Editorial pubs.acs.org/ac
Details Matter, and so Do Statistics
H
higher impact journals tend to have more retractions.4,5 Nonetheless, retracted articles still are very rare. Thankfully only a very small percentage of authors are involved in retracted studies. The much larger number of honest authors should ensure that they disclose all of the information needed to enable a fair evaluation of their work and to allow others to reproduce their experiments. Of course, this should be everyone’s goal when writing a research paper. The bottom line: make sure to provide all of the details needed to reproduce your work, including computer algorithms and statistical tests, and then deposit your raw data in the appropriate databases. I, and the scientists who read your papers, will thank you.
ow many times have you found an article that reports an exciting result but when reading it carefully, you discover that important information needed to evaluate and repeat what was done is missing? I find it especially frustrating when results are reported as significant but the statistical tests are not explained, and even worse, I cannot even find details on sample size and reproducibility. In an effort to address the issue of inconsistent method reporting, the Nature family of journals has just published several editorials,1,2 and released a set of guidelines and a checklist. The intention is to ensure that authors disclose enough information about their work, including a myriad of technical details, statistical tests, and even the raw data. Analytical Chemistry’s author guidelines and the ACS ethical guidelines already address most of these issues, but the Nature checklist goes further by asking that authors state exactly where in the manuscript the required details can be found during the submission process. Authors must provide access to any custom computer codes used for calculations, supply precise characterization of key reagents such as cell lines and antibodies, include detailed descriptions on sample size and justification for the statistical tests used, and comply with specific data deposition requirements. I certainly applaud such efforts and encourage all authors preparing an article for submission to any journal to look over the Nature guidelines and checklist. I am also considering what we need to do for Analytical Chemistry. Feel free to share your thoughts with me. It is important to note that the goal of journals in demanding such information is not to dictate science but to allow others to reproduce the science that is reported. In discussing these guidelines with colleagues, it was suggested that these requirements will prevent people from reporting results that have no repeats (the n = 1 experiment). Actually, the unique result still has a place in a scientific publication such as reports describing a single example of a new species of bird, sequencing portions of the genome from an extinct beetle trapped in amber, characterizing proteins from a dinosaur bone, or even describing a new hominoid skull. The point is not sample size per se but rather, that limitations in a study are clearly articulated (although for most studies, the sample sizes needs to be much larger). Of course, one reason for this policy appears to be to reduce retracted studies; a quick glance at Retraction Watch turns up some amazingly high-profile cases of large scale retractions during the past year. I was surprised by a study published in the Proceedings of the National Academy of Sciences3 that examined over 2000 articles in the PubMed database that were listed as being retracted by May 2012. The authors investigated the reasons for the retractions and concluded that fraud and suspected fraud accounted for 43% of the retractions, duplicate publication and plagiarism another 24%, but only about 21% were due to scientific error! My point is that if scientific misconduct is the major issue driving science retractions, perhaps requiring more detail during submission will not stop individuals willing to make up their results. Interestingly, the © 2013 American Chemical Society
■
Jonathan V. Sweedler AUTHOR INFORMATION
Notes
Views expressed in this editorial are those of the author and not necessarily the views of the ACS.
■
REFERENCES
(1) Enhancing reproducibility. Nature Methods 2013, 10, 367; http:// dx.doi.org/10.1038/nmeth.2471. (2) Announcement: Reducing our irreproducibility. Nature 2013, 496, (25 April), 398; http://dx.doi.org/10.1038/496398a. (3) Fang, F. C.; Steen, R. G.; Casadevall, A. Misconduct accounts for the majority of retracted scientific publications. Proc. Natl Acad. Sci. U.S.A. 2012, 109, 17028; http://dx.doi.org/10.1073/pnas. 1212247109. (4) Corbyn, Z. Misconduct is the main cause of life-sciences retractions. Nature 2012, 490, 21; http://dx.doi.org/10.1038/490021a. (5) Liu, S. V. Top Journals’ Top Retraction Rates. Scientific Ethics 2006, 1 (2),91−93; http://im1.biz/albums/userpics/10001/ SE2006V1N2A11_TopRetraction.pdf.
Published: May 21, 2013 5287
dx.doi.org/10.1021/ac4014513 | Anal. Chem. 2013, 85, 5287−5287