Putting the P -Value in its Place

Abstract

As the debate over best statistical practices continues in academic journals, conferences, and the blogosphere, working researchers (e.g., psychologists) need to figure out how much time and effort to invest in attending to experts' arguments, how to design their next project, and how to craft a sustainable longterm strategy for data analysis and inference. The present special issue of The American Statistician promises help. In this article, we offer a modest proposal for a continued and informed use of the conventional p-value without the pitfalls of statistical rituals. Other statistical indices should complement reporting, and extra-statistical (e.g., theoretical) judgments ought to be made with care and clarity.

Ten common statistical mistakes to watch out for when writing or reviewing a manuscript

“…However, banning p -values does not necessarily protect researchers from making incorrect inferences about their findings (Fricker et al, 2019). When applied responsibly (Kmetz, 2019; Krueger and Heck, 2019; Lakens, 2019), p -values can provide a valuable description of the results, which at present can aid scientific communication (Calin-Jageman and Cumming, 2019), at least until a new consensus for interpreting statistical effects is established. We hope that this paper will help authors and reviewers with some of these mainstream issues.…”

RNA-Seq analysis of transcriptome responses in Atlantic cod (Gadus morhua) precision-cut liver slices exposed to benzo[a]pyrene and 17α-ethynylestradiol

“…The statistical test such as Student's t-test requires a specific distribution of samples which can not always be satisfied [55]. Recent debates include alleged misuse of P-value [126][127][128][129][130][131]. The thresholds for Fold Change and P-value also significantly alter the gene expression data interpretations [132].…”

Spanish Scientists’ Opinion about Science and Researcher Behavior

“…The overall picture we gain from our results is that most respondents believe that researchers only publish statistically significant results (93.1%) and that science advances most when novel hypotheses are proposed (77.6%), that scientific journals are not interested in publishing null results (84.5%) but in publishing novel findings (82.8%), that replication studies are necessary when the published findings are contradictory (96.6%) but less so if the findings in the literature are unanimous (72.4%). In addition, over 50% of the respondents misinterpret the meaning of a statistically significant result and associate it with importance, the usefulness of the finding, and the size of the effect (Krueger & Heck, 2019). And while they may not agree with keeping a statistically non-significant result in the drawer (47.4%), they do not consider it a priority to publish these findings (66.4%).…”