摘要
与经典教育测量方法相比,基于项目反应理论(IRT)的教育统计与心理测量技术呈现出愈来愈多的优势.将Copula方法引入到IRT中来分析相依反应数据,对题目的边际反应概率和题目反应间的相依结构分别建模,更好地解决了项目局部相依性问题.
In educational and psychological measurement, most item response theory models are not robust to violations of local independence. A new class of models that makes use of Copulas to deal with local item dependencies is introduced. These models belong to the bigger class of marginal models in which marginal and association structure are modeled separately. It is shown how this approach overcomes some of the problems associated with other local item dependency models.
出处
《东北师大学报(自然科学版)》
CAS
CSCD
北大核心
2017年第2期41-46,共6页
Journal of Northeast Normal University(Natural Science Edition)
基金
国家自然科学基金资助项目(11201313
11571069
11501094
31400897)