Framing violence: the effect of survey context and question framing on reported rates of partner violence

In this dissertation, I investigated two explanations for the variability in levels of partner violence found by large community surveys. In Study 1, I examined the effect of how questions about partner violence are introduced (question framing: conflict, violence-in-relationships, or attacks) on reports of partner violence. Although there was not a reliable effect of question framing, the pattern of findings was consistent across 3 of 4 analyses. Counter to predictions, an attacks question framing yielded highest reported levels of violence and a conflict framing yielded lowest levels. In Study 2, I examined the effect of the overall nature of the survey (survey framing: family life and relationships, personal safety, or crime) on reports of partner violence. There was a significant effect of survey framing; however, the direction of effects was counter to predictions with reported levels highest using a personal safety survey framing. To clarify these findings, in Study 3, I investigated the effect of survey form on reports of partner violence, with survey and question framing paired as typically seen in community surveys (family life and relationships survey framing with conflict question framing, personal safety survey framing with violence question framing, crime survey framing with attacks question framing). Although there was not a reliable effect of survey form, the pattern of rates and means was consistent across analyses. Participants reported highest levels of violence with the personal safety-violence form and lowest levels with the family life-conflict form. Overall, the findings suggest that although the nature of the survey used appears to influence reports of partner violence, other contextual factors likely also play a role.