有关心理统计的三个疑问

Questioning Statistical Practice in Psychology

统计学在心理学研究中的作用似乎是确定无疑,然而,本文提出了三个疑问:心理量能统计吗?统计结果"显著"就足够了吗?统计模型与理论模型相距有多远?在对三个问题的分析中,试图说明:统计学作为心理学的工具,当然很重要,但更重要的是对心理进行有效测量,提高变量的测量水平和测量的有效性是心理学的根本任务。在数据分析过程中,应该加强对统计方法和结果的理论思考,纠正一些错误的统计规范和做法,弥补"零假设显著性检验"的不足。

Definitely statistics plays an important role in psychological research. Three basic questions,however,are proposed in the present paper: (1) Is it possible for psychological variables to be computed for statistical analysis? (2) Are statistically'significant'results enough? (3) How far is the statistical model from the theoretical model? It is suggested that the core task of psychological research is to improve the levels and validity of measurement rather than statistics,because the precondition of statistics is to collect high-quality data. Psychological researchers should give more theoretical concerns to the statistical approaches and results,and should correct some wrong statistical norms and practice. For example,numerous statisticians have pointed out the limited value of the approach of null hypothesis significance testing and its misuse and overuse,and they have also made practical suggestions to improve the approach.