三种心理测量理论的信度观

RELIABILITY VIEWS OF THREE PSYCHOLOGICAL MEASUREMENT THEORIES

目前，心理测量领域中主要存在三大理论派别。本文分别对这三种理论即经典测验理论、可概括性理论和项目反应理论作了简要介绍，着重分析这三种理论的信度观。文章讨论了这三种信度观的理论基础和研究方法，比较了它们的异同，指出经典测验理论存在的一些不足及概化理论和项目反应理论所作的改进。概化理论是对经典测验理论的扩展，它用多维的信度指标（概化系数）替代了经典测验理论的信度系数，项目反应理论则从信息量的角度出发，用项目信息函数、测验信息函数等指标更具体深入地反映项目、测验的测量可靠程度。

Nowadays,there are three major theories of psychological measurement. The paper gives a brief introduction to these three theories: classical test theory, generalizability theory and item response theory, and discusses in detail their different views of reliability. The paper points out some shortcomings of the classic test theory and the improvement of the other two theories. Generalizability theory,considered the extension of the classical test theory, replaces the classical reliability coefficient with more changeable generalizability coefficient. The item response theory uses item information function, test information function and other indexes to describe the precise degree (reliability)of single items and the whole test.