中国基础教育博硕士论文库

节点文献

初中生数学能力计算机自适应测验的编制研究

【作者】 张妮婕

【导师】 刘晓陵;

【作者基本信息】 华东师范大学, 发展与教育心理学, 2018, 硕士

【摘要】 多维项目反应理论、认知诊断、计算机自适应测验被视为现代测量理论发展的三大方向。计算机自适应测验以项目反应理论为基础,以计算机为媒介,为每个被试"量身定制"测验,提高了测量的效率和精度。但至今国内自主编制的计算机自适应测验还如凤毛麟角,因而笔者将目光投向基础教育中的数学能力测评,致力于形成可以应用于实践、对初中数学的学习与教学具有参考和指导作用的计算机自适应测验。本研究分为如下五部分内容:第一部分、研究背景概述。首先对本研究所涉及的计算机自适应测验和数学能力两个重要概念在总结前人研究的基础上作出明确的界定;其后简要阐述本次研究的研究目的和研究意义,并对研究思路进行了梳理。第二部分、理论基础研究。对测验编制所基于的两大理论一一经典测量理论和项目反应理论——的发展脉络与特点进行了简要的梳理。本研究在项目编制阶段主要利用经典测量理论的优势和经验;在题目入库阶段使用基于项目反应理论的参数估计与项目分析方法。第三部分、建立计算机自适应数学测验题库。建立题库包括如下几个步骤:首先,以初中数学大纲为基础,编制由中低难度的基础题和较高难度的提高题按一定比例组成的数学测验,共计500题;然后,采用锚题设计将总测验拆分为20个分测验在初中生中施测来完成预测,被试共计4443人;最后,将收集到的数据按照项目反应理论的假设进行模型拟合检验,依据检验结果选择恰当的项目反应模型进行参数估计、项目分析与测验等值,从中挑选拟合度优良、编制质量较高的448道题目组成题库。第四部分、计算机自适应测验系统的设计与实现。计算机测验系统采用B/S(浏览器/服务器)体系,在设计上针对管理员和被试分别设置管理和测试两个模块。管理模块用于对管理员权限、试题信息与被试信息进行维护;测试模块用于被试从登录到完成测试获得反馈的完整过程,其中包括在测试的各重要环节比较并选择所需的策略与方法:1)能力估计,本次研究选择极大似然估计法;2)选题策略:采用a分层b分块法与内容平衡法相结合的方法,以提高项目使用效率,提高选题的均衡性;3)终止规则:选用两条标准——一是学生作答题目的信息总量达16,二是总时长不超过40分钟;4)结果报告:将取值在[-3,3]之间被试能力θ值转换为取值范围为[0,1]、更加易于理解的ππ值作为向被试报告的结果。第五部分、计算机自适应测验系统的试用分析。将调试好的测试系统用于正式施测,对系统功能、测验信效度及题库安全性进行检验。结果表明,本研究所开发的计算机自适应测验系统可以实现对测验的管理并完成测试过程,且对数学能力水平有差异的个体具有较好的区分度,同时题目的曝光率相对较为均衡,能够在一定程度上保障题库的安全性。

【Abstract】 Multidimensional Item Response Theory,Cognitive Diagnosis and Computerized Adaptive Test are three important research directions in the future.Based on Item Response Theory and taking computer as medium,the computerized adaptive test chooses the item for students according to the updated ability estimation to customize the test for every student,thus simultaneously improves the test efficiency and accuracy.However,there are seldom self-development computerized adaptive test,so the present study focuses on the mathematical ability evaluate in compulsory education,devoting to developing a computerized adaptive test which can provide instructions for studying and teaching in practice.The study was conducted as the following five parts:First of all,the present study defines the concept of computerized adaptive test and mathematical ability based on the former study,especially the structure of the later,which is used to establish the item pool.Then the author clarifies the purpose and the significance of the study,listing the whole procedure of the research.Secondly,introducing the developmental history and characteristics of Classical Test and Item Response Theory which the study are based on and emphasizing the advantages of item response theory.We make up the items according to the guidance of CTT while use the method provided by IRT to estimate parameters and analyze items.Thirdly,establishing the item pool should follow several steps:first,organize items with different difficulty levels to develop a traditional paper and pencil test,in total 500 items;Then,adopt anchor item design to reconstruct the test to be 20 subsidiary test and conduct the tests among 4443 junior students to collect information for estimation of parameters of the items.Next,use statistical methods to test the model and item fitness,then select the appropriate model to calculate three parameters for item:difficulty,discrimination and guessing parameters.Finally,448 well-developed items are chosen to establish the item pool.In the fourth part,introducing how to design and realize the computerized test system.The system was divided to two modules:management module and test module.In the former,the admin can adjust the system function,browse student information and alter item details.In the test module,student will complete the entire test process from log-in to grades feedback.Also,the author have selected the appropriate techniques for the next four procedures:1)ability estimation:maximum likelihood method;2)item selection strategies:STR_b combining with content balancing to strengthen the security of the system;3)ending rule:the information amount arrive above 16 or the test time longer than 40minites;4)result transformation:transfer the theta which located between-3 to 3 to pai which located between 0 and 1 to be more meaningful for examinees.The fifth part is about the analysis of the test administration results.The author test and adjust the system until there is no visible questions and administer the test among junior students varying form Grade 7 to Grade 9,to examine the security of the item pool and the reliability and validity of the test.Finally,the results shows that the system works well to manage and conduct the test.Also,there proves to be a relatively high discrimination for students of different ability levels and a balance item exposure rate which promising for higher security.

移动知网

全球学术快报

  • 【网络出版投稿人】 华东师范大学
  • 【分类号】B842.3
  • 【下载频次】103
节点文献中: