国际应用语言学期刊效应值报告与解释现状研究

Effect Size Reporting and Interpreting Practices in International Academic Journals in Applied Linguistics

针对显著性检验存在的问题，美国心理学会出版手册第6版和国外社会科学领域的不少期刊都要求将定量研究报告效应值作为补充。分析6种重要的国际应用语言学期刊发表于2012-2014年的定量数据分析文章效应值的报告与解释现状，研究发现：6种期刊共239个统计检验平均有66％报告了效应值，其中53％对效应值进行了解释，仅1．3％报告了效应值的置信区间；相比相关分析和回归分析，t检验和非参数检验效应值报告比例较低；研究者存在过度报告Cohen’s d、η^2p等效应值的倾向；相比美国心理学会出版手册，期刊编辑对效应值报告的要求更为有效。

Considering the problems related to null hypothesis significance testing, the 6th APA Publication Manual （2010） and the editorial policies of many journals require the reporting of effect sizes （ES） to supplement the findings from inferential statistical analyses. This study investigated the effect size reporting and interpreting practices in the quantitative articles in six international applied linguistics journals from 2012 to 2014：. Overall, 66% of the statistical analyses reported effect size--53% of which interpreted effect size and only 1.3% of which reported the confidence intervals for the effect size. t tests and nonparametrie tests had much lower ratios of effect size reporting than correlation and regression analyses. Cohen＇s d and η^2p were reported much more frequently than other effect sizes. The 6th APA Publication Manual had some effect on ES reporting, and the requirements of journal editors were more effective.