稳健统计方法在外语教学研究中的应用

Application of Statistical Robust Method to Research into Foreign Language Teaching

满足统计分析假设是正确使用常规统计分析的必要条件。当常规统计分析假设违背时,稳健统计方法相对于常规的统计分析能够更精确地估计总体参数。利用外语教学研究中的真实数据,阐述在违背常规统计分析假设的情况下比较独立组平均数时一个稳健t检验和两个稳健方差分析与对应的常规t检验和常规方差分析统计结果的差异性,可以使外语教学研究者对稳健统计方法有一个基本的了解,给予常规统计分析假设应有的重视,选择恰当的统计分析方法。

To meet the statistical analytical assumption is the necessary condition for the proper use of conventional statistical analysis.When the assumptions of a conventional statistical analysis are violated,robust methods can give relatively accurate estimates of population parameters as compared to corresponding conventional statistical analyses.This article compares inde-pendent means by conducting not only one robust t test and two robust ANOVAs but also the corresponding conventional t test and ANOVAs with the real data in foreign language teaching research in an attempt to illustrate the differences in statistical re-sults between them when the assumptions of a conventional t test or ANOVA are violated.It is intended to enable language teaching researchers to have a basic understanding of robust methods,and to have due regard for the assumptions of a conven-tional statistical analysis so as to select an appropriate statistical method.