Logarithmic Level 1 Model Testing Group and Group by Time Interactions for AS Groups Using Only Six Time Periods
| Model Statistic | Estimate |
|---|---|
| Fit statistic (−2RLL) | 221.21 |
| AIC statistic | 239.21 |
| Repeated-measures error structure | |
| Autoregressive variance estimate | 0.04 (SE = .003, p < .001) |
| Autoregressive covariance estimate | 0.77 (SE = .27, p = .004) |
| Random coefficients variance matrix | |
| Intercept random variance | 0.11 (SE = .02, p < .001) |
| Intercept-change random covariance | −0.009 (SE = .006, p = .12) |
| Change parameter random variance | 0.01 (SE = .003, p < .001) |
| Random coefficients (Level 1 parameters) | |
| AS NMI intercept estimate | −0.16 (SE = .05, p = .001) |
| AS NMI change parameter estimate | −0.06 (SE = .02, p = .001) |
| Group intercepts effect (overall) | |
| AS MI difference from AS NMI intercept | 0.29 (SE = .07, p < .001) |
| Group by time interaction | |
| AS MI difference from AS NMI change | −0.02 (SE = .02, p = .40) |
All models (as in Table 2) were re-estimated for the AS groups with six time periods; the logarithmic model was still the best fitting model.