Talk or Text? Evaluating Response Rates and Sensitive Data Measurement: Evidence from a Randomized Controlled Experiment

Abstract

Researchers and policymakers face substantial challenges in selecting methods for conducting remote surveys, particularly when collecting sensitive information, reaching hard-to-access populations during emergencies, or working in vulnerable contexts. This study provides experimental evidence on the relative effectiveness of two survey methods: enumerator-led phone surveys and self-administered online surveys. In a randomized trial in El Salvador, we find that respondents assigned to phone surveys were, on average, 42 percentage points more likely to complete the survey, with particularly higher completion rates among women and older adults. Phone surveys also improved data quality and reduced the likelihood of respondents reporting high parental stress, without altering the composition of respondents. Although the direct cost of phone surveys is nearly double that of online surveys, the adjusted cost per completed response shows that phone surveys can be 25% less expensive when accounting for completion probabilities.

Keywords: Phone-based surveys, self-completion surveys, response rate, sensitive data, survey experiments