In a multilevel analysis with mixed, I can estimate the icc after the main analysis. However, I may have a hypothesis about the intraclass correlation such that:
i. a level-2 factor predicts tighter intraclass clustering, or
ii. a level-1 outcome is predicted by X and moderated by the intraclass correlation, such that the effect of X is stronger when the ICC is high or low

Is there a direct way to test such hypothesis using the ICC, or an indirect way of computing some other index of interdependence and using that variable in these analyses? I'm analyzing two longitudinal data sets. In one the clusters are 2-person teams and in the other the clusters are 8-person teams. Thanks.