There is a frequently demonstrated test score gap between African Americans (blacks) and European Americans (whites) in the United States. Several explanations have been proposed for the black-white test score gap. These explanations include black-white differences in socioeconomic status and family structure, genetic endowment, school quality, culture, and racial bias in testing. The present study pertains to the last explanation which suggests that the black-white test score gap may partly be accounted for by systematic errors in measurement. Specifically, I examine whether or not the race of the examiner is a source of bias in measuring the black-white test score gap.
A key hypothesis is that the race of the examiner provides an alternate explanation for the poor test performance of blacks. If blacks perform less well on cognitive measures when tested by whites than by blacks, and if a large proportion of black respondents in a survey were tested by white rather than black interviewers, then an underestimation of blacks’ test scores is expected. Such a bias against black respondents due to a mismatch between the race of the respondent and the race of the interviewer would lead to an overestimation of the black-white test score gap.
In this study, I assess the effects of interviewer race on the test performance of black and white respondents, using data from the General Social Survey (GSS) which is a national household survey for adults aged 18 and over. Vocabulary testing in the GSS is conducted at the home of the respondent, and it involves face-to-face and one-on-one interaction between the respondent and the interviewer.

Data from the 1998 GSS indicate that black respondents, in comparison to white respondents, were much more likely to be tested by interviewers of a different race. For whites, interviewer race does not appear to affect test performance significantly. For blacks, however, there is a negative effect on test performance when tested by a white interviewer. Consequently, the black-white test score gap is substantially overestimated when the effects of interviewer race are not taken into account. For example, for black and white respondents with comparable socioeconomic background and years of schooling completed, the test score gap between black and white respondents tested by white interviewers was about two-fifths of a standard deviation unit in GSS vocabulary test scores. This gap, however, was nearly closed for black and white respondents who were tested by interviewers of their own race.

Stereotype threat seems a plausible explanation for the results observed in this study because white respondents were not negatively affected by having a black interviewer. According to the theory of stereotype threat, black students, when taking a test which claims to measure intellectual ability or when their racial identity is revealed in taking a test, face the threat of being judged by a negative societal stereotype about their group’s intellectual ability. The stereotype threat causes anxiety and worry, interfering self-consciousness, and overcautiousness, which impair the test performance of black students.

The Washington Post covered this research in its Feb. 2, 2009 article, “How a Self-Fulfilling Stereotype Can Drag Down Performance” on page A5.