在醫學檢驗疾病並研發新藥時，常使用配對設計來判定此藥是否有效，若此配對設計資料有遺失值時(missing data)，則稱為部分成對資料。本文探討完全隨機遺失(missing completely at random, MCAR)的部分成對資料，此種資料因具相關性使得模型配適變得困難。本文主要的目的是利用強韌概似函數(robust likelihood function)方法，來分析部分成對資料。我們所建立的強韌概似函數，並不需要對該部分成對資料中的相關性建立模型假設，仍可得到正確的統計推論。;In medical research, the efficacy of the new drug is often decided by the paired design. The data of paired design that has missing values is called partially paired data. This article considers the partially paired data that is missing completely at random. It is hard to find a suitable model to analyze the correlated data.This article utilizes a robust likelihood function method to analyze the partially paired data. Using this robust likelihood function, we obtain correctly statistical inferences without modeling the correlated joint distribution of partially paired data.
