透過您的圖書館登入
IP:3.144.230.82
  • 學位論文

序位資料之樣本數演算法

Sample Sizes Algorithm for Ordered Categorical Data

指導教授 : 吳建華

摘要


對於臨床試驗中很普遍的序位資料,我們提供一個演算法,是針對兩組多項分配的序位資料,可以在特定顯著水準和期望檢定力之下,透過電腦做精確的計算來得到樣本數。 傳統的樣本數計算方法,多是利用近似值逼近,再代入樣本數公式計算,容易產生較大的誤差。本篇研究,是採取勝算比的假設檢定,並以Trend scores和Wilcoxon scores的線性排列(linear rank),來描述序位資料,再以p-value方法找出拒絕區,計算條件精確檢定力和非條件精確檢定力,透過這樣的演算法,直到達到期望檢定力,求得所需樣本數。此外,我們比較在相同情形下,Whitehead樣本數公式和在(Myoung-Keun Lee et al. (2002))中提供的樣本數數據以及本篇研究經電腦計算所得到的樣本數大小作比較,並加以分析。 本研究一共分成四章,第一章說明研究背景和臨床試驗的相關介紹、研究動機與目的,以及論文架構的概述;第二章探討文獻理論;第三章是實例比較,將電腦計算的結果和其他方法加以比較;第四章是結論。 關鍵字:多項分配、序位資料、勝算比、精確檢定力、Trend scores、Wilcoxon scores、 p-value。

並列摘要


Ordinal data are commonly used in the clinical experiments. In this thesis we develop an algorithm which can obtain sample size through accurate computation of computer under a specific level of significance and expected power. The traditional methods of computing sample size which are usually calculated by asym- ptotic methods and lead the sample size formula to obtain the results. In this thesis, we apply linear ranks tests based on Trend scores and Wilcoxon scores to test the ordinal data.To obtain the critical region by p-value method and calculate the exact conditional power and the exact unconditional power. Through this algorithm, we obtain the sample size when it achieves exp- ected power. Besides, we compare and analysis the sample sizes calculated by Whitehead’s formula ,the data from the paper (Myoung-Keun Lee et al.(2002) )and outcomes of this thesis. Keywords:Multinomial distribution、Ordinal data、Proportional odds、Exact power、 Trend scores、Wilcoxon scores、p-value

參考文獻


[8]楊喻純”費雪精確度檢定之樣本數演算法”中原大學應用數學系碩士論文,2007
[1] Lee, Myoung-Keun,Song, Hae-Hiang,Kang, Seung-Ho and Ahn, Chul W. (2002):The Determination of Sample Sizes in the Comparison of Two Multinomial Proportions from Ordered Categories. Biometrical Journal 44 (2002) 4, 395–409.
[2] Hilton, J. and Mehta, C. R., 1993: Power and sample size calculations for exact conditional tests
with ordered categorical data. Biometrics 49, 609–616.
[3]台灣藥品臨床試驗資訊網 http://www.cde.org.tw/ct_taiwan/knowledge.html

延伸閱讀