基于认知诊断的儿童数学学力结构及测评研究

发布时间：2018-02-26 16:28

本文关键词： 认知诊断儿童数学学力学力结构测评　出处：《内蒙古师范大学》2017年博士论文　论文类型：学位论文

【摘要】：“学力”是动态、发展的,对其内涵的解读与结构的测评,始终会受到特定时代理想与教育需求的双重制约。时至今日,伴随着学力观的本质超越与外延拓展,数学学力在基础教育综合质量测评研究中的地位也不断得到提升。而由此所引发的则是世界各国日益重视对儿童数学学力发展的大规模动态监测与过程性评估。在这一时代背景之下,新一代心理与教育测量理论的代表——认知诊断理论,逐渐显露出其特有的优势,即能够对测验总分背后所隐藏的内部心理加工过程进行更细致、准确的探测,从而实现对儿童个体认知水平的诊断和群体能力发展特征的比较。本研究以正处于基础性数学学力发展关键期的6?12岁儿童为研究对象,以其数学学力的发展水平与结构性特征诊断为研究目的,按照现代认知诊断理论的研究范式,首先构建了儿童关键数学学力认知模型;然后据此编制了《儿童数学学力认知诊断测验(1?6年级)》;最后对随机抽取的6所学校共8289名1?6年级儿童的数学学力发展现状进行了大规模实测。此外,为了更深入地分析儿童数学学力落差形成的原因,还自编《儿童数学学力影响因素调查问卷》(包括教师和家庭两个分问卷),并进行了同步调查。最终得到如下结论:(1)所构建的儿童数学学力认知模型基本完备且合理。本研究不仅从儿童数学问题解决的心理加工过程、《全日制义务教育数学课程标准(2011版)》和《全日制小学数学教科书(人教社,2013版)》等多方面,论证了该模型的理论基础,而且还基于实测数据对其完备性与合理性进行了量化验证。结果显示:所构建之认知模型具备了理论分析和实测数据的双重佐证。(2)自编《儿童数学学力认知诊断测验(1?6年级)》基本达到了测量学要求。本研究综合运用CTT、IRT和CDT等不同方法,对六个年级诊断测验的质量进行了综合分析。结果表明:自编诊断测验可以为进一步研究儿童数学学力发展提供较为稳定且可靠的信息源。(3)所选特定认知诊断模型对实测数据的分析结果较为有效且可信。本研究分别从宏观的儿童数学学力水平和微观的关键数学学力属性掌握模式及概率等维度,对各学校儿童数学学力的发展特征进行了比较分析。结果表明:所选广义诊断模型(General Diagnostic Model,GDM)与大规模实测数据具有良好的拟合性,且对各关键数学学力属性的分类准确性与一致性均较高。(4)对儿童数学学力影响因素的分析结果与理论假设基本一致。本研究分别从儿童智力、非智力、学业成就,其及家庭、教师等多方面因素,对儿童数学学力的影响因素结构及其路径关系进行了分析。结果表明:各年级儿童数学学力与其数学学业成绩均呈现极其显著的正相关,而其他各影响因素对数学学力的影响关系则在不同年级段呈现出不同的特点,但均与本研究理论假设基本一致。(5)基于认知诊断结果的教学补救建议具有较强的可操作性。本研究针对认知诊断测评中所发现的主要问题,分别从学校教育和家庭影响两个方面,对不同年级段儿童提出了一些兼具理论性与可操作性的教学补救建议。综上所述,本研究立足于新一代测验理论——认知诊断理论,对儿童数学学力的测评方法与实践路径进行了较为广泛的探索,并最终实现了对小学六个年级段儿童数学学力的跨年级参数等值与诊断分析。研究的创新之处主要体现在以下四个方面:第一,以对基于学习心理结构的儿童数学学力结构观的深入分析为依据,建构了具有可操作性的儿童关键数学学力认知模型,对于深化儿童数学认知诊断研究具有一定的理论价值与实践意义。第二,自编《儿童数学学力认知诊断测验(1?6年级)》,实现了对小学全年级段儿童数学学力的整体认知诊断与测评,既可以为后续相关研究提供必要的方法借鉴,也可以为认知诊断技术的实践推广积累必要的实践经验。第三,自编R语言程序,进行的基于大规模实测数据的认知诊断研究,可以为国内教育认知诊断测评实践探索新的技术路径。第四,基于IRT的儿童数学学力等值设计,有助于进一步深化国内测验等值技术的实践研究。当然,大规模教育认知诊断与测评是一项系统工程,其中任何一个环节出现纰漏都有可能影响研究的整体质量,甚至完全背离教育认知诊断的本质追求。虽然本研究的初衷是试图通过一次完整的儿童数学学力认知诊断与测评研究,从认知模型构建、诊断测验开发、实测数据分析、诊断结果报告等各个环节进行大胆尝试,以尽可能深刻地窥探新一代心理测量理论与教育认知诊断技术的核心奥秘。但由于研究者的时间、精力和学识所限,致使研究中仍不可避免地存在一些遗憾与不足:第一,虽然分别从质性分析和实证量化两个方面,对儿童数学学力认知模型的合理性与完备性进行了综合验证,但由于尚缺乏更多外部效度证据的支持,可能会使该模型在今后测量实践中的大范围推广应用受到一定程度的影响。第二,虽然各年级诊断测验编制的全过程都尽量严格按照认知诊断理论的技术规范进行操作,且在试测中也获得了较为理想的测量学指标。但在大规模正式测验数据的分析过程中,仍然发现各年级测验中均有个别项目的部分指标不够理想。第三,虽然本研究采用整群随机抽样的方法选取了六所不同类型学校近万名儿童及其家长参与了测评调查。但是在数据分析时仍然发现测试学校的代表性有限,所选优势学校儿童的数学学力水平及属性掌握概率并没有如所预期的那样大幅度高于薄弱学校儿童。第四,虽然已经对儿童数学学力认知诊断的前五个环节(认知模型构建、诊断测验编制、诊断模型开发/选择、诊断结果报告、提出补救教学建议)做了较为广泛而深入的探讨,但囿于研究者的时间与精力,并没有实施认知诊断评估的最后一个环节,即补救教学干预。以上不足将成为本研究后续努力的拓展方向。
[Abstract]:"Knowledge" is dynamic, development, evaluation and interpretation of the structure of its connotation, will always be influenced by the ideal and the educational needs of the specific times. Today, with the essence of educational view transcendence and extension, math in the study status of evaluation of comprehensive quality of basic education in continuous improvement. Caused by the countries all over the world pay more attention to monitoring and evaluation of large scale dynamic process on the development of children's mathematics ability. Under this background, a new generation of psychological and educational measurement theory is representative of cognitive diagnosis theory, gradually revealing its unique advantages, internal psychological processes that can be hidden on the test the total score behind the more detailed, more accurate detection, so as to realize the diagnosis and the ability of the group development characteristics of children's cognitive level. In this study, it is the foundation of The key period of math development in 6? 12 years old children as the research object, with the level of development of math and structural characteristics of diagnosis for research purposes, according to the research paradigm of modern cognitive diagnosis theory, firstly build the children key math cognitive model; then the preparation of the "children's math cognitive diagnostic test (1? Grade 6) >; finally 6 schools were randomly selected for a total of 8289 1? On a large scale measured math development status of grade 6 children. In addition, in order to more in-depth analysis of children's mathematics ability gap formation, but also by < math children factors questionnaire > (including teachers and two families), and the questionnaire survey. The final conclusions are as follows: (1) children's math cognitive model constructed by the basic complete and reasonable. This research not only from the children's psychological process of mathematical problem solving Process of "full-time compulsory education mathematics curriculum standard (2011 Edition) > and < full-time primary school mathematics textbooks (PEP, 2013 Edition) > etc, demonstrates the theoretical basis of the model, but also based on the measured data of the completeness and rationality of the quantitative validation. The results showed: with the dual evidence theoretical analysis and experimental data of the cognitive model. (2) self < math children cognitive diagnosis tests (1? 6 grade) > meets the measurement requirements. In this study the integrated use of CTT, IRT and CDT of different methods on the quality of the six grade test are analyzed. The results show that the self diagnostic test can provide a more stable and reliable source of information for further research on the development of children's mathematics ability. (3) the selected specific cognitive diagnosis model of measured data analysis results are effective and credible. This research respectively from the macro The children's math level and micro key math model and attribute mastery probability dimension, the development features of the school children's mathematical ability were analyzed. The results show that the selected generalized diagnosis model (General Diagnostic Model, GDM) and the massive measured data has good fitting, and the classification accuracy of each key math attributes and consistency were higher. (4) the results of the analysis factors of children's mathematics ability influence and theoretical assumptions are basically consistent. This study from children's intelligence, non intelligence, academic achievement, and the family, many factors such as teachers, the factors of influence on children's mathematics ability structure and path relationship the analysis. The results show that all children in grade math and math achievement showed extremely significant positive correlation, and other factors the impact of the mathematics study of the relationship between It shows different characteristics in different grades, but with the theoretical assumptions consistent. (5) teaching remedial suggestions based on the results of cognitive diagnosis has strong operability. This study focused on the main problems found in the evaluation of cognitive diagnosis, separately from the school education and family influence on two aspects. Children of different grades and puts forward some theoretical and practical teaching remedial suggestions. To sum up, this research is based on a new generation of test theory, cognitive diagnostic theory, the evaluation method of children's mathematical ability and the path of practice is explored extensively, and eventually realize the cross grade diagnosis and analysis of equivalent parameters primary school children from grade one to grade six in math. The research of innovation is mainly reflected in the following four aspects: first, to children's math learning psychology structure structure based on concept The in-depth analysis as the basis, with the construction of key children math cognitive model operation, and has certain theoretical and practical significance to deepen children's mathematical cognitive diagnosis research. Second, < math by children's cognitive diagnosis tests (1? 6 grades) >, to achieve the overall diagnosis and evaluation of children's Mathematical Cognition the whole academic grades of primary school, can not only provide the necessary method for the follow-up study of reference, can also promote the practice of cognitive diagnosis technology accumulated the necessary experience. Third, compiled by R language program, the study of cognitive diagnosis scale based on measured data, can provide the domestic education cognitive diagnosis evaluation practice new technology path. Fourth, children's math equivalent based on the IRT design, the practice research is helpful to further deepen the domestic test equating technology. Of course, the large scale cognitive education The diagnosis and evaluation is a systematic project, in which a part of any flaws are likely to affect the overall quality of the essence of education, or even a complete departure from the pursuit of cognitive diagnosis. Although the purpose of this study is trying to complete a children's math cognitive research on diagnosis and evaluation, constructed from a cognitive model, diagnostic test development and the data analysis, the diagnosis results were reported in all aspects of a bold attempt, as far as possible to the core mystery deeply into a new generation of psychological measurement theory and education cognitive diagnosis technology. But because of the time, energy and knowledge is limited, the research is still inevitably has some insufficiencies: first, though from two aspects of qualitative and quantitative, rationality and completeness of children's math cognitive model to conduct a comprehensive verification, but due to the lack of more Many external validity evidence, may make the model widely applied in surveying practice in a certain extent affected. Second, although each grade diagnosis test for the whole process to strictly in accordance with the technical specifications of cognitive diagnosis theory for operation, and the test also obtained the ideal measurement index. But in the process of analyzing large-scale formal test data, and still found that some indexes were all grade tests of individual projects is not ideal. Third, although this study using cluster random sampling method to select six schools of different types of nearly 10000 children and their parents participated in the survey. But still found the school test the representative in the data analysis, math level and the properties of the selected advantage of school children to master the probability is not as expected greatly higher than Weak school children. Fourth, although have on children's math cognitive diagnosis before five aspects (cognitive model, diagnostic test, diagnosis model development / selection, diagnosis report, proposed remedial teaching suggestions discussed) more and more widespread, but limited by the time and energy, and not the implementation of the last part of Cognitive Diagnostic Assessment, namely remedial teaching intervention. The above issues will be the development direction of the following research efforts.

【学位授予单位】：内蒙古师范大学
【学位级别】：博士
【学位授予年份】：2017
【分类号】：B842.1

【参考文献】