Evidence suggests that personalized invitations tend to increase response rates in web surveys (Cook et al., 2000). However, personalization may have an unintended impact on survey estimates. There is some evidence that personalization may reduce self-disclosure on sensitive items (Joinson, Woodley, & Reips, 2007) or increase socially desirable responding (Heerwegh et al. 2005), but other studies have been unable to replicate these findings (e.g. Heerwegh, 2005; Heerwegh & Loosveldt, 2006). To evaluate the impact of personalization on response rates and survey estimates of sensitive items, we compared the effects of personalized and generic greetings in a survey on an extremely sensitive topic: sexual experiences, including sexual assault victimization.
We conducted a Web survey with students at five universities. Sample members were randomly assigned to receive either a personalized greeting (“Dear John”) or a generic greeting (“Dear [Fill: School Name] Student”) in their survey invitation and reminders. Despite the literature suggesting personalization increases response rates, we hypothesized the personalized greeting would result in a lower response rate because our survey focused on such a highly sensitive topic. We also predicted personalization would result in lower rates of self-reported sexual assault victimization compared to a generic greeting. This is because we assumed sample members receiving the personalized greeting would perceive the survey as less anonymous, making them less likely to participate if they had experienced sexual assault victimization, or less likely to report their victimization experiences if they did participate.
We compared the effect of greeting on response rates and reported victimization. The personalized greeting resulted in a significantly higher response rate, but the generic greeting resulted in higher rates of sexual assault victimization; this difference is statistically significant for females. This experiment adds evidence to the divided literature on the effect of personalization on self-disclosure on sensitive items. Our findings suggest personalization increased response rates but decreased reported victimization, emphasizing that a higher response rate is not necessarily indicative of more accurate data.