A New Approach in Modeling a Categorical Response. I. Binary Case
提出一种基于内在优势比概念的分类响应变量建模新方法，引入两类非参数分布，并展示如何区分对称单峰分布，同时给出二元响应的递增和递减优势比模型。
Abstract A new approach in modeling categorical response variables is presented. This approach is based on characterizing symmetrical and unimodal distributions with respect to the biological mechanisms that they describe. This characterization is achieved through the notion of the intrinsic odds ratio. Two nonparametric classes of distributions are introduced, and discrimination among symmetric unimodal distributions is demonstrated. Two extended models for binary response variables—increasing and decreasing odds-ratio models—are presented.