Statement of purpose The purpose of this work is to determine the circumstances in which low income African American women are most likely to report sensitive or risky behaviours associated with intimate partner violence.

Methods Consenting African American women receiving Special Supplemental Nutrition Program for Women, Infants, and Children (WIC) services in WIC clinics were randomised to complete an IPV screening (Revised Conflict Tactics Scales, Short-Form) via computer-assisted self-interview (CASI) or face-to-face interview (FTFI).

Results Among the women in this study (n = 616), reports of sensitive or risky behaviours were generally higher when engaging women face-to-face verses via computer interview. For example, those who engaged in risky behaviours were more likely exposed to intimate partner violence (IPV) over the course of their life (aOR: 1.59). Reported rates of risk exposure were high for tobacco usage (23%,25%), drug usage (6%,8%), and alcohol usage (50%, 51%). Mode of screening, but not interviewer race, affected disclosure of these behaviours. Women screened via FTFI reported significantly more tobacco (aOR: 1.8) and more alcohol (aOR: 1.5) usage than CASI- screened women.

Significance African American women in a WIC setting disclosed sensitive information more often in face-to-face than computer screening, and race-matching of client and interviewer did not affect disclosure. These findings put into question the argument that computer based interviews increase exposure of high risk behaviours. This evidence suggests that trained interviewers could enhance disclosure above computer based assessments. Previous findings have observed similar outcomes for IPV disclosure. This might suggest that there are types of computer assessments that are not effective, possibly due to miscalculated assumptions of computer and/or reading literacy. This might also suggest that the quality of the face-to face interaction might be contributing to the improved disclosure of sensitive information.