Olejnik S, Algina J
University of Georgia
Contemp Educ Psychol. 2000 Jul;25(3):241-286. doi: 10.1006/ceps.2000.1040.
Although dissatisfaction with the limitations associated with tests for statistical significance has been growing for several decades, applied researchers have continued to rely almost exclusively on these indicators of effect when reporting their findings. To encourage an increased use of alternative measures of effect, the present paper discusses several measures of effect size that might be used in group comparison studies involving univariate and/or multivariate models. For the methods discussed, formulas are presented and data from an experimental study are used to demonstrate the application and interpretation of these indices. The paper concludes with some cautionary notes on the limitations associated with these measures of effect size. Copyright 2000 Academic Press.
尽管几十年来，人们对与统计显著性检验相关的局限性的不满一直在增加，但应用研究人员在报告其研究结果时，几乎仍然完全依赖这些效应指标。为了鼓励更多地使用效应的替代测量方法，本文讨论了几种效应大小的测量方法，这些方法可用于涉及单变量和/或多变量模型的组间比较研究。对于所讨论的方法，给出了公式，并使用一项实验研究的数据来演示这些指标的应用和解释。本文最后对这些效应大小测量方法的局限性提出了一些警示。版权所有2000年学术出版社。
