An analysis of common statistical errors in medical papers

Objective The common statistical mistakes seen in medical researches,including common misuse of t test and χ² test,are enumerated in this paper.The common misuse of t test includes t test instead of paired t test,t test instead of one-way ANOVA,and t test instead of a repeated measures analysis of variance.χ² test is incorrectly used instead of Fisher exact probability method,rank test,or paired χ² test.The case analysis on the common statistical problems in medical papers shows that the main reasons for misuse of statistical methods are the application premise error of a special statistical method,incorrectly choice of statistical methods divorced from the actual situation and the purpose of statistical analysis,not fully understanding whether the research data to meet the conditions of parameter testing,and incorrectly determine the measurement data corresponding to the experimental design type.To ensure the accuracy of statistical applications in a scientific journal,the editorial department should invite statistical review experts to control the quality of the manuscript; the editors should strengthen their own statistical professional quality,improve the level of statistical professional knowledge,deeply think and grasp the statistical analysis of the manuscripts in the process of editing manuscripts,timely look over the statistical data and consult the relevant statistical expert where you have a question.Finally,the authors should have more exchanges with the statistical experts in the research design,from the source to avoid the design of the loopholes and deviations.It is suggested to improve the application ability of statistics from the authors,editors and peer review experts to ensure the scientificalness,reliability and validity of scientific research,and reduce the induction of the error messages to readers.