January 1, 1999Structural Equation Modeling A Multidisciplinary Journal
Cutoff criteria for fit indexes in covariance structure analysis: Conventional criteria versus new alternatives
Authors
Discussion
Loading...
Member takes
Implication
This article evaluates conventional and new cutoff criteria for fit indexes in covariance structure analysis, impacting model adequacy assessment.
Key Points
- This research aims to assess the effectiveness of conventional cutoff criteria versus new alternatives for fit indexes in model evaluation.
- Utilized a 2-index presentation strategy combining standardized root mean squared residual and various supplemental fit indexes.
- Calculated rejection rates for diverse true-population and misspecified models.
- Evaluated cutoff values for multiple fit indexes based on the maximum likelihood estimation.
- Suggested cutoff values: .95 for TLI, BL89, CFI, RNI, and Gamma Hat; .90 for Mc; .08 for SRMR; and .06 for RMSEA.
- Indicated that the 2-index strategy is necessary to effectively reject certain true-population and misspecified models.
- Demonstrated that TLI, Mc, and RMSEA may overreject true-population models at small sample sizes.
Cite This Study
Hu et al. (1999) studied this question.