Bifactor Models in Psychometric Test Development
Fangfang Chen, Zugui Zhang · 2018
This chapter focuses on two application areas of bifactor models in psychometric test development: its suitability for testing and measurement of multidimensional constructs and its potential for addressing conceptual debates in psychological research. Within confirmatory factor analysis (CFA), two alternative models, bifactor models and second-order model, represents the factor structure of items that assess several related domains that are hypothesized to comprise a general construct. The chapter compares the similarities and differences between the two models and proposes the potential advantages of bifactor models over second-order models when researchers have an interest in the predictive power of the general construct as well as the specific domains. It then presents the bifactor models within the framework of exploratory factor analysis (EFA) and examines the conceptual differences among three EFA estimation methods: bifactor rotations, target rotations, and Schmid-Leiman transformations.