交叉設計是臨床實驗中常用的實驗手法。為了分析二元資料型態的交叉設計，本文使用獨立伯努利模型作為實作模型，利用強韌概似函數方法強韌化實作概似函數，得到強韌華德檢定統計量、強韌分數檢定統計量、強韌概似比檢定統計量，由此可得到正確的統計推論。 我們利用模擬研究與實例分析，比較強韌華德檢定統計量、強韌分數檢定 統計量、強韌概似比檢定統計量與Schouten and Kester(2009)提出的SK 檢定統計量分析結果的差異。;Crossover design is a commonly used experimental technique in clinical trials. In order to analyze crossover design for binary data, the independent Bernoulli model is used as the working model, and the robust likelihood function method is used to robustize the working likelihood function to obtain the robust Wald statistics, robust score statistics and robust likelihood ratio statistics in this paper. From these, we can get correct statistical inferences. Using simulation research and case analysis, we can compare the differences between the robust Wald test statistics, the robust score test statistics, the robust likelihood ratio test statistics and the SK test statistics proposed by Schouten and Kester (2009) analysis results.
