Assessing differential item validity of the AIDS-related social skills questionnaire among African adolescents

Using differential item validity (DIV) methodology, this study investigated whether the items of the AIDS-Related Social Skills (ASAS) questionnaire had the same interpretation by gender and country. Polytomous logistic regression was used to test if the coefficients for individual items were the same by gender and country. The subjects (n = 1133) were Black Anglophone African 10th-grade students from Nigeria (n = 396, Kenya (n = 280). Zimbabwe (n = 319). and Sierra Leone (n = 138). The analyses showed that many items had significant DIV, Because 12 of the 30 test items (i.e.. 40%) showed significant DIV. it suggested that the ASAS may not be internally valid by gender and across countries. Further qualitative work is needed to understand the extensive DIV that was statistically found in this study. (C) 2001 Academic Press.