The Measurement and Communication of Effect Sizes in Management Research

• Published in: Management and organization review Vol. 19; no. 1; pp. 176 - 197
• Main Authors: Fey, Carl F., Hu, Tianyou, Delios, Andrew
• Format: Journal Article
• Language: English
• Published: New York, USA Cambridge University Press 01.02.2023
• Subjects: Dependent variables; Estimates; Hypothesis testing; Independent variables; Perspectives; Research methodology; Researchers; Sample size; Statistical significance; 效应强度; 数据分析; 数据汇报; statistical reporting; confidence interval; data visualization; effect size; statistical analysis; 置信区间; 数据可视化
• ISSN: 1740-8776; 1740-8784
• DOI: 10.1017/mor.2022.2

The measurement and communication of the effect size of an independent variable on a dependent variable is critical to effective statistical analysis in the Social Sciences. We develop ideas about how to extend traditional methods of evaluating relationships in multivariate models to explain and illustrate the statistical power of a focal independent variable. Even with a growing acceptance of the need to report effect sizes, scholars in the management community have few well-established protocols or guidelines for reporting effect sizes. In this editorial essay, we: (1) review the necessity of reporting effect sizes; (2) discuss commonly used measures of effect size and accepted cut-offs for large, medium, and small effect sizes; (3) recommend standards for reporting effect sizes via verbal descriptions and graphical presentations; and (4) present best practice examples of reporting and discussing effect size. In summary, we provide guidance for authors on how to report and interpret effect sizes, advocating for rigor and completeness in statistical analysis. 在社会科学研究中进行有效的数据分析，一个关键点是测量和汇报自变量对应变量影响的强度大小，即效应强度。通过对多变量数据分析方法的汇总和拓展，我们致力于如何阐释、展示效应强度——这个重要的统计学功效。管理学界虽然逐步认可接受汇报效应强度，但仍缺乏完整的操作方法和汇报准则。在此篇社论中，我们主要做了以下工作：1）综述和重申汇报效应强度的必要性； 2）讨论常用的效应强度测量指标，以及学界普遍认可的大、中、小强度的临界值； 3）提出汇报效应强度的文字说明和绘图等具体操作标准； 4）列举一些优秀例子辅以说明。总之，我们在此提出汇报和阐述效应强度的指导性原则，以推进管理学研究中严谨和完整的数据分析。