Do survey interviewers really have to be strangers?

A standard methodological assumption in social and demographic surveys is that interviewers should be strangers to respondents. Bias is feared should the interviewer and respondent know each other prior to the interview. We refer to this assumption as the “stranger-interviewer norm” and note that it has never been systematically tested. In order to subject it to scrutiny, we fielded an experimental survey in a town of the Dominican Republic countryside in the summer of 2010. We employed a mix of local and outsider interviewers and hired enough locals to produce an adequate number of “insider” interviews, that is, interviews where interviewer and respondent knew each other. Systematic randomization in our sampling design gives our survey the rigor of an experiment. In this paper we use these data to evaluate how the accuracy of responses varies by the level of familiarity between interviewer and respondent, as we were able to validate a number of survey questions by checking official documents that confirmed (or disconfirmed) respondents’ reports. At the time of this submission our analyses have failed to find support for the stranger-interviewer norm. There is instead some evidence that respondents are more cooperative with interviewers when these are not outsiders.