具相關性的連續與個數資料，如一個人看病的次數與其血壓值，常見於醫學或其它研究領域中。這種相關性的資料在分析上較困難，原因是不易找到合適的統計模型。 本文嘗試利用Royall與Tsou (2003)的強韌概似函數法，來建立一個對迴歸係數做不需知道正確模型的有母數強韌推論法。 This thesis is concerned with regression analysis of bivariate correlated count and continuous data. The mixed Poisson-normal is chosen as the working model for the joint distribution of the bivariate data.We then show that this working model can be corrected to become asymptotically robust against model misspecifications. Full likelihood inference about regression parameters is therefore made available without knowing the true underlying joint distributions. Simulations and real data analysis are provided to demonstrate the efficacy of the new parametric robust likelihood approach for bivariate count-continuous data.
