Definitely an issue I sometimes spot in psych as well! If you want to control for higher-level characteristics, you’ll need a fixed effects model — or a multilevel model with tweaks https://www.statmodel.com/download/HamakerMuthenFixedVSRandom.pdf

Comments