臺灣博碩士論文加值系統

在臨床實驗上，如何取得有效的樣本數是我們所好奇的。在研究上，以往的樣本數計算多以近似法逼近，再帶入樣本數公式計算求得樣本數大小，但是這樣的方法容易產生較大的誤差，在傳統方法上則是會面臨龐大計算過程而浪費時間的問題，而且傳統上以討論2維度下的樣本數模擬為主。因此本文想提供一個3維度下的演算法來避免這些問題且求出較符合條件的樣本數。本文提出的方法是不需要透過近似值，藉由電腦演算的方式來解決龐大計算過程的問題，又可快速達到精準的樣本數大小。
在過去文獻中，是採用趨近法去計算檢定力（power），且步驟是先給定樣本數後去估計檢定力；本文則是直接採用分配，利用α方程式和β方程式及電腦演算的方式，直到符合條件停止，即可得到精確的樣本數大小，而並非先給定樣本數之後再針對公式去估算檢定力。文中運用了在2×2×2列聯表的精確度檢定，以電腦演算的方式來計算臨床上所需的樣本數，使得療效可以提高有效性卻不浪費資源。

In the clinical experiment, our primary objective is to obtain the effective sample sizes. The sample sizes is usually calculated by asymptotic methods, and leads the sample sizes formula to obtain the sample number size, but such method is easy to have a bigger error and huge computation process in the traditional method. It discusses the sample number simulations under two dimensions primarily in the traditional method. Therefore this article wants to provide the algorithm under three dimensions to avoid these questions and obtain the conditional sample sizes. This article proposed the method is does not need to penetrate the approximate value, and the way can solve the huge process problem and achieve the fine sample size by the computer algorithm.
In the past literature, it was based on asymptotic methods to calculate examines the strength in meditation (power), and also given the sample number first ; This article directly uses the distribution, α equation and β equation and the simulation, until conforms to the condition to stop, then obtains the precise sample number size. In the article has utilized the Exact test in 2×2×2 Tables which to calculates the sample sizes in clinically by the computer simulates and causes the curative effect to be allowed to enhance the validity actually not to waste the resources.

中文摘要………………………………………………………………………..… I
英文摘要…………………………………………………………………… …… II
謝誌……………………………………………………………………………… III
目錄…………………………………………………………………………....... IV
圖目錄……………………………………………………………………....... V
表目錄……………………………………………………………………....... VI
第一章 緒論…………………………………………………………………….. 1
第一節 背景...………………………………………………………………. 1
第一項 死刑犯之例……………………………………………….... 1
第二節 研究目的與動機………………………………………………….. 3
第三節 研究架構………………………………………………………….. 3
第二章 樣本數計算…………………………………………………………….. 4
第一節 勝算比（odds ratio）…………………………………………….. 4
第二節 在2×2×2列聯表下的精確度檢定………………………………. 6
第一項 在2×2×2列聯表下精確度檢定之計算檢定力……….……. 7
第二項 樣本數之演算法……….……………………………………. 8
第三章 結果分析……………………………………………………………….. 12
第四章 結論…………………………………………………………………….. 14
參考文獻………………………………………………………………………….. 15
基本資料………………………………………………………………………….. 16
圖目錄
圖一：演算法流程圖………………………………………………..…………… 11
表目錄
表1：死刑判決按被告及被害者種族分類…………………………………….. 1
表2：病人在試用新療效時分為有治癒成功及沒有治癒成功………………. 6
表3：Exact and approximated powers（multiplied by 1000）for α=0.05………. 12
表4：電腦模擬之所求樣本數…………………………………………………… 13

［1］劉應興,”類別資料分析導論” 臺北市, 華泰, 2003.
［2］Michael Haber. Sample-Sizes for the Exact Test of 'No Interaction' in 2 x 2 x 2
Tables. Biometrics, Vol. 39, No. 2 (Jun., 1983), pp. 493-498
［3］Heilbron,D.C.（1981）.The analysis of ratios of odds ratios in stratified
contingency tables. Biometrics 37,55-66.
［4］Zelen,M.（1972）.Exact significance tests for contingency tables embedded in a
2n classification. In Proceedings of the Sixth Berkeley Symposium on
Mathematical Statistics and Probability, Vol.I,L.M.Lecam, J.Neyman and E.
Scott(eds),737-757.Berkeley:University of California Press.
［5］Bartlett, M.S.（1935）.Contingency table interactions. Journal of the Royal
Statistical Society ,Supplement 2,248-252.
［6］Good man,L.A.（1964）.Simple methods for analyzing three-factor interaction in
contingency tables. Journal of the American Statistical Association 59,319-352.
［7］Zacks, S. and Solomon, H.（1976）. On testing and estimating the interaction
between treatments and environmental conditions in binomial experiments: the
case of two stations. Communications in Statistics, Series A 5, 197-223
［8］簡玉芝,”以電腦模擬計算臨床試驗之樣本數”中原大學碩士論文,94學年