Treatment Effect Heterogeneity in Paired Data
针对配对数据中处理效应同质性的常见假设，开发了检验方法，并以泊松和二项分布为例，研究了存在过度离散时的混合模型推断，尤其关注计数配对数据。
Data from matched pairs are often analysed under the assumption that treatment effects are homogeneous. We develop tests of this assumption, with Poisson and binomial examples. Mixture models for inference when overdispersion is present are investigated, with stress on matched pairs of counts.