Gender-biased clustering of attitudes towards physical intimate partner violence: A social network analysis in south-central Ethiopia.

Changing social norms, shared beliefs about what is acceptable, is a key focus of global health campaigns aimed at ending intimate partner violence against women (IPVAW). In Ethiopia, it is estimated that one in four women have been assaulted by a male partner, and over half the population hold attitudes supportive of this form of violence. To date, efforts to change people's attitudes towards IPVAW have been hindered by uncertainty over how social norms are acquired or learned. Here, we consider whether people's acceptance of IPVAW is maintained through social influence or "contagion," using large-scale sociocentric social network data from 5,163 Arsi Oromo farmers in south-central Ethiopia. Bayesian analyses reveal that IPVAW attitudes cluster within social networks. People are more likely to accept IPVAW if the people they chat to, respect, or live with do too. However, exploration of the relationships between social ties indicates that this effect is gender stratified, i.e. driven by same-gender connections. Meanwhile, having IPVAW-accepting social ties of the opposite gender is predictive of a person rejecting IPVAW. Our results indicate that transmission paths may exist among social ties of the same gender: between friends and neighbors, from key respected community figures, and within and beyond households. This suggests that IPVAW prevention interventions that seek to target men