The impact of respondent–interviewer familiarity and repeated survey participation on abortion reporting: Evidence from rajasthan, india

Abstract

CONTEXT: Researchers have long assumed that familiarity between an interviewer and a survey participant reduces the validity of responses, especially for such sensitive behaviors as abortion. However, little empirical evidence exists on this issue. METHODS: Data on 6,041 women aged 15–49 and 133 interviewers who took part in the second (2017) round of the Performance Monitoring and Accountability 2020 survey in Rajasthan, India, were used to examine the effect of interviewer–respondent acquaintance and participation in the prior survey round on women’s reporting of induced abortion. Associations were identified using multivariate, multilevel models that adjusted for respondent, interviewer and community characteristics, and that included interviewer random effects. RESULTS: On average, interviewers completed interviews with 41 respondents from their assigned cluster; they reported that they were acquainted with 61% of respondents and that 13% of respondents had participated in the prior survey round. Four percent of women reported having had an abortion. Neither interviewer–respondent acquaintance nor participation in the previous survey round was associated with abortion reporting in any of the multivariate models or in additional sensitivity analyses. CONCLUSIONS: The findings do not support the hypothesis that respondent familiarity with the interviewer or the survey process is associated with lower reporting of sensitive behaviors, like abortion. Future studies should further explore these and other design features to identify those that provide statistically significant improvements in the reporting of abortion and other sensitive behaviors.