Intimate partner violence prevalence estimation using telephone surveys: understanding the effect of nonresponse bias.

To assess the effect of nonresponse bias in telephone prevalence studies of intimate partner violence, the authors asked women visiting a health center in Albany, New York, during 1998 about their willingness to participate in telephone surveys. Women physically victimized by a male partner were more likely than other women to say they would participate in telephone surveys (66.7% vs. 44.4%, p = 0.03). Among women severely victimized, those living with their partner were less willing to participate than those not cohabiting (45.5% vs. 91.7%, p = 0.03). Including questions about willingness to participate in telephone surveys in studies of other kinds may be a useful method of identifying nonresponse bias.