The Realism of Reliability : The Contradiction between Definition and Estimation of Reliability

Abstract

The purpose of this study is to show how the estimation of reliability from the real data is different from the definition of reliability.  Theoretically, reliability is defined as the ratio of true score and observed score variances.  From the definition of reliability, we can conclude that the reliability index ranged from 0 to 1.  In fact, there are many cases in which the estimation of reliability from the real data is different from the conclusion. Subsequently, there is no reason to limit the reliability index between 0 to 1.  Nowadays there is no reason to explain why the reliability index is less than -1.0.  This is the main issue for all measurers to find the reasons why there is a significant contradiction between the definition and estimation of reliability.