The effect of the sex of interviewers on the quality of data in a Nigerian family planning questionnaire

Abstract

Traditionally, female interviewers have been preferred to men for conducting fertility and family planning surveys. However, in West Africa, evidence for their superiority over male interviewers is mixed. In Nigeria, as part of a four-state pretest of the national family planning questionnaire, an experimental design was incorporated to quantify effects of the sex of the interviewer. In one state, reinterviews were also performed to measure the reliability of responses. In the conservative northern state of Kano, the use of male interviewers was problematic. However, in the other three states, only weak evidence was found to mitigate against the use of male interviewers. In fact, in two states, the proportion of respondents reporting knowledge of several contraceptive methods was significantly higher when male interviewers conducted the survey. Respondents' reports of contraceptive use were very unreliable for interviewers of both sexes.

PIP: While it has been generally assumed that survey questions directed to women about sensitive topics such as contraceptive usage and sexual behavior must be asked by female interviewers, several factors, such as the lack of trained women, serve to make such surveys more expensive than those conducted by men. To determine whether data quality is affected by the sex of the interviewer, therefore, an experimental design was embedded into a four-state pretest of a family planning questionnaire in Nigeria. In addition, reinterviews were performed in 33 of the 39 original enumeration areas of one state. Of those originally interviewed by women, half were reinterviewed by men and half by women. The same pattern was followed for those originally interviewed by men. The most significant differences by sex of interviewer were found in the conservative state of Kano where the respondent's partner or other men were present for 79% of the interviews. The women frequently refused to answer questions about sexual intercourse and the durations of male-conducted interviews were significantly longer, but women in Kano were more likely to refuse to tell a woman interviewer about current contraceptive usage. In Imo and Kwara states, female interviewers elicited more accurate (higher) numbers of deceased children, but the number of spontaneous reports of knowledge of contraceptive methods was higher for males. However, in Kwara State, reports of knowledge and use of contraceptives were least reliable when men conducted both the initial and the reinterview. The present study is limited by its inability to control for other characteristics of the interviewers and by the fact that reinterviews were only conducted in one state. However, results from the reinterviews suggest that respondents are highly unreliable in their reporting of family planning use, regardless of the sex of the interviewer. Small reinterview surveys should be included as part of large-scale demographic surveys in developing countries to determine response variability of key measures.