【摘要】 影响统计检验力的因素包括研究设计因素、研究工具因素和统计学因素。统计检验力分析是实验设计中非常重要的一部分:先验统计检验力分析可以帮助研究者在实验开始之前确定样本量以节约人力物力;后验统计检验力分析可以在研究完成之后帮助研究者审视研究效力,为后续研究提供帮助。当研究问题或实验设计较为复杂时,可借助Optimal Design设计多阶层统计检验力分析。建议在本科及研究生阶段重视统计检验力分析的教学,在科研中注重统计检验力分析的应用,以优化实验设计并增加研究结果的可靠性。更多还原

【Abstract】 Factors that can affect power could be design factors, instrumental factors, and statistical factors. The power analysis is important to a study. To be more specific, the priori power analysis can help researchers to determine the necessary number of participants needed to detect the differences between treatments, which can help researchers save money and efforts, before a study has been conducted. In addition, a post-hoc power analysis could help researchers evaluate the experimental design, which is helpful for the future use. There are some problems in the application of power analysis. It is hard to conduct the power analysis when the research or the design is complicated. In addition, there are some misinterpretations of the analysis results. Furthermore, there are limit numbers and features of software that can conduct power analysis. One example of conducting a multilevel power analysis has been shown with Optimal Design. The teaching of power analysis has been recommended to undergraduate and graduate students. Researchers should also get the importance of the application of power analysis to optimize the experimental design and improve the reliability of research results.更多还原