Lynch Joan, Ramjan Lucie M, Glew Paul, Salamonson Yenna
School of nursing and Midwifery, Western Sydney University, Campbelltown, New South Wales, Australia.
Western Sydney University, Campbelltown, New South Wales, Australia.
Nurse Res. 2020 Oct 22. doi: 10.7748/nr.2020.e1745.
The P -value is frequently used in research to determine the probability that the results of a study are chance findings. A value less than 0.05 was once typically considered only to mean that results are 'statistically significant', as it indicates the chance they are false positives is less than one in 20 (5%). However, P<0.05 has transcended into meaning a study has had positive findings and its results are true and meaningful, increasing the likelihood it will be published. This has led to researchers over-emphasising the importance of the P-value, which may lead to a wrong conclusion and unethical research practices.
To explain what the P -value means and explore its role in determining results and conclusions in quantitative research.
Some researchers are calling for a move away from using statistical significance towards meaningful interpretation of findings. This would require all researchers to consider the magnitude of the effect of their findings, contemplate findings with less certainty, and place a greater emphasis on logic to support or refute findings - as well as to have the courage to consider findings from multiple perspectives.
The authors argue that researchers should not abandon P -values but should move away from compartmentalising research findings into two mutually exclusive categories: 'statistically significant' and 'statistically insignificant'. They also recommend that researchers consider the magnitudes of their results and report whether findings are meaningful, rather than simply focusing on P -values.
Lessening the importance of statistical significance will improve the accuracy of the reporting of results and see research disseminated based on its clinical importance rather than statistical significance. This will reduce the reporting of false positives and the overstatement of effects.
P值在研究中经常被用来确定一项研究结果是偶然发现的概率。曾经，小于0.05的值通常仅被认为意味着结果具有“统计学显著性”，因为这表明它们是假阳性的概率小于二十分之一（5%）。然而，P<0.05已经演变成意味着一项研究有阳性结果且其结果是真实且有意义的，这增加了它被发表的可能性。这导致研究人员过度强调P值的重要性，这可能会导致错误的结论和不道德的研究行为。
解释P值的含义，并探讨其在定量研究结果和结论判定中的作用。
一些研究人员呼吁从使用统计学显著性转向对研究结果进行有意义的解读。这将要求所有研究人员考虑其研究结果效应的大小，思考确定性较低的结果，并更加强调用逻辑来支持或反驳研究结果——同时要有勇气从多个角度考虑研究结果。
作者认为研究人员不应摒弃P值，而应避免将研究结果划分为两个相互排斥的类别：“统计学显著”和“统计学不显著”。他们还建议研究人员考虑结果的大小，并报告结果是否有意义，而不是仅仅关注P值。
降低统计学显著性的重要性将提高结果报告的准确性，并使研究基于其临床重要性而非统计学显著性进行传播。这将减少假阳性结果的报告以及对效应的夸大。
