Fit Indexes: A Comparative Study between First-Order, Hierarchical, and Bifactor Confirmatory Factor Analysis Models in Self-Reported Measures

Author

Assistant Professor of Educational Psychology College of Education - Minia University

Abstract

  The current research aimed at comparing fit indexes between the measurement models from confirmatory factor analysis models as follows: the zero-order model which included the unidimensional model of the first order and the multidimensional model of the first order, the hierarchical model and the bifactor model. To verify this, the researcher selected three measurements that measure multiple aspects: a three- dimensional classroom goal structure scale, a four-dimensional self-esteem scale, and the five major factors of personality scale consisting of five dimensions. They were applied to a sample consisting of (894) participants distributed on the three scales; (178) for the five major factors of personality scale, (265) for the self-esteem scale, and (451) for the classroom goal structure scale. A series of confirmatory factor analysis models were conducted on each scale of the three models (First_order , hierarchical, and bifactor). Results from fit indexes that compare between the three models of the three scales revealed in general that the unidimensional model has lower fit indexes than the multidimensional model of the first order, and that the bifactor model has better fit indexes than zero-order and hierarchical models. Further research on bifactor modeling is recommended to investigate the general factor in personality and self-reported scales.

Keywords

Files

Share

How to cite

Statistics