Endogenous social effects on intimate partner violence in Colombia

Intimate partner violence (IPV) is now recognized as a worldwide public health problem. Most theories ascribe IPV to individual, family, or cultural factors. Empirical support for these theories is inconsistent; however, community-level clustering of IPV behaviors is commonplace, indicating these behaviors are autocorrelated. In this study I use social interaction models and DHSIII data from Colombia to jointly model household and neighborhood effects on individual reports of beatings and forced sex. I find the probability a Colombian woman ever experienced IPV is dependent on her union status, schooling, number of live births, area of residence, and her partner's schooling. Net of these factors, the odds of ever being beaten were 64% higher for a woman living in a cluster where the proportion of other women reporting beatings exceeded the sample mean. These results show that endogenous social processes affect IPV levels and indicate the need for community-level IPV control efforts. (C) 2002 Elsevier Science (USA). All rights reserved.