24.3. The end of this section might be a natural place to compare level-2 variance parameters. Consider computing percent-of-total-variance distributions for all three level-2 variances for fit2. These would need to be in a sigma^2 metric, like as in an ICC. W/r/t the constrained model, fit3, you could demonstrate how the posterior for sigma_{a:b} was clearly not zero.
24.3. The end of this section might be a natural place to compare level-2 variance parameters. Consider computing percent-of-total-variance distributions for all three level-2 variances for
fit2. These would need to be in a sigma^2 metric, like as in an ICC. W/r/t the constrained model,fit3, you could demonstrate how the posterior for sigma_{a:b} was clearly not zero.