Statistical inference in zero-inflated random effects models with missing data

点击翻译按钮获取中文摘要

Correlated data are common in a variety of disciplines such as biological, ecological, health and medical studies. Data are correlated due to, for example, repeated-measures, or a nest design, or a natural hierarchy of units. Random effects models (or mixed models or hierarchical models) are widely used to analyse these data, in which random effects are introduced to account for correlation within a hierarchy. For example, in medical longitudinal studies, random effects are often used to model the heterogeneity between subjects. These random effects characterize the varying susceptibilities of subjects to certain diseases.***In modelling correlated data, it is common to encounter a large number of zeros. For example, in studies of insurance claims, a large number of people will have zero claims during a certain period of time. In addition, some data may be missing for certain reasons. The presence and zero-inflation and missing data poses challenges to data analysis. In the proposed research I will develop efficient data analysis methods for random effects models with the presence of zero-inflation and missing data. I will also pursue variable selection and model assessment in this context. As these data complexities are common in many disciplines, the methods to be developed will be widely applicable. The methods will be easily accessible to applied researchers through worked examples and user-friendly packages.**