Are self-reported fertility preferences biased? Evidence from indirect elicitation methods

Desired fertility measures are routinely collected and used by researchers and policy makers, but their self-reported nature raises the possibility of reporting bias. In this paper we test for the presence of such bias by comparing responses to direct survey questions with indirect questions offering a varying, randomized, degree of confidentiality to respondents in a socioeconomically diverse sample of Nigerian women (N = 6,256). We find that women report higher fertility preferences when asked indirectly, but only when their responses afford them complete confidentiality, not when their responses are simply blind to the enumerator. Our results suggest that there may be fewer unintended pregnancies than currently thought, and that the effectiveness of family planning policy targeting may be weakened by the bias we uncover. We conclude with suggestions for future work on how to mitigate reporting bias.