Table 2.
Tests of measurement invariance across time
| χ2 | df | CFI | TLI | RMSEA | Model comparison | SB Δχ2 | Δdf | |
|---|---|---|---|---|---|---|---|---|
| LMX-Contribution | ||||||||
| Model 1: Configural invariance | 9.89 | 6 | .99 | .98 | .04 | – | – | |
| Model 2: Weak invariance (loadings) | 11.18 | 8 | 1.00 | .99 | .03 | 2 vs. 1 | 1.27 | 2 |
| Model 3: Strong invariance (loadings, thresholds) | 11.61 | 10 | 1.00 | 1.01 | .02 | 3 vs. 2 | 0.24 | 2 |
| Model 4: Strict invariance (loadings, thresholds, uniquenesses) | 14.89 | 13 | 1.00 | 1.01 | .02 | 4 vs. 3 | 3.29 | 3 |
| AOC | ||||||||
| Model 1: Configural invariance | 186.76* | 47 | .94 | .92 | .08 | – | – | |
| Model 2: Weak invariance (loadings) | 196.96* | 52 | .94 | .92 | .08 | 2 vs. 1 | 4.94 | 5 |
| Model 3: Strong invariance (loadings, thresholds) | 205.70* | 57 | .94 | .93 | .08 | 3 vs. 2 | 7.86 | 5 |
| Model 4: Strict invariance (loadings, thresholds, uniquenesses) | 206.70* | 63 | .94 | .94 | .07 | 4 vs. 3 | 2.01 | 6 |
Full information maximum likelihood estimation was used. df, degrees of freedom; CFI, comparative fit index; TLI, Tucker-Lewis index; RMSEA, root mean square error of approximation; SB, Santorra-Bentler scaled
*p < .05