基于项目反应理论的贝叶斯统计推断方法在教育中的应用

The Application of the Bayesian Statistical Inference Method Based on Item Response Theory in the Subject Education

与经典教育测量方法相比,基于项目反应理论的教育统计与心理测量技术呈现出愈来愈多的优势。目前,项目反应理论核心框架主要基于频率统计方法。然而,贝叶斯统计方法已逐渐显示出巨大的潜力和吸引力,尤其适用于对统计建模的深层改进。鉴于项目反应理论模型的复杂性和应用广泛性,国外已经成功开发了一些常用的软件包。然而这些软件包本身也存在一些缺陷和问题。因此,在取得的理论研究成果(构建新的模型、提出新的估计技巧)的基础上,亟待配套研发一套拥有知识产权的软件平台,并应用于教育部基础教育质量监测中心中小学教育质量(数学)的纵向跟踪调查数据,为相关职能部门决策提供科学依据。

Compared with the classical education measurement methods,the educational statistics and psychological measurement technology based on item response theory show more and more advantages.Item response theory is currently based on the frequency statistical approach.However,a Bayesian statistical method has gradually shown its great potential and attractiveness,which is especially suitable for the improvement of the statistical models.In view of the complexity and the extensive application of item response theory models,a number of software packages abroad have been successfully developed.However,these packages also have some defects and problems.Therefore,it is essential todevelop a set ofsoftware platformpossessed intellectual propertyrights based on previous theoretical results(new model construction and new estimation methods proposed).And the software platform will be applied to the longitudinal data tracing and survey in the national assessment for the education quality of the primary and secondary education(Mathematics),which will help toprovide scientific basis for decision-makingin the relevant departments.