The Effects of Prompt Interventions on Web Survey Response Rate and Data Quality Measures

Previous survey research has found that prompt interventions for speeding and straightlining were effective at reducing these undesirable response behaviors in web surveys. However, the effects of prompt interventions on data quality measures are mixed, and it is unclear how prompt interventions affect key survey estimates. We conducted an experiment on prompt interventions using the National Cancer Institute's 2019 Health Information National Trends Survey Push-to-Web Pilot Study. We used two types of prompts, one targeted speeding and the other targeted straightlining. We found no significant differences between the prompt and no-prompt conditions in overall web response rates. Also, we found that web respondents assigned to the prompt condition spent more time on the survey, had a lower percent of straightlining, and a lower percent of speeding on grid questions. Regarding key survey estimates, there were significant differences on estimates for one out of 40 items tested.