Rasch模型在物理试题质量分析中的应用
发布时间：2018-09-15 19:23
【摘要】：本研究把国际先进的科学的测量理论Rasch模型引入到物理试题测评中。验证试题的质量如何,试题是否科学、客观、公平,试题的结构、试题难度、试题中知识点分配及覆盖面是否符合考试大纲的要求等。通过学习Rasch测量软件Conquest3.0,了解软件的命令编码,软件的实际操作流程。利用Conquest软件对我省某次大规模考试理科综合物理主观题做一次较全面和系统的分析。统计的样本数为32442个,该测验主观题总分为72,样本均值为26.48,标准差为15.22,方差为231.65,偏度为0.04,峰度为-0.76,均值标准误为0.09,信度为0.81,Outift MNSQ和Infit MNSQ值都在可接受的范围内,说明本次测验信度较好,可以对题目难度和学生能力做出较全面、客观的评价。Rasch模型对题目难度和被试者能力进行分析,结果表明题目和学生能力都近似于正态分布,题目的难度相对高于学生能力水平。题目难度分布均匀,既有难题,又有容易题,第20题(T20)最容易,第19题(T19)较容易,第21题(T21)较难,第22题(T22)最难。各分数段的考生分布非常合理。因此,这份试卷能很好地区分考生的能力水平,达到选拔人才的目的,其难度符合考试大纲要求。
[Abstract]:In this study, the international advanced scientific measurement theory Rasch model is introduced into the physics test. To verify the quality of the test questions, whether the questions are scientific, objective, fair, the structure of the test questions, the difficulty of the test questions, whether the distribution and coverage of knowledge points in the test questions meet the requirements of the examination outline, etc. Through the study of Rasch measurement software Conquest3.0, to understand the software command coding, the actual operation of the software flow. This paper makes a comprehensive and systematic analysis of the comprehensive physical subjective questions of a large scale science examination in our province by using Conquest software. The sample number was 32442, the total score of the subjective question was 72, the sample average was 26.48, the standard deviation was 15.22, the variance was 231.65, the deviation was 0.04, the kurtosis was -0.76, the mean standard error was 0.09, the reliability was 0.81outift MNSQ and Infit MNSQ were all within acceptable range. The results show that the reliability of this test is good and can make a more comprehensive assessment of the difficulty of the question and the ability of the students. The objective evaluation .Rasch model analyzes the difficulty of the question and the ability of the subjects. The results show that both the questions and the students' abilities are similar to the normal distribution. The difficulty of the problem is relatively higher than the ability level of the students. The difficulty distribution is even, both difficult and easy, 20 (T20) is the easiest, 19 (T19) is easier, 21 (T21) is more difficult, 22 (T22) is the most difficult. The distribution of candidates for each fraction is very reasonable. Therefore, this test paper can distinguish the ability level of examinee well, achieve the purpose of selecting talents, its difficulty accords with the examination outline request.
【学位授予单位】：福建师范大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：G633.7
本文编号：2244256
[Abstract]:In this study, the international advanced scientific measurement theory Rasch model is introduced into the physics test. To verify the quality of the test questions, whether the questions are scientific, objective, fair, the structure of the test questions, the difficulty of the test questions, whether the distribution and coverage of knowledge points in the test questions meet the requirements of the examination outline, etc. Through the study of Rasch measurement software Conquest3.0, to understand the software command coding, the actual operation of the software flow. This paper makes a comprehensive and systematic analysis of the comprehensive physical subjective questions of a large scale science examination in our province by using Conquest software. The sample number was 32442, the total score of the subjective question was 72, the sample average was 26.48, the standard deviation was 15.22, the variance was 231.65, the deviation was 0.04, the kurtosis was -0.76, the mean standard error was 0.09, the reliability was 0.81outift MNSQ and Infit MNSQ were all within acceptable range. The results show that the reliability of this test is good and can make a more comprehensive assessment of the difficulty of the question and the ability of the students. The objective evaluation .Rasch model analyzes the difficulty of the question and the ability of the subjects. The results show that both the questions and the students' abilities are similar to the normal distribution. The difficulty of the problem is relatively higher than the ability level of the students. The difficulty distribution is even, both difficult and easy, 20 (T20) is the easiest, 19 (T19) is easier, 21 (T21) is more difficult, 22 (T22) is the most difficult. The distribution of candidates for each fraction is very reasonable. Therefore, this test paper can distinguish the ability level of examinee well, achieve the purpose of selecting talents, its difficulty accords with the examination outline request.
【学位授予单位】：福建师范大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：G633.7
【参考文献】
相关期刊论文 前4条
1 李中权;孙晓敏;张厚粲;张立松;;多面Rasch模型在主观题评分培训中的应用[J];中国考试(研究版);2008年01期
2 王蕾;;Rasch测量原理及在高考命题评价中的实证研究[J];中国考试(研究版);2008年01期
3 杜文久;;数学测验中主观题的评分问题[J];数学教育学报;2006年03期
4 孙晓敏;薛刚;;多面Rasch模型在结构化面试中的应用[J];心理学报;2008年09期
，本文编号：2244256
本文链接：https://www.wllwen.com/jiaoyulunwen/chuzhongjiaoyu/2244256.html
