在醫學與其他研究領域中，常會出現群集成對的連續與離散資料，如家庭中每一位成員的血壓值與是否得到癌症。由於這種相關性資料不容易找到合適的模型，因此在分析上較為困難。 本文主要的目的是在廣義線性模型下，利用強韌概似函數方法，來分析群集中成對的混合型資料。我們建立參數之強韌概似函數，在不需要特別對於群集中成對的反應變數間及群集中不同成員的反應變數間之二層相關性建立模型的假設下，仍可得到正確的統計推論。 ;In medicine and other fields of research, cluster pairs of continuous and discrete data are common, such as the blood pressure value of each member of the family and whether they have cancer. It is hard to find a suitable model to analyze this correlated data. In this thesis, we propose a robust likelihood function method under the generalized linear model to analyze the cluster paired mixed data. Using this robust likelihood function, we can make correct statistic inferences without modeling the two-layer correlation between the same unit within a cluster and for different members within the same cluster.
