IPV perpetration (IPVPerp). Risk and promotive factors from multiple ecological levels of influence (i.e., individual, family, workplace, community) were hypothesized to be important in the prediction of IPVPerp. Although clinically significant IPVPerp and general IPVPerp were hypothesized to relate, specific risks for clinically significant IPVPerp were hypothesized. U.S. Air Force active duty members and civilian spouses (N = 34,861 men; 24,331 women) from 82 sites worldwide completed the 2006 Community Assessment, an anonymous online survey assessing IPVPerp along with a variety of potential risk and promotive factors. Final structural equation models for men and women, cross-validated in holdout samples, clearly supported the relevance of an ecological approach to IPVPerp. Factors from all 4 levels were associated with both general IPVPerp and clinically significant IPVPerp, with relatively distal community and workplace factors operating via more proximal individual and family level variables (e.g., relationship satisfaction). The results suggest a variety of both established and novel potential targets for indirectly targeting general and clinically significant IPVPerp by improving risk profiles at the individual, family, workplace, and community levels.