通过Mplus计算几种常用的测验信度  

Estimating Test Reliabilities Using Mplus

作　　者：王孟成[1] 叶宝娟[2] 

机构地区：[1]广州大学心理与脑科学研究中心,广州510006 [2]江西师范大学心理学院,南昌330022

出　　处：《心理学探新》2014年第1期48-52,共5页Psychological Exploration

基　　金：国家社会科学基金教育学国家青年项目(CBA130124)

摘　　要：Cronbach’s alpha系数作为信度估计指标存在诸多弊端。为了克服其不足,研究者提出了多种信度估计,而流行的统计软件尚未直接提供这些参数,以致在实践中并未被广泛采用。为了缩小理论和实践的差距,文章通过具体实例给出几种常用的信度估计(合成信度,单个指标信度和ωh)的Mplus程序。Cronbach＇ s alpha as a reliability estimate resides in many shortages. To overcome those drawbacks, researchers have introduced a plentiful of alternative estimators of reliability. At present, however, those alternative reliabilities are not available in popular statistical packages（ e. g. ,SPSS） ,leading to less adopted by applied researchers. In order to narrow the gap between theory and practice,in this paper, several popular estimators of reliability （composite reliability;single item reliability and reliability ωh ）were demonstrated though analyzing an example using Mplus.

关 键 词：信度 Cronbach's ALPHA 合成信度 信度ωh Mplus 

分 类 号：B841.2[哲学宗教—基础心理学]