Date of Award
Quantitative Research Methods
Reliability generalization (RG) is a meta-analytic method that aims to assess the variability of test score reliability across studies and identify the sources of this variability. In this study, a reliability generalization analysis was performed on studies of the Brief Symptom Inventory-18 (BSI-18) to examine the variability in Cronbach's alpha reliability estimates reported in the literature. This inventory was chosen because of its extensive use in counseling and medical settings and documented reliability and validity. The database that was consulted to collect articles was PsycInfo. The reported Cronbach's alphas were obtained to assess whether defined moderator variables affected reliability estimates. Out of the 161 references located, 48 studies met the selection criteria. For the Global Severity Index (GSI), the mean reliability was 0.91, 0.77 for the Somatic subscale, 0.85 for the Depression subscale, and 0.83 for the Anxiety subscale. The moderator analyses led to a predictive model where the type of population (clinical vs. nonclinical) for the GSI, and gender for the Somatic subscale were significant. Finally, clinical implications of the results are discussed.
Alzahrani, Dareen Taha, "A Reliability Generalization Study of the Brief Symptom Inventory-18" (2016). Electronic Theses and Dissertations. 1175.
Recieved from ProQuest
Dareen Taha Alzahrani