2021/4/26
What does a reliability of .90 mean? Analyses of twenty-three psychological measurements for the purposes of demonstration
系上很榮幸地邀請銘傳大學諮商與工商心理學系顏志龍老師至系上演講，演講相關內容如下：
===========================================
時間：4/30（五）13：10
地點：社科院北棟2樓心理系階梯教室
講題：What does a reliability of .90 mean? Analyses of twenty-three psychological measurements for the purposes of demonstration
講者：顏志龍老師(銘傳大學 諮商與工商心理學系教授)
摘要：
Analogy between statistical index and real-life experience is useful to increase scientists’ understanding of some concepts. A good example would be Rosenthal and Rubin’s (1982) binomial effect size display (BESD) which demonstrated that an r2 value of only .10 was surprisingly equivalent to increasing 32% success rate of an experimental treatment. In the past, psychologists tend to rely on rules of thumb to define the acceptable criterion for reliability (e.g., .70 to .90), This rule of thumb was not empirically verified. In this study, we analogized reliability to real-life experiences to illustrate the meaning of a reliability value. Large scale data of twenty-three psychological measurements were used to illustrate the manner in which reliability coefficients manifest in the real world. For instance, if Beck Depression Inventory II with a reliability of .90 is being used as an assessment tool, the 95% confidence interval of a testee's estimated true score would cover three assessment categories of mild mood disturbance, borderline clinical depression, and moderate depression. In summary, this study shows that the previous criteria for reliability in the field of psychology may occur measurement error larger than psychologists expected. Some practical suggestions are provided to address this issue.
===========================================
敬請準時出席，謝謝。