Rudy E B, Kerr M
School of Nursing, Case Western Reserve University, Cleveland, OH 44108.
Heart Lung. 1991 Sep;20(5 Pt 1):517-22.
Power analysis provides one method for assessing the efficacy of alternative research designs. The purpose of this article is to simplify the methods for calculating power analysis to determine an adequate sample size. Although attention to type I errors (alpha error) is prevalent among nurse researchers, there is less appreciation for research problems resulting from a type II error. Without the a priori examination of the power of the test of significance, there is no assurance that the sample is sufficient to discern significant differences. Formulas for calculating effect size are provided for t tests, correlations, chi-square, analysis of variance, and regression. Examples of calculating the effect size by using four different statistical tests based on research studies are presented: t tests with unequal variance between groups, chi-square, an analysis of variance, and regression. Power analysis is an additional procedure to ensure that the sample size is adequate for the research project about to be undertaken.
功效分析提供了一种评估替代研究设计有效性的方法。本文的目的是简化计算功效分析以确定足够样本量的方法。尽管护士研究人员普遍关注I型错误（α错误），但对II型错误导致的研究问题的认识较少。如果没有对显著性检验功效进行事先检验，就无法保证样本足以辨别显著差异。文中提供了t检验、相关性分析、卡方检验、方差分析和回归分析的效应量计算公式。还给出了基于研究实例使用四种不同统计检验计算效应量的示例：组间方差不等的t检验、卡方检验、方差分析和回归分析。功效分析是一项额外的程序，可确保样本量足以用于即将开展的研究项目。
