Social norms and women's risk of intimate partner violence in Nepal.

Social norms increasingly are the focus of intimate partner violence (IPV) prevention strategies but are among the least examined contextual factors in quantitative violence research. This study assesses the within-community, between-community, and contextual effect of a new measure of social norms (PVNS: Partner Violence Norms Scale) on women's risk of IPV. Data come from baseline surveys collected from 1435 female, married, reproductive-age participants, residing in 72 wards in three districts (Chitwan, Kapilvastu, Nawalparasi) in Nepal who were enrolled in a cluster randomized trial testing the impact of a social behavioral change communication intervention designed to prevent IPV. Results of unconditional multilevel logistic regression models indicated that there was cluster-level variability in the 12-month prevalence of physical (ICC = 0.07) and sexual (ICC = 0.05) IPV. Mean PVNS scores also varied across wards. When modeled simultaneously, PVNS scores aggregated to the ward-level and at the individual-level were associated with higher odds of physical (ORind  = 1.12, CI = 1.04, 1.20; ORward  = 1.40, CI = 1.15, 1.72) and sexual (ORind  = 1.15, CI = 1.08, 1.24; ORward  = 1.47, CI = 1.24, 1.74) IPV. The contextual effect was significant in the physical (0.23, se = 0.11, t = 2.12) and sexual (0.24, se = 0.09, t = 2.64) IPV models, suggesting that the ward-level association was larger than that at the individual-level. Adjustment for covariates slightly attenuated the ward-level association and eliminated the contextual association, suggesting that individual perceptions and the collective community phenomena were equally strong predictors of women's risk of IPV and should be taken into consideration when planning interventions. PVNS is a promising measure of social norms underpinning women's risk of IPV and warrants further psychometric testing.