Is it time to put a moratorium on list experiments for domestic violence elicitation?

Using data from over 24,000 respondents in the Norwegian Crime Victimization Survey, we conducted a double list experiment to measure domestic violence (DV). Both list experiments revealed a statistically significant decrease in reporting when including a sensitive DV item. This clear violation of the “no design effects” assumption is not only explained by floor effects. One possibility is that the results indicate a "fleeing" behavior whereby respondents try to avoid association with DV. Combined with the inherent power limitations of list experiments in many contexts, these results underscore the need for caution in employing list experiments to measure DV, even in large samples.