Year of Publication

2014

Degree Name

Doctor of Philosophy (PhD)

Document Type

Doctoral Dissertation

College

Education

Department

Educational, School, and Counseling Psychology

First Advisor

Dr. Pamela L. Remer

Abstract

The purpose of this study was to revise the Awareness of Privilege and Oppression Scale (Montross, 2003) and to improve upon the psychometric properties of the original instrument. The APOS-2 is a diversity training outcome measure that is designed to measure the social justice-related construct awareness of privilege and oppression. I retained 26 items from the original APOS (Montross, 2003) and utilized an expert focus group to generate new test items for the APOS-2. Feedback from an expert rater group was solicited and then incorporated into the APOS-2 to help reduce the number of items, improve item content, and evaluate content validity. The newly revised scale was then administered to a combined sample of 484 undergraduate students at a large public university through an internet-based survey. Item-analysis procedures and an exploratory factor analysis (EFA) with direct oblim oblique rotation were utilized to further reduce the number of items and then determine the psychometric properties of the final solution. The EFA of the APOS-2 data provided support for the theoretical four-factor solution. The observed Cronbach alpha reliability estimates for the final 40-item total score and subscale scores were as follows: Total score (.92), Awareness of Heterosexism (.84), Awareness of Sexism (.73), Awareness of Classism (.84), and Awareness of Racism (.86). The APOS-2 correlated low and positively (r = .29) with a measure of openness to diversity and negatively and close to zero (r = -.10) with a social desirability measure. These collective data suggest the APOS-2 may be a viable alternative to the original APOS with a stronger initial effort to link item content to the extant literature, improved subscale reliability estimates, continued support for the use of the theoretically derived subscales, and a predictable relationship with measures of convergent and discriminant validity.