Pre-Survey Messaging to Improve Response: Appeals to Authority, Self-Interest, and Salience (WP-25-12)
Andrew Dillon, Steven Glazerman, Dean Karlan, and Christopher Udry
Contacting respondents via text messaging before survey administration is potentially a low-cost way to increase contact and response rates. In high-income countries, pre-survey messaging is often used to improve survey response via postcards, letters, advertisements, or gifts. For low-income countries, these strategies do not apply. The researchers conducted two experiments on pre-survey messaging. In the first, they randomized cases from Random Digit Dial (RDD) surveys in four countries (Colombia, Mexico, Philippines, and Rwanda) to receive messages that tested whether respondents better responded to surveys organized by researchers or government. In the second experiment, the researchers randomized pre-survey message content for second-round surveys of 7,000 respondents originally identified through RDD surveys in five countries: Burkina Faso, Colombia, Côte d’Ivoire, Rwanda, and Zambia. The content variations included information about survey participation compensation, key statistics from the survey’s previous round about food access and household finance, and general encouragements about survey participation. While pre-survey messages do increase response rates by 2 percentage points on average, they find no impact of message content on rates of contact, survey completion, composition of sample of respondents, or estimated study outcomes from the survey.