ISSN 0439-755X
CN 11-1911/B
主办:中国心理学会
   中国科学院心理研究所
出版:科学出版社

心理学报 ›› 2002, Vol. 34 ›› Issue (02): 46-50.

• • 上一篇    下一篇

差数显著性t检验与元分析方法的模拟对比

郭春彦;朱滢;王全珍   

  1. 首都师范大学教科院,北京大学心理系,北京行政学院 北京100037 ,北京100871 ,北京100044
  • 发布日期:2002-04-25 出版日期:2002-04-25
  • 通讯作者: 郭春彦

A SIMULATED COMPARATIVE STUDY OF SIGNIFICANCE T-TEST AND META-ANALYSIS

Guo Chunyan 1, Zhu Ying 2 , Wang Quanzhen 3 ( 1Institute of Educational Sciences,Capital Normal University, Beijing 100037 ; 2Department of Psychology, Beijing University,Beijing 100871; 3English Department , Beijing Administrative College,Be   

  • Online:2002-04-25 Published:2002-04-25

摘要: 利用计算机程序构造被试总体、模拟实验研究程序进行抽样研究 ,探讨差数显著性t检验方法与元分析方法在检验实验结果数据和进行实际应用方面的差异。模拟实验结果表明 ,差数显著性t检验与总体效果大小和样本容量有明显关系 ,但与随机抽样分布样本数目基本无关 ;抽样分布样本效果大小的平均值可以作为总体效果大小的估计值 ,它与样本容量和抽样样本数目有密切关系 ;本研究提供的实际应用数据结果可以作为研究者进行元分析方法实验研究的参考依据

关键词: 元分析, 显著性检验, 效果大小

Abstract: Computer programs were used to construct overall population and simulated experimental procedures were applied to conduct sampled studies in order to probe into the differences between the methods of significance t testing and meta analysis in testing the data of an experimental result and its practical application. The simulated experimental results showed that significance t testing is highly related to population distribution effect size and sample capacities, yet rarely related to the numbers of randomly selected samples; mean value of the selected samples effect size can be used as an estimated value of population effect size, and it is closely related to sample capacities and numbers of selected samples; the data result of practical application provided in this study can be used as a basic reference for researchers who are engaged in an experimental study of meta analysis method.

Key words: meta analysis, significance t test, effect size