在各種領域中常會接觸到具有相關性的二分法資料，這些相關性可能來自於生物間的基因、環境、重複測量或是時間所造成。Royall and Tsou (2003)所提出的強韌概似函數方法，在大樣本下，使得二分法資料具有相關性或者資料並非來自二項分配時，皆能對迴歸參數做出正確的推導。 本文之目的在利用二項分配模型的修正項，得到估計群集內資料間相關性的新方法。 Correlated data are commonly encountered in many fields. The correlation may come from the genetic heredity, familial aggregation, environmental heterogeneity, or repeated measures. Royall and Tsou (2003) proposed a parametric robust likelihood technique. With large samples, the adjusted binomial likelihood is asymptotically legitimate for correlated binary data. In this work, we use the adjustment by the binomial working model and obtain a new method for estimating the correlation between data in a cluster.
