Studying behaviour in economics, sociology, and statistics often involves fitting a model in which the outcome is an ordinal response which is only observed for a subsample of subjects. (For example, questions about health satisfaction in a survey might be asked only of respondents who have a...