DINA模型等值的逻辑关系初探

发布时间：2018-03-25 00:05

本文选题：DINA模型　切入点：等值　出处：《江西师范大学》2014年硕士论文

【摘要】：认知诊断理论将认知诊断模型、认知心理学和学科教学研究成果相结合,实现了对被试内部心理加工成分、认知过程、认知成分或认知加工的测量。与经典测量理论和项目反应理论相比,能够提供更加详细的诊断信息,已经成为心理与教育测量学界最重要的研究领域之一。然而,要在大规模评价活动中应用认知诊断理论,首先需要解决的就是测验等值的问题。等值是进行大规模评价中经常用到的测量技术,以实现测验分数间的可比性为目的。在众多的认知诊断模型中,DINA模型(Deterministic Inputs, Noisy and Gate Model)由于其在拥有简洁的项目参数的同时,诊断准确率较高,已成为目前被应用的较为广泛的认知诊断模型之一。De la Torre和Lee(2010)指出,DINA模型(Haertel,1989; Junker和Sijtsma,2001)只有在以下两种情形里没有进行等值的必要：第一,模型和被试反应数据完全拟合；第二,两次测试下的被试特质分布大致相同(如属性分布类似等)。但是,在实际测验情境中,这两种条件基本无法满足。这也就成为了在基于DINA模型的大规模的心理或教育评价活动中需要进行等值的原因。本研究旨在探索DINA模型里等值的逻辑关系,为建立DINA模型的等值转换关系式打下基础。本研究分为6个实验,分别探讨了DINA模型中的项目的失误参数s与猜测参数g、被试的IMP的掌握概率与项目的失误参数s、被试的IMP的掌握概率与项目的猜测参数g之间的相互关系。主要研究结论如下： (1)项目的失误参数s与猜测参数g之间相互独立； (2)被试的IMP的掌握概率的变化(x)对项目j的失误参数S(ygi)的影响为：其中,sj为被试的IMP的掌握概率未变化时项目j的失误参数s的值。 (3)被试的IMP的掌握概率的变化(x)对项目j的猜测参数g(ygi)的影响为：其中,.9j为被试的IMP的掌握概率未变化时项目j的猜测参数g的值。 (4)项目j的失误参数s的变化(x)对被试的IMP的掌握概率(YPAi)的影响为：其中,PAi：项目j的失误参数s未变化时被试的IMP的掌握概率的值。 (5)项目j的猜测参数g的变化(x)对被试的IMP的掌握概率(YPAi)的影响为：其中,PAi：项目j的猜测参数g未变化时被试的IMP的掌握概率的值。
[Abstract]:The cognitive diagnosis theory combines the cognitive diagnosis model, cognitive psychology and the research results of subject teaching and learning to realize the mental processing and cognitive process of the subjects. The measurement of cognitive component or cognitive processing, which can provide more detailed diagnostic information than classical measurement theory and item response theory, has become one of the most important research fields in the field of psychological and educational measurement. In order to apply the cognitive diagnosis theory to the large-scale evaluation, the problem of test equivalence, which is often used in large-scale evaluation, is the first problem to be solved in order to achieve the comparability between test scores. Among the many cognitive diagnostic models, Noisy and Gate models (Noisy and Gate models) have higher diagnostic accuracy because of their concise project parameters. One of the more widely used cognitive diagnostic models. De la Torre and Leehs2010) points out that the Dina model is Haertell 1989; Junker and Sijtsmai 2001).) there is no need for equivalence only in the following two situations: first, the model and the test response data are fully fitted; and second, there is no need for equivalence in the following two cases: first, the model and the data of the test response are completely fitted; second, The distribution of traits in the two tests was roughly the same (such as similar distribution of attributes, etc.). However, in the actual test situation, These two conditions can not be satisfied, which is the reason for the need for equivalence in large-scale psychological or educational evaluation activities based on the DINA model. The purpose of this study is to explore the logical relationship of equivalence in the DINA model. It lays the foundation for establishing the equivalent transformation relation of DINA model. This study was divided into six experiments. The relationship between the project error parameter s and guess parameter g in DINA model, the grasping probability of IMP and the error parameter s of the project, the grasp probability of IMP and the guess parameter g of the project are discussed respectively. The main findings are as follows:. (1) the error parameter s and the guess parameter g are independent of each other; (2) the influence of the change of the mastery probability of IMP on the error parameter of item j is: where Sj is the value of the parameter s of item j when the probability of mastery of IMP does not change. (3) the influence of the change of the mastery probability of IMP on the guess parameter of item j is: the value of the guess parameter g of item j is the value of the item j when the mastery probability of IMP of the subject does not change. (4) the influence of the error parameter s of item j on the mastery probability of IMP was as follows: the value of mastery probability of IMP when the error parameter s of item j did not change. (5) the influence of the change of the guess parameter g of item j on the mastery probability of IMP of the subjects is as follows: the value of the mastery probability of the IMP of the subjects when the guess parameter g of item j does not change.
【学位授予单位】：江西师范大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：B842.1

【参考文献】