摘要
P值是教育研究中十分重要的推断依据,但在使用过程中存在诸多问题.通过对241篇属于教育领域的CSSCI期刊(2017-2018)中有关P值使用情况的统计发现,这些问题主要包括:P值错误表达,根据P值选择显著性水平,忽视样本对P值的影响,对P值及其相关指标的报告不够全面等.为此,P值的使用应该理解P值及相关概念,在计算P值之前给出α,进行重复性试验并且引入效用量,尽可能全面地报道实验结果.此外,效应量作为一种度量效应大小的重要指标,常为教育研究者所忽视.为此,在教育研究中可以引入5类效应量,通过P值来做定性判断,通过效应量做定量判断,提高统计推断的准确性和教育研究的科学性.
P-value was an important tool for statistical inference in educational research,but there were many problems in its application.Based on statistical analysis of the use of P-value in the 241 articles belonging to the CSSCI journals in the field of education(2017—2018),it was found that these problems in the use of P-value,including P-value error expression,selecting the significance level by the P-value,ignoring the effect of samples to P-value,incompletely reporting the information of P-value and related indicators,etc.Therefore,it was necessary to understand P-values and related concept before using P-values,give alpha before calculating P-value,repeatability test and introduce the effect size,provide full reports as much as possible.Besides,as an important index to measure the effect size,the effect size was often ignored by educational researchers.Therefore,in order to improve the accuracy of statistical inference and the scientificity of educational research,five types of effect size could be introduced to educational research,both qualitative judgment and quantitative judgment should be considered,by using P-value and effect size.
作者
沈光辉
范涌峰
陈婷
SHEN Guang-hui;FAN Yong-feng;CHEN Ting(School of Mathematics and Statistics,Southwest University,Chongqing 400715,China;Southwest University Branch Center of National Innovation Center for Assessment of Basic Education Quality,Chongqing 400715,China;Faculty of Education,Southwest University,Chongqing 400715,China)
出处
《数学教育学报》
CSSCI
北大核心
2019年第4期92-98,共7页
Journal of Mathematics Education
基金
中国基础教育质量监测协同创新中心重大成果培育性项目——高中监测制度建设及监测方式研究（2019-06-021-BZPK01）
中国基础教育质量监测协同创新中心重点培育项目——西藏小学生数学学业质量监测（2018-06-020-BZPK01）
中国博士后基金面上项目——学校特色发展水平常模构建研究（2017M622934）
中央高校专项资金项目——西藏初中生数学学业质量监测研究（SWU1709664）
关键词
P值
问题
假设检验
效应量
P-value
problem
hypothesis testing
effect size
