Residual Covariance (R) Matrix

Figure 1. Residual covariance (R) matrix for unstructured covariance model
Residual covariance (R) matrix for unstructured covariance model
Figure 2. Residual covariance (R) matrix for autoregressive covariance model
Residual covariance (R) matrix for autoregressive covariance model

The R matrices show the greatest differences, but even these are not particularly large. The most notable differences are the correlations between weeks at least two weeks apart, and the variance of the fourth week.

Figure 3. Estimates of covariance parameters for unstructured model
Estimates of covariance parameters for unstructured model

Looking at the covariance parameter estimates for the unstructured model, you can see that the confidence intervals for the parameters which are most different include the values of the corresponding parameters from the autoregressive matrix. Thus, the simpler autoregressive model gives similar results to the unstructured model, so you can be very confident in preferring the autoregressive covariance model to the unstructured model.

Next