Rasch Analysis of the Brief Symptom Inventory-18 with African Americans.

Abstract

Although the United States offers some of the most advanced psychological services in the world, not everyone in U.S. shares equally in these services, and health disparities persist when assessments do not appropriately measure different populations' mental health problems. To address this assessment issue, we conducted factor and Rasch analyses to assess the psychometric characteristics of the Brief Symptom Inventory-18 (BSI-18) to evaluate whether the BSI is culturally appropriate for assessing African Americans' psychological distress. The dimensional structure of the BSI was first identified and held up under cross-validation with a second subsample. The measure was unidimensional among African Americans. Our results also suggested minimal person separation, stability across subsamples, and little differential item functioning. Most African Americans identified themselves on the low end of the categories in a 0-4 rating scale, indicating their low endorsement of the items on the BSI. Rasch analyses were completed with the original scale but also collapsing the scale to three points, with some increase in separation and reliability for the collapsed scale. Differences in mean person position were found for mental health-related variables, consistent with hypotheses. Implications for theory and research on multicultural health scales are discussed as are effects of severe item skewness on analyses.