兩相關變量三元反應配對資料邊際比例差異之研究

None

李柏青(Po-Ching Li)

指導教授 ： 楊明宗

國立中央大學/理學院/統計研究所/碩士(2008年)

摘要

健康保健與用藥安全是近代社會兩個重要的課題，故在臨床醫學研究上對藥物控制研究特別受到重視，常需藉由比較不同劑量的藥劑或不同藥劑產生的影響作進一步的討論，用以增進生活的品質。本文針對兩相關變量三元反應配對資料邊際機率相關問題作深入的研究，提出常用之統計方法以比較不同樣本間的差異，並透過統計方法比較與數值模擬的過程找尋較簡易且適當的統計方法，供實務應用者方便使用。

關鍵字

邊際同質性 ； 配對資料

並列摘要

The health care and drug safety are two important subjects for modern society. Recently, clinical medical research is especially paid attention to pharmacy control. In order to improve quality of life, it is necessary to investigate the difference between dosages or drugs. The goal of this paper is to test the marginal homogeneity having two dependent variables of matched pair data with three responses. Some well-known statistic methods are studied. Comparisons between those methods are conducted by simulation studies, and simple and suitable methods are recommended for practical applications.

並列關鍵字

marginal homogeneity ； matched pair

參考文獻

distributions of multivariate categorical responses. Journal of the American Statistical

1. Agresti, A. (2002). Categorical data analysis. Wiley, New York.

probabilities, with application to safety studies for drugs. Applied Statistics 54, 691-816.

3. Agresti, A. and Klingenberg, B. (2006). Multivariate extensions of McNemar’s test. Biometrics 45, 629-636.

4. Agresti, A. and Min, Y. (2005). Simple improved confidence intervals for comparing matched proportions. Statistics in Medicine 24, 729–740.