The Importance of Complexity in Modeling

Rotten Bush and Bright wrote about these diagnostics in the first edition of their book. They have a really cool example where you get empirical base estimates of level two random effects do residual plots for trends but when you bring in a cross-level interaction that's been omitted it actually accounts for the curvature. I can't even remember the last application I've seen published that anybody is even commented on doing diagnostics to examine meeting assumptions in the MLM.