摘要
以一种新的量表分析方法——非参数项目反应理论模型(NIRT)分析艾森克人格问卷情绪稳定性分量表,并与参数项目反应理论模型(PIRT)进行比较。统计检验结果显示量表中24个项目与Mokken模型拟合,与2PLM有4个项目不拟合,Mokken模型可准确分析这4个项目。根据Mokken模型拟合程度删除项目后,平均项目信息量的增量明显高于根据2PLM拟合程度删除项目后的增量。分析过程展示了非参数项目反应理论模型用于分析人格量表的可行性、优势和适用性。
The study analyzes Neuroticism scale (N scale)of Eysenck Personality Questionnaire (EPQ)using Nonparametric Item Response Theory(NIRT) model ,and compares NIRT model with Parametric Item Response Theory(PIRT) models. The results show that the 24 items in N scale all fit the Mokken model and 4 items don' t fit the 2PLM,while these 4 items can be accurately measured using Mokken model. After removing the misfit items based on NIRT model, the increment of mean item information is much higher than that based on 2PLM model. The analytic procedure and results reveal the feasibility, advantage and applicability of NIRT model in analyzing personality questionnaire.
出处
《心理学探新》
CSSCI
北大核心
2015年第1期62-67,共6页
Psychological Exploration
基金
全国教育科学"十二五"规划教育部重点课题(GFA111009)
广州市基础教育学业质量监测系统项目(GZIT2012-ZB0292)