Interviewer Effects on Racially-Sensitive Questions and Biomeasures: Results from a Semi-Interpenetrated Design

• Activity Number: 437 - Session in Honor of Jim Lepkowski's Retirement
• Type: Invited
• Date/Time: Wednesday, August 2, 2017 : 8:30 AM to 10:20 AM
• Sponsor: Survey Research Methods Section
• Author(s): Matt Jans* and Trivellore Raghunathan and Mick P. Couper
• Companies: Abt Associates and University of Michigan School of Public Health and Michigan Program in Survey Methodology
• Keywords: interviewer effects ; race-of-interviewer ; survey methodology ; interviewer variance

Abstract

Public health surveys often seek to measure racial disparities in health conditions, but estimates related to race can be clouded by the effects of interviewers. This paper focuses on the effect of interviewers generally (i.e., interviewer variance) and interviewer race on estimates of health characteristics in a survey of rural African Americans. Interviewers were assigned to respondents using a semi-interpenetrated design allowing estimates of interviewer variance to be mostly disentangled from geographical sample point effects. Questions in the survey (n = 292) were rated for their racial sensitivity by 6 independent reviewers who had a range of ethnic, racial, and linguistic backgrounds. Interviewer effects were assessed for racially-sensitive and racially-nonsensitive questions. Results show that interviewer race affects responses to a range of survey questions, particularly those that are racially-sensitive (e.g., reports of discrimination), but also those that are racially-nonsensitive (e.g., self-rated health). Surprisingly, it also affected blood pressure readings, suggesting a possible physiological link in the mechanism of interviewer effects.