混合IRT模型的研究进展与应用  

Research Progress of Mixture IRT Models and its Applications

作　　者：黄明明[1] 王立君[1] 

机构地区：[1]浙江师范大学教师教育学院

出　　处：《考试研究》2015年第4期61-68,共8页Examinations Research

摘　　要：传统的项目反应理论模型由于不能很好地处理非连续数据而影响了对具有潜在类别属性的特质进行精确估计。混合项目反应理论不仅能够精确地估计项目参数和能力参数,而且可以实现按照不同类别属性的潜在特质与行为对被试进行自动鉴别。随着研究的发展,混合分部评分模型、混合Logistic模型、多水平项目反应理论模型以及带协变量的混合项目反应理论模型等相继诞生,并在教育测验分析与编制、项目功能差异分析以及其他拓展性实践应用中展现出优良的品质。开发多维混合项目反应理论模型、多维混合认知诊断模型以及混合题组模型等并对其进行本土化研究与应用将是混合项目反应理论的一大研究热点与方向。The traditional item response theory models can＇t deal with the inconsistent data which has influenced the accurate estimation of trait with latent category. Mixture IRT model can not only estimate the item and ability parameter but also identify the subjects according to the latent trait with different categories automatically. With the improvement of study about mixture IRT model, mixture partial credit model, mixture Logistic models, multilevel mixture IRT models and mixture IRT models with covariant have been found out, they all usually have advantages to he used to analyze and form the educational and psychological tests, detect DIF and other extended practical application. Developing the multidimensional mixture IRT models, multidimensional mixture cognitive diagnosis models and mixture testlet models will be hot issues and study directions.

关 键 词：项目反应理论 混合项目反应理论模型 混合Rasch模型 潜在类别分析 潜在特质 

分 类 号：G424.74[文化科学—课程与教学论]