Agree-response bias can inflate prejudice scale scores artificially

Evidence

Social desirability biases in measures of racial attitudes have been reasonably well documented. But do they normally occur in sufficient magnitude to threaten the main findings of survey research? Race-of-interviewer comparisons show that Whites express more liberal attitudes to Black than to White interviewers. But relatively few Whites are interviewed by Blacks. For example, in the 1986 NES survey that is often analyzed in studies of racial attitudes, only 4 percent of the White respondents had non-White interviewers. Social desirability pressures could lead some Whites to express no opinion at all rather than expressing their true prejudices (Berinsky, 2004). But in practice relatively few Whites fail to answer questions on prejudice or racism (e.g., in the 1986 NES, an average of only 3 percent of White respondents gave a “no opinion” response to the symbolic racism and old-fashioned racism items). At this point, the existing evidence does not indicate that such biases are powerful enough to render survey measures invalid, though clearly more research is needed. — , Rudman, Laurie A - On the Nature of…, 2200 (#complicates, via linked claim)

Categorization

• XREF: connects to: Vague survey wording can make racial prejudice undetectable in response data; Measuring stereotypes indirectly reveals prejudice explicit surveys miss — Anthony G. Greenwald and Mahzarin R. Banaji, Implicit Social Cognition: Attitudes,…, 194; Survey correlations between prejudice scales may be inflated by response style, not real relationships — John T. Jost and Jim Sidanius, Political Psychology, 1044 (+4 more)

Broader Connections

• Social desirability bias does not measurably invalidate prejudice survey measures #complicates (0.90) — , Rudman, Laurie A - On the Nature of…, 2200