January 1, 1990Psychological Bulletin
Comparative fit indexes in structural models.
Authors
Discussion
Loading...
Member takes
Implication
Comparative fit indexes evaluate model fit, highlighting improved estimations for structural models.
Key Points
- The aim is to propose new fit indexes to better evaluate the performance of structural models compared to existing methods.
- Introduction of a new coefficient summarizing the reduction in noncentrality parameters.
- Development of new normed (CFI) and nonnormed (FI) fit indexes from two estimators.
- Generalization of existing fit indexes to include Wald and Lagrange multiplier statistics.
- CFI demonstrates improved accuracy for small samples compared to Bentler and Bonett's normed fit index (NFI).
- FI linearizes the performance of the non-normed fit index (NNFI), preventing extreme fitting errors.
- All proposed fit indexes show equivalence under correct specifications and perform well across various sample sizes.
Cite This Study
Peter M. Bentler (1990) studied this question.