Small steps to help improve the caliber of the reporting of statistics
2017; American Physical Society; Volume: 41; Issue: 3 Linguagem: Inglês
10.1152/advan.00049.2017
ISSN1522-1229
Autores Tópico(s)Data Analysis with R
ResumoEditorialSmall steps to help improve the caliber of the reporting of statisticsDouglas Curran-EverettDouglas Curran-EverettDivision of Biostatistics and Bioinformatics, National Jewish Health, Denver, Colorado; and Department of Biostatistics and Informatics, Colorado School of Public Health, University of Colorado Denver, Denver, ColoradoPublished Online:05 Jul 2017https://doi.org/10.1152/advan.00049.2017MoreSectionsPDF (102 KB)Download PDF ToolsExport citationAdd to favoritesGet permissionsTrack citations ShareShare onFacebookTwitterLinkedInWeChat in 2004 dale benos and i published guidelines for reporting statistics in journals published by the American Physiological Society (6). In so doing, we had hoped the guidelines would help improve the caliber of statistical information reported in those journals. By 2007, an admittedly short interval, it was clear that the mere publication of the guidelines was unlikely to impact reporting practices (7).The guidelines themselves sparked unsolicited comment (see Refs. 7, 15, 16, 18, and 20). Our sequel to the guidelines (7) provided for invited commentary (see Refs. 2, 17, 19, and 21). Through it all, Dale and I believed that the guidelines reflected mainstream statistical best practices (6, 8).1 I still do. I am sure Dale would too.Back then, to us and to the Editors-in-Chief, authors complained most vigorously aboutGuideline 5. Report variability using a standard deviation.They did so for two reasons. They failed to appreciate the distinction between a standard deviation and a standard error:2I do not agree with the edict about presenting data as [standard deviations] rather than [standard errors of the mean]. These presentations are for visual effect only. . . . To me, this edict is silly, particularly since showing [standard deviations rather than standard errors of the mean] is a cosmetic issue only. (Comment cited in Ref. 7.)or they preferred to report a standard deviation using the format90±10 mmHgwhen the guidelines and other papers (1, 3, 10, 11, 13) advocated notation of this form:90 mmHg (SD 10).As I begin my second term as Editor-in-Chief of Advances in Physiology Education, a full 10 years after Dale Benos and I published our sequel to the guidelines for reporting statistics, I remain committed to improving statistical practice and reporting among researchers. I am but one of many. Our collective efforts have gone on for decades. There is a reason for that. As educators, we know only too well how difficult it is to correct a misconception one of our students may hold. Why should we be any different?I equate trying to change the reporting practices of statistics with trying to change the direction of an ocean liner with a kayak. Good luck with that. When I mentioned this analogy to a colleague, he said,What we need are more kayaks. A lot more kayaks.I understand it is difficult to change entrenched practices (7, 9). I get that change is slow. But that does not mean we should not try.With this Editorial, I am announcing that I have asked the Associate Editors and Editorial Board of Advances—effective June 2017—to actively promote two of the 2004 guidelines for reporting statistics:Guideline 5. Report variability using a standard deviation.Guideline 7. Report a precise P value.Below is my rationale for doing so.Guideline 5. Report variability using a standard deviation.3 Suppose the random variable Y represents the physiological thing we care about. Let us simplify our lives and assume that Y is distributed normally with mean μ and standard deviation σ. The mean μ describes the location of the center of the distribution of Y, and the standard deviation σ—the square root of the variance σ2—describes the spread of the normal distribution (Fig. 1).Fig. 1.Two normal distributions. These distributions differ in their standard deviation σ, but for each distribution, 68% of the possible values are within 1σ of the mean μ (gray area).Download figureDownload PowerPointThese two parameters, μ and σ, determine a normal distribution. We can describe the theoretical distribution of possible outcomes of our random variable Y with the normal probability density function f (y):f(y)=1σ2π exp[−(y−μ)22σ2],for −∞ < y < +∞ (12). Bear in mind that the standard deviation σ is a single positive number.Guideline 7. Report a precise P value. The 2004 guidelines said a precise P value does two things: it communicates more information with the same amount of ink, and it permits each reader to assess a statistical result (6). Moreover, only with a precise P value can we estimate the chances that we will reproduce someone else's scientific result (5, 22). Table 1 reiterates guidelines for the appropriate rounding of precise P values.Table 1. Guidelines for rounding P valuesP Value RangeRounding Precision0.01 ≤ P ≤ 1.00Round to 2 decimal places: round P = 0.024 to P = 0.02.0.001 ≤ P ≤ 0.009Round to 3 decimal places: round P = 0.0048 to P = 0.005.P < 0.001Report as P < 0.001.Adapted with permission from Ref. 7.As Dale and I wrote in 2004 and in 2007 (6, 8), these guidelines embody fundamental concepts in statistics (10), and they are consistent with Scientific Style and Format (3), the style manual used by the American Physiological Society. These guidelines are also fully supported by the American Statistical Association (14, 23) which only recently issued its first-ever position statement on a specific component of statistical practice (23).In 2007 Tom Lang (17) wrote:Meeting high standards should be required in all research and publication efforts, not merely recommended. We require investigators to use the scientific method; we do not just recommend that they do. We require investigators to explain their experimental procedures; we do not just recommend that they do. We even require investigators to format their references correctly; we do not just recommend that they do. Authors should be required to report statistics as completely and as accurately as every other aspect of the research. To allow ignorance, tradition, personal preference, or the practices of other journals to justify anything less is to legitimize the very forces that science attempts to overcome.Given that I am a relentless optimist, I am hopeful—even expectant—that Advances readers and authors will embrace these two guidelines. As always, if you are moved to comment, I am happy to listen.DISCLOSURESNo conflicts of interest, financial or otherwise, are declared by the author(s).ACKNOWLEDGMENTSI thank Ronald Wasserstein (Executive Director, American Statistical Association), Rita Scheman (Director of Publications and Executive Editor, American Physiological Society), Gordon Drummond (School of Medicine, University of Edinburgh, UK), Tom Lang (Medical Writing and Editing Program, University of Chicago), Calvin Williams (Clemson University, Clemson, South Carolina), past Editor-in-Chief Rob Carroll, past Deputy Editor Jon Kibble, Deputy Editor Barb Goodman, and Associate Editors David Harris, Mohammed Khalil, Jodie Krontiris-Litowitz, Bryan Mackenzie, Nancy Pelaez, Kathy Ryan, Arif Siddiqui, and Dee Silverthorn for their helpful comments and suggestions.REFERENCES1. Altman DG, Gardner MJ. Presentation of variability. Lancet 328: 639, 1986. doi:10.1016/S0140-6736(86)92470-0. Crossref | PubMed | ISI | Google Scholar2. Clayton MK. How should we achieve high-quality reporting of statistics in scientific journals? A commentary on "Guidelines for reporting statistics in journals published by the American Physiological Society". Adv Physiol Educ 31: 302–304, 2007. doi:10.1152/advan.00084.2007. Link | ISI | Google Scholar3. Council of Science Editors, Style Manual Committee. Scientific Style and Format: The CSE Manual for Authors, Editors, and Publishers (7th Ed.). Reston, VA: Council of Science Editors, 2006, p. 167–170.Google Scholar4. Curran-Everett D. Explorations in statistics: standard deviations and standard errors. Adv Physiol Educ 32: 203–208, 2008. doi:10.1152/advan.90123.2008. Link | ISI | Google Scholar5. Curran-Everett D. CORP: Minimizing the chances of false positives and false negatives. J Appl Physiol (1985) 122: 91–95, 2017. doi:10.1152/japplphysiol.00937.2016. Link | ISI | Google Scholar6. Curran-Everett D, Benos DJ. Guidelines for reporting statistics in journals published by the American Physiological Society. Adv Physiol Educ 28: 85–87, 2004. doi:10.1152/advan.00019.2004. Link | ISI | Google Scholar7. Curran-Everett D, Benos DJ. Guidelines for reporting statistics in journals published by the American Physiological Society: the sequel. Adv Physiol Educ 31: 295–298, 2007. doi:10.1152/advan.00022.2007. Link | ISI | Google Scholar8. Curran-Everett D, Benos DJ. Last Word on Perspectives "Guidelines for reporting statistics in journals published by the American Physiological Society: the sequel". Adv Physiol Educ 31: 306–307, 2007. doi:10.1152/advan.00089.2007.Link | ISI | Google Scholar9. Curran-Everett D, Benos DJ. Statistics, authors, and reviewers: the heart of the matter. Adv Physiol Educ 33: 80, 2009. doi:10.1152/advan.90216.2008. Link | ISI | Google Scholar10. Curran-Everett D, Taylor S, Kafadar K. Fundamental concepts in statistics: elucidation and illustration. J Appl Physiol (1985) 85: 775–786, 1998. Link | ISI | Google Scholar11. Gardner MJ. Letter: Understanding and presenting variation. Lancet 305: 230–231, 1975. doi:10.1016/S0140-6736(75)91414-2.Crossref | Google Scholar12. Hogg RV, Craig AT. Introduction to Mathematical Statistics (4th ed.). New York: Macmillan, 1978.Google Scholar13. Iverson C, American Medical Association. AMA Manual of Style: a Guide for Authors and Editors (10th Ed.). New York: Oxford University Press, 2007, p. 894.Google Scholar14. Kass RE, Caffo BS, Davidian M, Meng X-L, Yu B, Reid N. Ten simple rules for effective statistical practice. PLOS Comput Biol 12: e1004961, 2016. doi:10.1371/journal.pcbi.1004961. Crossref | PubMed | ISI | Google Scholar15. Kay B, Curran-Everett D, Benos DJ. The ongoing discussion regarding standard deviation and standard error. Adv Physiol Educ 32: 334–335, 2008. doi:10.1152/advan.90191.2008. Link | ISI | Google Scholar16. Koehnle T, Curran-Everett D, Benos DJ. The proof is not in the P value. Am J Physiol Regul Integr Comp Physiol 288: R777–R778, 2005. doi:10.1152/ajpregu.00511.2004. Link | ISI | Google Scholar17. Lang T. The need for accurate statistical reporting. A commentary on "Guidelines for reporting statistics in journals published by the American Physiological Society: the sequel". Adv Physiol Educ 31: 299, 2007. doi:10.1152/advan.00032.2007. Link | ISI | Google Scholar18. Ludbrook J. Comments on journal guidelines for reporting statistics. Clin Exp Pharmacol Physiol 32: 324–326, 2005. doi:10.1111/j.1440-1681.2005.04221.x. Crossref | PubMed | ISI | Google Scholar19. Mackenzie B. Sustained efforts should promote statistics literacy in physiology. A commentary on "Guidelines for reporting statistics in journals published by the American Physiological Society: the sequel". Adv Physiol Educ 31: 305, 2007. doi:10.1152/advan.00087.2007. Link | ISI | Google Scholar20. Morton JP. Reviewing scientific manuscripts: how much statistical knowledge should a reviewer really know? Adv Physiol Educ 33: 7–9, 2009. doi:10.1152/advan.90207.2008. Link | ISI | Google Scholar21. Rangachari PK. Statistics: not a confidence trick. A commentary on "Guidelines for reporting statistics in journals published by the American Physiological Society: the sequel". Adv Physiol Educ 31: 300–301, 2007. doi:10.1152/advan.00069.2007. Link | ISI | Google Scholar22. Wagner PD. Cores of Reproducibility in Physiology (CORP): Advancing the corpus of physiological knowledge. J Appl Physiol (1985) 122: 89–90, 2017. doi:10.1152/japplphysiol.01046.2016. Link | ISI | Google Scholar23. Wasserstein RL, Lazar NA. The ASA's statement on p-values: context, process, and purpose. Am Stat 70: 129–133, 2016. doi:10.1080/00031305.2016.1154108.Crossref | ISI | Google ScholarFOOTNOTES1The aforementioned papers and the first three papers in my Explorations in Statistics series are included in a Reporting Statistics collection which is available at http://advan.physiology.org/Reporting-Statistics.2The distinction between a standard deviation and a standard error is substantive: a standard deviation estimates the variability among observations in a sample, but a standard error of the mean estimates the uncertainty about the actual value of some population mean. This distinction has been emphasized repeatedly (1, 4, 6, 10, 11).3If the standard deviation is not a meaningful estimate of variability, as when the data are skewed, then the range or interquartile range provide more meaningful estimates.AUTHOR NOTESAddress for reprint requests and other correspondence: D. Curran-Everett, Div. of Biostatistics and Bioinformatics, M222, National Jewish Health, 1400 Jackson St., Denver, CO 80206–2761 (e-mail: [email protected]org). Download PDF Back to Top Next FiguresReferencesRelatedInformation CollectionsAdvances in Physiology Education CollectionsReporting StatisticsStatistics Cited ByAfferent renal innervation in anti-Thy1.1 nephritis in ratsKristina Rodionova, Roland Veelken, Karl F. Hilgers, Eva-Maria Paulus, Peter Linz, Michael J. M. Fischer, Martina Schenker, Peter Reeh, Gisa Tiegs, Christian Ott, Roland Schmieder, Mario Schiffer, Kerstin Amann, and Tilmann Ditting3 November 2020 | American Journal of Physiology-Renal Physiology, Vol. 319, No. 5Evolution in statistics: P values, statistical significance, kayaks, and walking treesDouglas Curran-Everett15 May 2020 | Advances in Physiology Education, Vol. 44, No. 2Statistical considerations in reporting cardiovascular researchMerry L. Lindsey,* Gillian A. Gray, Susan K. Wood, and Douglas Curran-Everett*8 August 2018 | American Journal of Physiology-Heart and Circulatory Physiology, Vol. 315, No. 2 More from this issue > Volume 41Issue 3September 2017Pages 321-323 Copyright & PermissionsCopyright © 2017 the American Physiological Societyhttps://doi.org/10.1152/advan.00049.2017PubMed28679565History Received 29 March 2017 Accepted 30 March 2017 Published online 5 July 2017 Published in print 1 September 2017 Metrics
Referência(s)