4-6年级多重成就测验的编制II:信度考验  

Multiple Achievement Tests for Pupils of Grades 4-6 II:Rialibility

作　　者：范晓玲[1] 龚耀先[1] 

机构地区：[1]中南大学湘雅二医院医学心理中心

出　　处：《中国临床心理学杂志》2006年第6期553-555,559,共4页Chinese Journal of Clinical Psychology

基　　金：湖南省教育厅科学研究项目(项目号:B/19-14020742)

摘　　要：目的:以CTT和GT作4-6年级MATs的信度检验。方法:重测信度161例,复本信度180例,重测复本信度200例。项目有效样本2002例,其中A本1004例,B本998例。结果:A、B本分量表和总量表重测信度0.91～0.95,复本信度0.87～0.94,重测复本信度0.82～0.89,分半信度0.79～0.90,系数0.90～0.96,评分者信度0.94～0.98,真分数变异0.82和0.86;概化研究表明分测验条目15个左右为宜,分量表与总量表条目量50与100即可。结论:各种信度考验结果基本符合测量学的计量标准。Objective： MATs reliability of the 4-6 Grades was tested. Methods： 161 subjects were given the MATs test-retest; 180 subjects, parallel test; and 200 subjects retest and parallel test. item samples consist of 2002 students. Results： The reliabilities of the subscales and full scales ranged from 0.91 to 0.95, in Spearman test-retest correlations, the parallel reliability was 0.87 to 0.94; the retest and parallel reliability 0.82 to 0.89; the split-half reliability 079 to 0.90 and the Cronbach＇s coefficient 0.90 to 0.96. The scorers reliabilities were 0.94-0.98. The true scores of MATs-A and MATs-B were 0.82 and 0.86, respectively. Generalization analysis showed that it was proper to have about 15 items in each suhtest, and 50 and 100 items in the suhscales and the fullscale. Conclusion： A series of reliahilities testing in the MATs conformed to the psychometric standards.

关 键 词：成就测验 信度 经典测验理论 概化理论 

分 类 号：G449.5[哲学宗教—心理学]