Abstract: Based on the idea of "assessment for learning" and aiming at promoting students' learning, the assessment pattern of objectively quantifying the learning status and providing diagnostic feedback has been increasingly valued. Cognitive diagnostic assessments (CDAs) aim to provide individualized diagnostic feedback for each student, clarify students' learning status, and provide a reference for follow-up students' self-remedy and teachers' targeted remedial teaching. To achieve the ultimate goal of effectively promoting students' learning, the effectiveness of remedial teaching based on cognitive diagnosis is very important. However, currently, limited studies focus on evaluating the effectiveness of remedial teaching. One of the main reasons is that the cross-sectional design is adopted by most CDAs, which means that only one assessment is done at a specific testing time point. This issue may also reflect on the development of existing cognitive diagnosis models (CDMs). Although various CDMs have been proposed, most of them are only applicable to cross-sectional data analysis. In recent years, the longitudinal data collected from the CDAs throughout the learning process provides researchers with the chance to develop learning diagnosis models, that can be adapted to track individual growth over time and to evaluate remedial teachings' effectiveness. Compared to the cross-sectional CDA, longitudinal CDA is more helpful when aiming to promote students' development. Under such a background, in longitudinal CDAs, the effect analysis of individualized remedial teaching is a very interesting topic. To explore the effectiveness of individualized remedial teaching based on the longitudinal CDA, this research is divided into four stages: test development stage, preliminary testing stage, formal testing stage, and individualized remedial teaching stage. In test development stage, two parallel cognitive diagnostic tests were compiled by taking the chapter of "linear equation with one unknown" as an example, including the Volume A and Volume B. Two parallel tests both require six attributes and have the same Q-matrix (the order of the rows in two Q-matrices is different), including (A1) basic concept of linear equation; (A2) algebra rules; (A3) solving linear equation; (A4) recognition of implicit conditions; (A5) simple application of the linear equation; and (A6) complex application of the linear equation. In the preliminary testing stage, there are 74 seventh-grade students take part in the testing. The two tests are combined into one and the test time is 80 minutes. Based on the analysis via classical test theory, the results indicated that two tests both have good quality and meet the requirements of parallel tests. In the formal testing stage, there are 329 seventh-grade students (85 in the urban area and 244 in the rural area) take part in the testing. Volume A is used and the test time is 40 minutes. Three CDMs, including the DINA, DINO, and GDINA models, were used to fit the data. The model-data fit indices (AIC and BIC) preferred the DINA model as the best fitting model. Further, the results of the DINA model found that the mastery of attributes by urban students is better than that by rural students. In the individualize remedial testing stage, 22 rural students were sampled and divided into two groups on average: the CDA-based remedial teaching group and the traditional remedial teaching group. In the first group, students received one-to-one remedies from the teacher according to the diagnostic feedback that focuses on non-mastery attributes. By contrast, students in the second group received one-to-one remedies from the teacher according to the incorrect item responses. By comparing the effect of CDA-based individualized remedial teaching and traditional remedial teaching, it is found that both remedial methods can improve students' performance and the effect of CDA-based individualized remedial teaching is significantly better than that of traditional remedial teaching(F(1) = 51.62, P < .001). Overall, the results of this study indicated that the CDA-based individualized remedial teaching can effectively promote students' learning. The results of this study also provide a practical basis for practitioners to use CDA to promote students' learning.

Key words: cognitive diagnosis, individualized remedial teaching, assessment for learning, cognitive diagnostic model

摘要： 基于“为学习而测评”理念，以促进学生学习为目的，本研究进行了基于认知诊断测评的个性化补救教学效果分析。首先，以“一元一次方程”章节为例，编制两份平行的认知诊断测评试卷。然后，通过对不同地区（城市和农村）七年级学生的施测与分析，发现城市学生对属性的掌握情况优于农村学生对属性的掌握情况。之后，选择农村学生为补救对象，通过对比基于认知诊断测评和传统教学两种个性化补救教学的效果，发现两种补救教学方法均能提高学习成绩，但前者的补救效果显著优于后者的。总之，本研究结果表明采用基于认知诊断测评的个性化补救教学能够有效促进学生学习，为实践者应用认知诊断测评促进学生学习提供了实践依据。

关键词: 认知诊断, 个性化补救教学, 为学习而测评, 认知诊断模型