Sex, lies, and measurement: Consistency tests for indirect response survey methods

Abstract

Abstract Social scientists seeking to analyze socially sanctioned behaviors or attitudes increasingly rely on indirect response survey methods, meant to veil the answers of individual respondents. We propose simple internal consistency tests for two such methods, the list experiment and the randomized response technique (its Warner and Crosswise variants). We implement these tests in two studies on sexual and reproductive health behavior in Cameroon and Cote d’Ivoire. Non-compliance with instructions among surveyed individuals appears high and not easily char-acterizable. The tests we propose can be easily and cheaply embedded in measurement tools, allowing researchers to at least know whether their data is reliable before using it.