Hi, I am performing a variance components analysis of a 3 level nested model of binary data. Given the lack of homoscedasticity at the individual data, how do I calculate ICCs?