Johnson Samantha L, Stone Whitley J, Bunn Jennifer A, Lyons T Scott, Navalta James W
Department of Health and Human Performance, Middle Tennessee State University, Murfreesboro, TN, USA.
School of Kinesiology, Recreation, & Sport, Western Kentucky University, Bowling Green, KY, USA.
Int J Exerc Sci. 2020 Sep 1;13(1):1-5. doi: 10.70252/HMZN3851. eCollection 2020.
Statistical reporting of quantitative research data has been plagued by potential bias and reporting suppression due to a single numerical output: the -value. While there is great importance in its merit, creating a pass-fail system (set at point of .05) has created a culture of researchers submitting their project's data to a filing cabinet if it does not yield "statistical significance" based on this value. The editors of the International Journal of Exercise Science are following the American Statistical Association's call for statistical reform by adjusting our reporting guidelines to the following requirements: [1.] make an intentional effort to move away from statements "statistically significant" or "not significant;" [2] all -values are to be reported in their raw, continuous form; [3.] measures of the magnitude of effect must be presented with all -values; [4.] either an a-priori power analysis with relevant citations should be included or post-hoc power calculations should accompany -values and measures of effect. The ultimate goal of this editorial is to join with other scholars to push the field toward transparency in reporting and critical, thoughtful evaluation of research.
定量研究数据的统计报告一直受到潜在偏差和报告抑制的困扰,原因在于单一的数值输出:p值。虽然p值有其重要性,但建立一个通过/失败系统(设定为0.05)已经营造了一种文化,即如果研究项目的数据基于该值未产生“统计学显著性”,研究人员就会将其提交的数据束之高阁。《国际运动科学杂志》的编辑们响应美国统计协会的统计改革呼吁,将我们的报告指南调整为以下要求:[1.] 有意避免使用“统计学显著性”或“不显著”的表述;[2] 所有p值都应以原始的连续形式报告;[3] 效应大小的度量必须与所有p值一起呈现;[4] 应包含带有相关引用的先验功效分析,或者p值和效应度量应附带事后功效计算。这篇社论的最终目标是与其他学者一道,推动该领域在报告方面实现透明度,并对研究进行批判性、深思熟虑的评估。