MacArthur R D, Jackson G G
J Infect Dis. 1984 Mar;149(3):349-54. doi: 10.1093/infdis/149.3.349.
One hundred fourteen articles published in the Journal of Infectious Diseases in 1982 were evaluated for the occurrence of eight commonly made statistical errors. Seventy-one percent of Original Articles and 50% of presentations in the Data Forum used statistical methods to analyze results. Almost all of the articles that used statistics contained at least one statistical error. The most common inadequacy, which occurred in 95% of the articles with statistical data, was the statement of a probability value without a complete summary of the statistical results. The most common error was the failure to include a correction for multiple comparisons. These results suggest that a more clearly stated statistical policy, a more explicit set of instructions to authors, and closer editorial attention to statistical methodology, perhaps at the prepublication phase, would improve the validity of articles published in the Journal.
对1982年发表在《传染病杂志》上的114篇文章进行了评估，以检查是否存在八种常见的统计错误。71%的原创文章和50%的数据论坛报告使用了统计方法来分析结果。几乎所有使用统计方法的文章都至少包含一个统计错误。最常见的不足之处是在95%有统计数据的文章中出现的，即只陈述概率值而没有对统计结果进行完整总结。最常见的错误是未对多重比较进行校正。这些结果表明，制定更明确的统计政策、为作者提供更明确的一套说明，以及编辑在出版前阶段对统计方法给予更密切的关注，将提高该杂志发表文章的有效性。
