The P value is the most common method used in medical literature for the result of a statistical test. It is the probability of the data with a true null hypothesis and is calculated using a formal statistical test after the appropriate model has been determined to analyze study data. The P value is dependent on the effect size, sample size, and a measure of variability within the outcomes. For many years, the P value has been set at 0.05, which is an arbitrary cutoff. It is important to understand that setting the cutoff at 0.05 may be correct for some study designs but not in others. Therefore, we recommend that in addition to the P value, another metric should be reported that specifies the magnitude of the effect such as effect size, confidence interval of the effect size, or fragility index.
From the Department of Medicine, Division of Physical Medicine and Rehabilitation, University of Toronto, Toronto, Ontario, Canada (DK, SMA); and Neural Engineering and Therapeutic Team, Lyndhurst Centre, UHN-Toronto Rehabilitation Institute, Toronto, Ontario, Canada (SMA).
All correspondence should be addressed to: Dinesh Kumbhare, MD, PhD, FRCPC, FAAPMR, University of Toronto, Suite 7-131, 550 University Ave, Toronto, Ontario, Canada M5G 2A2.
Financial disclosure statements have been obtained, and no conflicts of interest have been reported by the authors or by any individuals in control of the content of this article.
Online date: August 28, 2019