TABLE 5.
Results of structural equation models for cognition.
| Model | df | X2 | SRMR | RMSEA | TLI | CFI | Δdf | ΔX2 |
| No age model | 20 | 23.15 | 0.021 | 0.023 | 0.994 | 0.997 | ||
| Multi-group factor analysis | ||||||||
| Model 1. Paths, means (LV, IV, DV), variances free | 48 | 61.33 | 0.051 | 0.043 | 0.98 | 0.99 | – | – |
| Model 1 vs. Model 2 | – | – | – | – | – | – | 1 | 7.70** |
| Model 1 vs. Model 3 | – | – | – | – | – | – | 3 | 7.00 |
| Model 2. Paths, means (IVs, DV), variances free | 49 | 69.03* | 0.076 | 0.052 | 0.97 | 0.98 | – | – |
| Model 3. Paths, means (LV, DV), variances free | 51 | 68.33 | 0.055 | 0.048 | 0.98 | 0.98 | – | – |
| Model 3 vs. Model 4 | – | – | – | – | – | – | 1 | 0.14 |
| Model 3 vs. Model 5 | – | – | – | – | – | – | 1 | 8.15** |
| Model 4. Paths, mean LV, variances free | 52 | 68.47 | 0.055 | 0.046 | 0.98 | 0.98 | – | – |
| Model 4 vs. Model 6 | – | – | – | – | – | – | 1 | 8.06** |
| Model 4 vs. Model 7 | – | – | – | – | – | – | 10 | 19.21** |
| Model 4 vs. Model 8 | – | – | – | – | – | – | 11 | 16.45 |
| Model 5. Paths, mean DV, variances free | 52 | 76.48* | 0.083 | 0.056 | 0.97 | 0.97 | – | – |
| Model 5 vs. Model 6 | – | – | – | – | – | – | 1 | 0.05 |
| Model 6. Paths, variances free | 53 | 76.53* | 0.083 | 0.054 | 0.97 | 0.98 | – | – |
| Model 7. Variances, mean LV, free | 62 | 87.68* | 0.091 | 0.053 | 0.97 | 0.97 | – | – |
| Model 8. Paths, mean LV free | 63 | 84.92* | 0.133 | 0.048 | 0.97 | 0.98 | – | – |
| Model 8 vs. Model 9 | – | – | – | – | – | – | 10 | 19.60** |
| Model 8 vs. Model 10 | – | – | – | – | – | – | 6 | 8.44 |
| Model 9. Mean LV free | 73 | 104.52** | 0.127 | 0.054 | 0.97 | 0.97 | – | – |
| Model 10. Cognitive paths, mean LV free | 69 | 93.362* | 0.14 | 0.05 | 0.97 | 0.97 | – | – |
The first model is without age. The rest of the models are multiple group analyses testing measurement equivalence on different sets of parameters. The best fitting multiple group model as determined by ΔX2 is in italic.
SRMR, Standardized Root Mean Square Residual; RMSEA, Root Mean Square Error of Approximation; TLI, Tucker-Lewis Index; CFI, Comparative Fit Index; Δdf, change in df; ΔX2, change in X2; LV, latent variable; IV, Independent Variable; DV, Dependent Variable in this case latent cognition.
*p < 0.05, **p < 0.01.