检验能力对统计显著性检验的影响

THE INFLUENCE OF TESTING POWER ON STATISTICAL SIGNIFICANCE TEST

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，探讨抽样样本总体达到ｔ检验显著性的数目与统计检验能力之间的一致性。模拟实验结果表明，统计检验能力与样本总体ｔ检验显著性的数目具有很高的一致性，但两者同时受到显著性水平α、样本容量ｎ和总体效果大小δ的影响，从而有可能影响统计推断的可靠性。因此，在进行显著性检验过程中，应对统计检验能力进行估计，这将有利于心理学研究成果的积累。

Using the computer to construct population subject groups to simulate experimental research procedures in a sampled study,we probed into the consistence between the number of population samples with a t-test significance and the statistical testing power.The simulated experimental result indicated that a high consistence existed between statistical testing power and the number of population samples with a t -test significance; yet, both of them were affected at the same time by the significant level a, the sample capacity n and the population effect size althus the reliability of statisticalinference might be affected. Therefore, in tbe process of statistical tests, the statistical testing power should be estimated for the accumulation of psychological research achievements.