ABSTRACT
Multivariate multilevel regression models contain multiple response variables, like multivariate analysis of variance (MANOVA). Multivariate analysis controls the type I error better than using several univariate tests. Multivariate analysis often has more power. Multilevel multivariate analysis uses an additional level for the dependent variables. Even in single-level data, it is a useful alternative to MANOVA when the response variables have missing values, or if the dependent variables are dichotomous. It can also be used to construct multilevel measurement models, by including questions that form a scale as multivariate responses in a multilevel model.
