Measuring self-reported racial/ethnic discrimination in multicultural health surveys

Background: Although self-reported discrimination has been associated with numerous health problems, little work has validated discrimination measures across multiple racial/ethnic and language groups. The National Cancer Institute is collaborating with the California Health Interview Survey (CHIS) to develop a psychometrically validated and short instrument for telephone administration. CHIS is an ideal setting given the linguistic and ethnic diversity of the state. We report on our efforts to develop a measure of self-reported racial/ethnic discrimination.

Methods: Following a review of the literature, cognitive interviews and psychometric analyses of the Everyday Discrimination Scale were conducted. We then obtained feedback from an expert panel to develop a multidimensional instrument of self-reported racial/ethnic discrimination for field testing in CHIS 2007. Two versions were tested to evaluate whether it was more appropriate to reference the participant's race/ethnicity before asking about discrimination or after. A stratified subsample of respondents including African-Americans, Latinos/Hispanics, Asian Americans, American Indians/Alaska Natives, Whites and multiracial individuals were randomly assigned to receive one of the two 5-minute discrimination module versions. Five-hundred interviews were recorded for behavior coding.

Results: Our literature review identified four dimensions of discrimination most directly related to health outcomes and behaviors which we included in our field test instrument: 1) frequency, 2) recency, 3) stress appraisal and 4) response. Preliminary analyses of the Everyday Discrimination Scale revealed that the participants reported relatively infrequent experiences of discrimination, although there were ethnic differences in these patterns. Differential item functioning was found for several items by race/ethnicity and gender. Further, some items were redundant. Cognitive testing provided guidance on modifying item structure, word phrasing, and response categories. Our expert panel supported the comparison of two common—but divergent—measurement approaches to accurately eliciting reports of racial/ethnic discrimination. Currently, we have completed over 5,300 interviews in CHIS 2007. Depending on the specific item, between 16-46% and 23-53% of respondents reported recent and lifetime experiences of discrimination, respectively.

Discussion: We anticipate that our mixed-methods approach will yield a valid and reliable telephone-administered instrument to assess self-reported racial/ethnic discrimination for all major racial/ethnic groups. The results of our CHIS 2007 experiment will inform the development of a final instrument for administration in CHIS 2009. The 2009 discrimination data will be publicly released with CHIS. We anticipate that these data will contribute to improving our understanding of discrimination's impact on the health of the nation's diverse populations.