在醫藥研究中，通常用敏感性和特異性來描述檢測儀器或藥劑的準確性。而為了比較兩種儀器或藥劑，成對的設計可以減少人與人之間的差異性所造成的偏差。Perera et al. (2017) 提出一個新的測量雙眼（成對器官）的準確性的方法，并介紹了雙眼的敏感性和特異性來描述如何進行篩檢測試。但是他們文中採用的模型過於複雜。在本篇論文中，我們利用只包含了伯努利分佈的簡單模型來分析資料。我們的方法在使用上極為便利，完全不需要上述文章中必須採用的複雜數值方法便可完成。同時，在推論成對器官準確性前需對成對器官個別的準確性參數做相同的檢定。本論文亦提出不需假設相關性的強韌概似推論法。;In medical research, the accuracy of a machine or a therapy is commonly measured by sensitivity and specificity. In order to compare the accuracy of the machines or therapies, paired design can reduce the effect of heterogeneity among individuals. A new binocular accuracy measurement was proposed by Perera et al. (2017), who introduced the binocular versions of sensitivity and specificity to describe how screening tests are conducted in practice. However, the models they adopted are excessively complicated. In this thesis, we proposed an easier model, which only involves Bernoulli distributions, to accomplish the intended tasks.
