The recent commentary by Beaujean and Farmer (2020) on the original paper by Dixon et al. (2019) serves a cautionary tale of selective p-values, the law of small N sizes, and the type-II error. We believe these authors have crafted a somewhat questionable argument in which only 57% of the original Dixon et al. data were re-analyzed, based on a series of assumptions that removed significant power from the original statistical analysis. We here provide an additional 12 re-analyses of the original data and demonstrated that depending on the testing assumptions and data inclusion criteria, p-value may or may not exceed the commonly used .05 level. Although we support healthy discussion on experimental methods, and appreciate the reinterpretation of our findings, the reader is cautioned that because Beaujean and Farmer’s conclusions are limited to such a restricted range of data and different pre-analytic assumptions, the actual importance of their obtained p-value must remain in the eye of the beholder.

Beaujean 和 Farmer (2020) 最近对 Dixon 等人的原始论文的评论。(2019) 提供了一个关于选择性 p 值、小 N 大小定律和 II 型错误的警示故事。我们相信这些作者精心设计了一个有点可疑的论点，其中只有 57% 的原始 Dixon 等人。数据被重新分析，基于一系列假设，这些假设从原始统计分析中去除了显着的功效。我们在这里提供了对原始数据的额外 12 次重新分析，并证明根据测试假设和数据包含标准，p 值可能会或可能不会超过常用的 0.05 水平。虽然我们支持对实验方法的健康讨论，并感谢对我们发现的重新解释，p值必须留在旁观者的眼中。