Question Wording Matters in Measuring Frequency of Fear of Crime: A Survey Experiment of the Anchoring Effect

Abstract

For decades, fear of crime researchers have disagreed about how to best measure fear of crime. One approach proposed that measuring frequency of fear of crime within the past year has the highest validity. We argue that a frequency approach is vulnerable to the anchoring effect, in which participants base their numerical estimate on an available anchor. We conducted a survey experiment to test the effect of question wording on reported frequency of fear of crime. Participants were randomly assigned to report the number of times they felt fearful of crime within either a year, a month, or a week. There was also a fourth condition that asked a forced-choice question with many response options. They also reported the intensity of their most recent instance. We hypothesized that the year condition would yield lower frequency and higher intensity reports, followed by the month condition, and then the week condition. We did not find differences in intensity between conditions, but we found stark differences in frequencies between the year, month, and week conditions in the hypothesized direction. This is consistent with the anchoring effect: the word “year,” “month,” or “week” signaled an anchor to participants, and they adjusted their estimates from those anchors. These findings have important implications for measuring fear of crime and for survey methodology generally. We advise against asking about the frequency of fear of crime because such questions will lead participants to anchor and adjust. We also strongly caution researchers who wish to measure the frequency of other emotions, feelings, or behaviors.