Reliability generalizations of the Attitudes Toward Women Scale and the Racial Identity Attitudes Scales

The statistical technique, reliability generalization (RG), was employed to analyze and synthesize reliability coefficients from previous studies using the Attitudes Toward Women Scale (AWS) and Racial Identity Attitudes Scale (RIAS), which also included the Black Racial Identity Attitudes Scale (BRIAS), People of Color Racial Identity Attitudes Scale (PRIAS), and White Racial Identity Attitudes Scale (WRIAS). The purpose of this research was to investigate both measurement and sample characteristics that impact reliability coefficients across studies and sample populations, for better or worse. Cronbach's alpha, a specific type of reliability, was assessed in the current research. In the AWS-RG study, publication year and number of items were significant predictors of reliability and accounted for 26% of the total variance in reliability scores. The results indicate that reliability has deceased for the AWS over time, likely the result of item wording when the scale was written in contrast to spoken language, today. In the RIAS and related scales RG study, our results suggested that individuals who identified as Black produced significantly lower reliability scores than did individuals of other races on the Immersion/Emersion subscale. Additionally, examinees enrolled in undergraduate programs produced higher reliability scores than did those enrolled in or who had previously completed graduate school on the Pre-encounter/Conformity subscale. Overall, RIAS, BRIAS, PRIAS, and WRIAS produced scores with poor internal consistency, measured by Cronbach's alpha, in which future researchers may consider calculating reliability for the RIAS, BRIAS, PRIAS, and WRIAS using a different form of reliability. Findings from these studies can help assist researchers and clinicians in selecting appropriate instruments better suited for their population.