How to Assess Endogeneity

Endogeneity is the correlation between your random effects and your predictors or the random effects in your outcome depending on how you want to define it. You mostly blow that if you don't include all of the relevant covariates at the second level in the model. I think there's this misconception that if I include random intercepts, then I'm accounting for clustering and all the reasons are explained. And that always drives me crazy because the random intercept variance is just saying it's a residual, but now it's a remnant somewhere else. It's one of these phantom assumptions that you have to meet, but at the same time you're kind of difficult to test.