Rasch analysis of the Outcome Questionnaire with African Americans.

Effectively diagnosing African Americans' mental health with a single intake checklist has posed an unresolved challenge, as most intake checklists were developed from White perspectives. In this study, Rasch analysis was used to assess the psychometric characteristics of a common measure of clinical distress, the Outcome Questionnaire (OQ; Lambert, Lunnen, Umphress, Hansen, & Burlingame, 1994), for a sample of African American students split into a calibration and validation subsample. OQ subscales were first identified and were then held up under cross-validation with a second subsample. Rasch analysis of the OQ clearly indicated the measure was multidimensional among African American students with 2 subscales titled Well-Being and Psychological Distress. Our results also indicated appropriate response scale use, adequate person separation, strong stability across subsamples, and little differential item functioning. Moreover, our analysis showed items of the 2 subscales to be well-targeted for African American students. However, if subscales were to be revised for African American students, some items at the same logit position might be deleted and replaced with either very easy or more difficult items or with items at intermediate positions to extend and to fill in gaps in construct coverage. Implications for theory and research on multicultural mental health scales were discussed.

Development and structural validation of a shortened version of the Participation Scale.

PURPOSE: To validate a shortened version of the Participation Scale (P-scale) that will be quicker to use and to describe the factor structure found in the P-scale data in various study samples. METHODS: A large multi-country and multi-cultural database was compiled consisting of 5125 respondents. Item analysis, explanatory factor analysis and confirmatory factor analysis were applied to identify items for deletion and investigate the factor structure of the P-scale. RESULTS: The multi-country database included 11 databases from six different countries. Respondents were affected by a range of health conditions, including leprosy, HIV/AIDS, dermatological conditions and various disabilities. Of the respondents included 57% were male. The P-scale Short (PSS) contains 13 items. A two-factor structure, with factors named "work-related participation" (three items) and "general participation" (10 items), showed the best model fit (Comparative Fit Index = 0.983, Tucker Lewis Index = 0.979, Rooted Mean Square Error of Approximation = 0.061). The Cronbach's alphas were very good for both the whole scale and the subscales, 0.91, 0.83 and 0.90, respectively. Correlation between the two factors was high (r = 0.75) indicating that interpreting the P-scale as measuring an overall factor "participation" is still valid. A very high correlation (r = 0.99) was found between the full P-scale and the PSS. CONCLUSIONS: The findings suggest good validity of the P-scale across a range of languages and cultures. However, field testing needs to confirm the validity of the PSS to measure the level of social participation restrictions across cultures and health conditions.