Recruitment Intensity and Representativeness in Phone Surveys: Evidence from the Philippines

Abstract

When pandemic restrictions made face-to-face data collection too risky or costly in 2020, global poverty researchers started relying more heavily on mobile phone surveys, most often Computer Assisted Telephone Interviewing (CATI). These surveys provide several obvious benefits in terms of marginal cost and speed of deployment, but samples recruited via these methods may differ from the target population for several reasons (Gibson et al. 2017). To be included in a phone survey sample, respondent must own a phone, be able to pay for service, have reliable connectivity and power, and be willing to answer their phones. These barriers have been shown to bias the sample towards younger, more urban respondents with higher income. The choice of phone survey will thus introduce bias into the analysis when these factors are correlated with outcomes of interest.

Keywords: poverty, research methods, development economics