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 ›› 1997, Vol. 29 ›› Issue (04): 436-442.

• • 上一篇    

差数显著性t检验与元分析的对比研究

郭春彦;朱滢   

  1. 首都师范大学教科所,北京大学心理系
  • 出版日期:1997-12-25 发布日期:1997-12-25
  • 通讯作者: 郭春彦

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

Guo Chunyan Zhu Ying (Capital Normal University) (Beijing university)   

  • Online:1997-12-25 Published:1997-12-25

摘要: 利用计算机构造被试总体、模拟实验研究程序进行抽样研究,探讨显著性t检验方法与元分析方法在检验实验结果数据方面的差异。在模拟实验过程中,t验受到显著性水平、样本容量和总体效果大小的影响,因此最终影响了统计推断的可靠性,建议:在进行显著性检验过程中,应对统计检验能力进行估计;元分析方法以样本为元素对总体进行推断,因此具有很高的准确性和可靠性,它将很有可能成为今后心理学研究的重要统计工具。

关键词: 显著性检验, 元分析, 统计检验能力, 效果大小

Abstract: A sampled study through computer constructed population subject group and simulated experimental study procedures probed into the difference between significance t-test and meta-analysis in testing data rusults of experiments. In the process of simulated experiment, the t-test result was affected by significance level, sample capacity and population effect size. Thus reliability of statistical inferences was affected. Suggestions: In the process of significance test, statistical testing power should be estimated; inferring population subject group througlt samples as factors, meta-analysis has high accuracy and reliability, it will very likely be a very important statistical method in later research in psychological science.

Key words: significance t-test meta-analysis, statistical testing power, effect size