Examining the Psychometric Properties of the Chemistry Self-Concept Inventory Using Rasch Modeling

The Chemistry Self-Concept Inventory (CSCI) is a widely used instrument within chemistry education research. Yet, agreement on its overall reliability and validity is lacking, and psychometric analyses of the instrument remain outstanding. This study examined the psychometric properties of the subscale and item function of the CSCI on 1140 high school chemistry students in the midwestern United States. Using exploratory factor analysis, we found that a five-factor model, distinct from previous studies' models, fit our data best. Using this model in Rasch analysis, we found several items within the Mathematics subscale that showed unusual item difficulty, significant misfit, and low item discrimination. Implications for future use of the CSCI are discussed.