Race-of-Interviewer Effects Among Southern Adolescents

Abstract

Bias arising from the race-of-interviewer effect was found to occur among both black and white high school seniors living in the South. The magnitude of the effect was comparable to that observed among non-Southern adults. Such bias was limited to items mentioning the race of the interviewer, and the direction of bias was always toward deference to the interviewer's race. Low status respondents showed a sporadic tendency toward increased deference. The magnitude of the effect of the interviewer's race rarely exceeded 3 percent of variance explained. Bruce A. Campbell is Associate Professor of Political Science and Acting Director of the Survey Research Center at the University of Georgia. The author wishes to thank Susan Dukes for her able assistance. The data reported here were collected with the support from the National Institute of Education (NEG-00-3-0188). Public Opinion Quarterly Vol. 45 231-244 ? 1981 by The Trustees of Columbia University Published by Elsevier North-Holland, Inc. 0033-362X/81/0045-231/$2.50 This content downloaded from 207.46.13.57 on Fri, 09 Sep 2016 04:26:37 UTC All use subject to http://about.jstor.org/terms 232 BRUCE A. CAMPBELL goal is to measure the existence of race-of-interviewer effects in a sample of high school seniors, and to determine whether racial or interpersonal deference provides the most satisfactory explanation. The extent to which status amplifies the race-of-interviewer effect is also tested. In addition to this, there are five characteristics of the dataset that provide an opportunity to broaden our understanding of this potential impediment to validity in survey research. First, the respondents are southerners, a group that in spite of obvious reasons to suspect race-of-interviewer effects, has not been much studied in this literature over the past 20 years. Second, the respondents are adolescents: while Schuman and Hatchett (1974) found no effect for the interviewers age, it may be that deference is greater among adolescent respondents than it is among adults. Thus, the youth of the sample allows us to address the problem of race-of-interviewer effects as it may bear on research in adolescent socialization. Third, the effect of status is often obscured by the absence of upper-status blacks in small samples; the present dataset contains 116 upper-status black families. Fourth, the study assesses opinion on a wide variety of measures. Since Schuman and Hatchett (1974) demonstrate in their data that race-of-interviewer effects emerge only in connection with certain types of items, the presence of multiple criterion variables confers important advantages. Finally, a method using paper and pencil is tested in the face-to-face interview context, to see whether race-of-interviewer effects can be reduced.