We consider estimation of mixed-effects logistic regression models for longitudinal data when missing outcomes are not missing at random. A typology of missingness mechanisms is presented that ...