A central challenge to telephone surveys is low response rates. This is particularly true for random digit dial (RDD) surveys, which have especially low response rates. For researchers designing RDD survey protocols, there is a clear tradeoff between effort and composition, where surveys can achieve a higher response rate by calling fewer numbers repeatedly or by calling more numbers less intensively. This brief explores this tradeoff by measuring the effects of (i) repeated attempts per case, and (ii) rescheduling a call, on completion rates and sample composition. Using data from nine low- and middle-income countries (LMICs), we find that repeated and rescheduled attempts result in lower completion rates than new attempts, on average. However, the respondents who complete the survey in later attempts or after rescheduling have statistically significant differences in observable characteristics. This suggests that more call attempts may be needed to adequately represent the respondents who are harder to interview, even if those call attempts produce fewer completions per case.
This paper was originally posted on the SSRN website.