“假设检验”决策的误区——一场由全国大学生数学建模竞赛引发的争论

A Frequently Mistaken Concept in "Hypothesis Test" ——A dispute triggered by China Undergraduate Mathematical Contest in Modeling

针对2012年全国大学生数学建模竞赛A题第一问中"两组评酒员的评价结果是否有显著性差异"的问题,研究了湘、鄂两省答卷中解决该问题的主流评价方法——方差分析和配对t检验的方法,并对两种方法进行了对比和说明。

In the 2012 China Undergraduate Mathematical Contest in Modeling （CUMCM） problem of A, the first question is ＂analyzing two sets of evaluation of wine member of the evaluation results to see if there is significant difference＂. Analysis of variance （ANOVA） and Paired Sample t test （ t- pa） are two main methods in solving this problem in Hunan and Hubei area. But there is a misunder- standing in evaluating the significance of difference between two groups of assessment result. This arti- cle gives a detailed comparison and description over the two methods.