An Investigation of the Psychometric Properties of Full-Information Bifactor Primary and Domain Scores

• In Event: 25.082 - Fairmont Royal York Roundtable Session Three
  In Roundtable Session: 25.082-5 - Dimensionality and Estimation: Advancements and Applications in (Multidimensional) Item Response Theory and Factor Analysis

Abstract

The bifactor model is a multidimensional item response theory model capable of producing primary and domain-specific scores to characterize performance on the overall test and subscales. Empirical studies have predominantly focused on primary trait scores, with less consideration of domain-specific scores beyond a nuisance dimension. In response, this simulation study investigated the accuracy of bifactor primary and domain-specific scores under varying conditions. Bifactor primary dimension and unidimensional IRT (UIRT) scores were also compared. Bifactor primary and UIRT scores produced similar results when item slopes on the domain specific factors were small, with decreased accuracy of bifactor primary scores with medium to high item slopes on the domain factors. Across conditions, domain-specific scores were not accurate for score reporting purposes.

Authors

• Jason C. Immekus, University of Louisville

• Ruben Castaneda, University of California - Merced