Response rates remain low in phone surveys compared to face-to-face data collection (see here for a similar brief on response rates). This is especially true for random digit dial (RDD) or similar “cold call” phone surveys, which are necessary in the absence of a sample frame of reliable phone numbers. This brief presents early evidence from a series of experiments IPA conducted in 4 countries during 2020 to learn whether pre-survey messages, typically SMS texts, improve the rates at which respondents answer the phone, and complete the interview, with the ultimate goal of increasing the productivity of phone surveys. We find that, on average, SMS messages improve the rate at which respondents complete the survey relative to no message. This change is not driven by the rate at which respondents answer the phone, but by survey completion conditional on starting the survey. Random variation in message content had no significant effect on the rate at which respondents answer and complete phone surveys. The sample that resulted from the SMS message treatment had a composition that was on average younger and less likely to be predicted to be in poverty than the other treatment and control groups.

This paper was originally posted on the SSRN website.

Andrew Dillon, Steven GlazermanMichael Rosenbaum
Publication type: 
Phone Survey Methods Resource
April 30, 2021