论文部分内容阅读
讨论了显著性、效果量和原因量在研究设计和统计分析中的意义与作用。指出 ,不应将显著性检验中规定的 0 .0 5水平绝对化 ,对刚刚高于或刚刚低于此水平的研究结果 ,应持谨慎或宽容的态度。同时 ,根据Pollard(1993)的分析 ,显著性水平提供的是I型错误的条件先验概率和I型错误的整体先验概率的上限 ,而不是I型错误的条件后验概率。因此 ,应注意先验概率和后验概率的区分。效果量与显著性有本质区别 ,研究者不仅应关注和报告显著性 ,也应关注和报告效果量。多大的效果量才有实际意义 ,取决于研究者对特定研究领域特定研究问题的专业判断。研究者进行研究设计时 ,还必须考虑效果量和原因量的关系问题 ,即因果效能问题 ,设法用最小的代价产生最大的效益
Discusses the significance of the amount of effect and the amount of research in the design and statistical analysis of the significance and role. It was pointed out that the 0.05 level set forth in the significance test should not be absolute and should be cautious or tolerant of findings just above or just below this level. In the meantime, according to Pollard (1993), the significance level provides the upper limit of the priori probability of type I error and the overall prior probability of type I error, not the conditional posterior probability of type I error. Therefore, attention should be paid to distinguish between prior probability and posterior probability. There is an essential difference between effectiveness and significance. Researchers should not only pay attention to and report on the significance, but also pay attention to and report on the effects. The magnitude of the effect is of practical importance and depends on the professional judgment of the researcher on a particular research question in a particular field of study. When investigating and designing, researchers must also consider the relationship between the amount of effect and the amount of cause, that is, the problem of causal efficiency, and try to produce the maximum benefit with minimum cost