Hojat Mohammadreza, Xu Gang
Center for Research in Medical Education and Health Care and Department of Psychiatry and Human Behavior, Jefferson Medical College of Thomas Jefferson University, 1025 Walnut Street, Philadelphia, PA 19107, USA.
Adv Health Sci Educ Theory Pract. 2004;9(3):241-9. doi: 10.1023/B:AHSE.0000038173.00909.f6.
Effect Sizes (ES) are an increasingly important index used to quantify the degree of practical significance of study results. This paper gives an introduction to the computation and interpretation of effect sizes from the perspective of the consumer of the research literature. The key points made are: 1. ES is a useful indicator of the practical (clinical) importance of research results that can be operationally defined from being "negligible" to "moderate", to "important". 2. The ES has two advantages over statistical significance testing: (a) it is independent of the size of the sample; (b) it is a scale-free index. Therefore, ES can be uniformly interpreted in different studies regardless of the sample size and the original scales of the variables. 3. Calculations of the ES are illustrated by using examples of comparisons between two means, correlation coefficients, chi-square tests and two proportions, along with appropriate formulas. 4. Operational definitions for the ES s are given, along with numerical examples for the purpose of illustration.
效应量(ES)是用于量化研究结果实际意义程度的一个日益重要的指标。本文从研究文献使用者的角度介绍效应量的计算与解读。要点如下:1. 效应量是研究结果实际(临床)重要性的有用指标,可操作性地定义为从“可忽略”到“中等”再到“重要”。2. 效应量相对于统计显著性检验有两个优势:(a)它独立于样本大小;(b)它是一个无尺度指标。因此,无论样本大小和变量的原始尺度如何,效应量在不同研究中都能得到统一解读。3. 通过两个均值比较、相关系数、卡方检验和两个比例的例子以及适当的公式来说明效应量的计算。4. 给出了效应量的操作性定义,并附带数值示例用于说明。