摘要：Bounded outcome scores are often encountered in health-related survey studies. Such scores are usually bounded and discrete and are often treated as categorical or ordinal data, which is not satisfactory in some scenarios. The binomial-logit-normal distribution, as a parametric model, is a useful alternative for the bounded outcome scores. The prop osed model converges to the continuous logit-normal mo del and hence bridges the gap between discrete modeling and continuous modeling. This result is useful when the score is dense and smo oth within its bounded support. A quality of life data is shown as an application.
关键词：health outcome ; binomial-logit-normal ; ordinal data ; logistic regression
