相關係數與信度係數的關聯及問題

The Relationships and the Problems between Correlation Coefficient and Reliability Coefficient

林冠群(Guan-Chyun Lin) ； 吳裕益(Yuh-Yih Wu)

《測驗學刊》 52卷2期 (2005/12) Pp. 29-60

https://doi.org/10.7108/PT.200512.0029

摘要

利用相關係數是估計信度係數的一種簡捷方式，但也潛藏著許多可能的問題。統計模擬的結果顯示，組內相關係數是不錯的信度估計量，但皮爾遜相關係數較會有高估的情況。因此，使用任何方式估計信度時，皆需弄清它的假設與其目的，以及所需的實驗設計，才不會在未加以審視的情況下，產生應可避免的謬誤。

關鍵字

測驗 ； 信度 ； 相關係數

並列摘要

Applying correlation coefficient is a simple method to estimate the reliability coefficient, but many problems may sneak in it. The results of statistical simulation reveal that intraclass correlation coefficient is a good estimator of reliability, while Pearson correlation coefficient over estimates the reliability in many conditions. Therefore when we apply any method to estimate reliability, we must look at the assumption, the purpose and the required experimental design of it, otherwise some avoidable fallacy will arise from the inattentive situation.

並列關鍵字

test ； reliability ； correlation coefficient

參考文獻

Brennan, R. L.(2001).Generalizability theory.New York, NY:Springer-Verlag.

Choosing an Intraclass Correlation Coefficient

Safrit, M. J.(1976).Reliability Theory.Washington, D.C.:American Alliance for Health, Physical Education and Recreation.

Shavelson, R. J.,Webb, N. M.(1991).Generalizability theory: A primer.Newbury Park, CA:SAGE.

Shrout, P. E.,Fleiss, J. L.(1979).Intraclass correlations: uses in assessing rater reliability.Psychological Bulletin.86(2),420-428.