Issues in estimating the prevalence of intimate partner violence: assessing the impact of abuse status on participation bias.

This study examines the potential implications of household interviews on participation bias for estimates of intimate partner violence (IPV). Using a variety of scales, IPV prevalence for the 135 women interviewed in a street-intercept survey was compared with the IPV prevalence of a subsample of these women who reported willingness to participate in a household survey with their partner in another room or when their partner also would be interviewed. A potential self-selection bias showed an 8% to 13% deflated prevalence of moderate to highly abused women and a 8% to 11% inflated prevalence of nonabused women among this subsample who would willingly participate in a household survey. Understanding who is included in household surveys, and who is not, is essential to compute and interpret national prevalence estimates for IPV.