Search

Understanding how and why physical intimate partner violence (IPV) persists in high-risk communities has proven difficult. As IPV is both sensitive and illegal, people may be inclined to misreport their views and experiences. By embedding a list randomization experiment (LRE), which increases respondent privacy, in a survey of 809 adult Arsi Oromo men and women in rural south-central Ethiopia, we test the reliability of direct questioning survey methods (e.g., used in the Demographic and Health Surveys) for measuring attitudes that underpin the acceptability of IPV. Participants were randomly assigned versions of the survey in which they were asked either directly or indirectly about the acceptability of wife-beating. By comparing responses across these surveys, we identify the extent to which views are being misreported using direct questioning methods, as well as identifying the "true" predictors of continued support for wife-beating. Indirect questioning reveals that almost one third of the sample believe that wife-beating is acceptable. Adults (particularly men) who are less educated (<3 years schooling) or living in households where women do not participate in economic decision making are among those most likely to identify wife-beating as justifiable (>50% endorsement). These individuals, however, are also more inclined to hide their approval when asked directly by an interviewer. That we find high but underreported support for wife-beating among some members of the community demonstrates a clear need to encourage a more open dialogue, to prevent violence toward women remaining undetected and thus unchanged. This finding also raises questions about the accuracy of traditional direct questioning for capturing information on IPV attitudes and norms. Of relevance to policy, we find that wife-beating is entirely absent only among adults with higher levels of education, living in households where decision making is shared between couples.