Table 3. Confirmatory Factor Analysis Model Fit for the 3D-WFI (Study 2).
χ2 | df | Δχ2 | Δdf | p | CFI | TLI | RMSEA (90% CI) | SRMR | |
---|---|---|---|---|---|---|---|---|---|
Model 1: One factor | 5,892.49 | 135 | <.001 | .62 | .57 | .131 (.128, .134) | .12 | ||
Model 2: Two factors (Physical vs. Psychological Fatigue) | 3,639.27 | 134 | 291.95 | 1 | <.001 | .77 | .74 | .103 (.100, .106) | .08 |
Model 3a: Three factors | 963.68 | 132 | 294.19 | 2 | <.001 | .95 | .94 | .050 (.047, .053) | .04 |
Model 3b: Three factors with design-based correlated errors | 687.54 | 114 | 289.91 | 18 | <.001 | .96 | .95 | .045 (.042, .048) | .03 |
Note: N = 2,477. See text regarding correlated measurement errors. Two and Three factor models allowed the factors to correlate. CFI = comparative fit index; TLI = Tucker-Lewis Index; RMSEA = root mean square error of approximation; SRMR = standardized root mean square residual. Δχ2= Satorra-Bentler scaled chi-square difference test for robust maximum likelihood estimation each model to the prior model.