Nature: the importance of studying repeatability

The objectivity of academic research results requires not only the professional ethics of researchers, but also the cooperation of academic community, which requires scholars to ensure the repeatability of papers as much as possible The so-called repeatability is that others can reproduce the research results with the schemes and materials provided by the author In fact, if we follow this standard, we don't need so many so-called research papers at all The development of science needs to be objective and real, and doesn't need any leap and rapid progress on the surface Therefore, to attach importance to and adhere to repeatability should become the iron rule in the academic field The repeatability of research papers is the most important foundation of scientific research The repeatability of the paper requires the author to provide a detailed description of the research process, research objects and statistical analysis methods, otherwise it will bring great difficulties to other scholars to repeat experiments, but the poor repeatability of biomedical research has been plagued in this field Some magazines have made some explorations in this field, but still can't avoid the problem of poor research repeatability Recently, the most prominent case is the Japanese scholars' event about STAP cells, and finally to the awkward situation that even the paper publishing institutions are difficult to repeat themselves Reproducibility, preciseness, transparency and independent verification are the cornerstones of scientific methods Of course, only relying on the reproducibility of the research results or the repeatability of the research, can not be used as a basis to judge a certain judgment is correct, nor can it be used as a standard to judge a certain judgment because the results are not repeatable Preciseness and transparency are the guarantee of repeatability Preciseness and transparency are the basis of ensuring the development of science Independent test is the most powerful tool and means to counterattack opponents and eliminate subjective factors In June 2014, editors from more than 30 biomedical academic journals, heads of funding institutions and some academic institutions discussed the principles and guidelines of preclinical biomedical research and how to achieve the goal of research repeatability at the headquarters of the American Association for the advancement of science The meeting was attended by NIH, nature and science The meeting discussed the research repeatability issues from the current methods used by the magazine, such as the effectiveness of these methods, to the cost of using these methods As the publication policy of these magazines, the standard lists a series of requirements for improving the transparency and repeatability of research     It is suggested that academic journals should inform the authors of relevant policies on statistical analysis and evaluation of statistical accuracy The length of papers should not affect the repeatability evaluation due to the limitation of the number of printed pages It is also suggested that the use of checklists should be encouraged to ensure that papers can provide important experimental data, such as standard use, repetition types and times, statistical methods, random grouping methods, whether or not Blind method, sample size, data selection or exclusion criteria, etc The journal should also suggest that the author upload the data to the public database, and make a two-way link with the paper after publication In particular, the journal should encourage all experimental materials to be shared with peers who wish to repeat the experiment One of the relatively loose suggestions in the specification is about the identification of image data, such as screening other people for forged images, storing high-resolution documents, etc., and the other is about the requirements of comprehensive description experiment For example, animal experiments must provide animal sources, species, gender, age, transgenic animals must give a detailed description of the source of genes With regard to cell research, the author must provide information on cell source, identity and mycoplasma contamination Although these specifications do not guarantee the requirements of repeatability and independent verification, they can make it easier to repeat these studies Some magazines have implemented all these standards before However, it is still of great significance for so many magazines to jointly implement this standard Journal unit for reproducibility Nature, 06 November 2014; doi: 10.1038/515007a