Scoring Method Based on IRT Models with Auxiliary Information and Optimal Sequencing of Test Items for CAI

提出了项目反应理论中的几个新模型和测试项目最优排序的方法。将被试的认知过程分为三种类型：(A)完全、知识、(B)部分知识和(C)无知，提出了一个新的项目反应模型，并将其应用于日语词汇测试。在上述模型的基础上，开发了硕士研究生和非硕士研究生的统计程序。提出了一种考虑试题信息量和受试者掌握程度的试题最优排序新方法。提出了另一种新的项目反应模型，该模型以正确/错误反应和有无自信反应为两个标准变量，使项目或被试能够从多个方面进行描述。提出了一种新的以时限为辅助信息的项目反应模型。以上模型均可视为潜在变量模型。讨论了这些模型中常见的一些统计问题。

Several new models in item Response Theory and methods for optimal sequencing of test items were proposed.1. Dividing subjects' cognitive process into three kinds ; (a) full , knowledge, (b) partial knowledge and (c) ignorance, a new item response model was proposed and applied to Japanese vocabulary test.2. Based on the above model, a statistical procedure was developed to classify masters and non-masters.3. A new procedure for the optimal sequencing of test items was developed, which takes into account the item information amount and the subjects' mastery level.4. Another new item response model which has two criterion variables, namely, correct/wrong response and "confident or not" response was proposed so that the items or subjects could be described by several aspects.5. A new item response model with the time limit as auxiliary information was proposed.6. All above models can be considered as latent variables models. Some Statistical problems common in these models were discussed.