共變異數分析準確信賴區間之樣本數計算

Abstract

對於共變異數分析下的標準化處理均差，本研究提供精確信賴區間兩種方法樣本數計算 分別為寬度期望值和容忍率。在計算樣本數時，利用信賴區間的方法可提供 相當多的參數資訊，信賴區間隨著不同參數下所產生的寬度皆相，此時如何決定信賴區間的品質相當重要，藉由綜合性考量包含期望寬度或是容忍率方式就可以保證在這樣的準則底下所找到寬度和最適本數能達合理的目標，是一種品質上的保證。

This article provides the sample size calculation for precise confidence interval of Standardized analysis of covariance (ANCOVA) contrasts in two methods including CI width expectation and width tolerance rate. When calculating the appropriate sample size with the method of confidence interval could provide more information. By adopting different sample size to acquire distinct confidence interval, and the estimation width would vary from different parameters. How to evaluate the quality of the confidence interval is a crucial issue. This research consider both CI width expectation and width tolerance rate methods to ensure appropriate width and sample size under the regulation. It is an assurance on quality.