對於分析長期追蹤資料或族群資料時，某些應變數之間並非是獨立分布的，此時可在迴歸模式中加入隨機效用來說明相關性。本計畫將探討廣義線性混合效用測量誤差模式(GLMMeM)的參數估計，雖然已有一般常用於分析測量誤差模式的統計分法例如迴歸校正，模擬外插或是校正分數函數被應用在相關的問題， 包含線性及廣義線性的混合效用模式(mixed effect model)，但卻沒有使用條件分數函數於GLMMeM上的相關討論，然而在沒有隨機效用的廣義線性測量誤差模式上，除了計算可能較複雜以外，條件分數函數所需的假設不強而且結果經常較其它方法精準，因此我們也預期條件分數函數在GLMMeM上也會有相同的優點，值得發展。 In analyzing a longitudinal data or clustered data, one can introduce the random effect components into the regression model to account for the correlation between the individuals within the subgroup. In this project, we consider the estimation of the generalized linear mixed model when the covariate is subject to measurement error which is abbreviated to GLMMeM (Generalized Linear Mixed Measurement error Model). Some conventional approaches in the context of measurement error model, for example, “Regression calibration” , “SIMEX” and “Corrected score” had been applied to GLMMeM with distributional assumptions on the miss-measured covariate. However, the conditional score approach usually performs better than these methods in a fixed effect measurement error model, besides, the conditional score may require less assumptions about the distribution of miss-measured covariate. Thus, it is worthwhile to develope a conditional score estimation in the GLMMeM problem for it may perform better in the GLMMeM than the existent methods.