Abstract
Gender-based violence (GBV) is widespread globally and has myriad adverse health effects but is vastly underreported. Few studies address the extent of reporting bias in existing estimates. We provide bounds for underestimation of reporting of GBV to formal and informal sources conditional on having experienced GBV and characterize differences between women who report and those who do not. We analyzed Demographic and Health Survey data from 284,281 women in 24 countries collected between 2004 and 2011. We performed descriptive analysis and multivariate logistic regressions examining characteristics associated with reporting to formal sources. Forty percent of women experiencing GBV previously disclosed to someone; however, only 7% reported to a formal source (regional variation, 2% in India and East Asia to 14% in Latin America and the Caribbean). Formerly married and never married status, urban residence, and increasing age were characteristics associated with increased likelihood of formal reporting. Our results imply that estimates of GBV prevalence based on health systems data or on police reports may underestimate the total prevalence of GBV, ranging from 11- to 128-fold, depending on the region and type of reporting. In addition, women who report GBV differ from those who do not, with implications for program targeting and design of interventions.