在分層病例對照研究下，針對羅吉斯迴歸模型我們推論迴歸參數的半參數最大概似估計量(SMLE)，並提出動差型式檢定統計量以診斷模型的適合度。我們的方法主要是推廣Qin & Zhang (1997) 的概念，將模型重參數化後得到多組的雙樣本半參數化模型，以此推論SMLE，並利用描述反應變數在非參數和半參數化模型下二階動差的差距來建立檢定統計量。我們推論得到SMLE的大樣本性質，及檢定統計量將分配收斂至卡方分配。模擬研究中我們發現即使在有限樣本的情況下，該檢定統計量亦執行良好。最後呈述一個關於痲瘋病研究的範例。 We inference the semiparametric maximum likelihood estimate (SMLE) and present the moment specification test of the logistic regression model for stratified case-control data. By generalizing the concept of Qin & Zhang (1997), we get the two-sample semiparametric model in each stratum and then inference the SMLE base on this finding. The test statistic is constructed via a discrepancy between two second moments under nonparmetric and semiparametric model. The large sample properties of SMLE are described. The proposed test statistic converges in distribution to the Chi-squared distribution. Simulation studies demonstrate that the test statistic perform well even in finite sample. Illustration with a leprosy disease study is provided
