项目反应理论在题库建设中的应用研究  

ON APPLICATION OF PROJECT RESPONSE THEORY IN ITEM BANK CONSTRUCTION

作　　者：罗永莲[1] 贾玉芳[1] 

机构地区：[1]晋中学院计算机科学与技术分院,山西晋中030600

出　　处：《计算机应用与软件》2015年第1期86-88,152,共4页Computer Applications and Software

基　　金：山东省教育科学"十一五"规划课题(GH-08072)

摘　　要：通常的题库管理系统都要根据经典测量理论,选择样本组进行测量,然后根据测量统计分析的结果,修改或选择合适的试题以完成题库建设。这种方法对测试组的能力有一定的依赖性,所得到的修正参数也存在偏差。针对经典测量理论存在的问题,提出以项目反应理论为基础,通过多次迭代,利用最大似然估计法估计被测试组的能力参数,在此基础上产生题目参数。用此题目参数对试题相应初始值进行修正,并根据这些参数指标组合成卷。实验结果表明,此方法估计出的题目参数相对稳定,不因测试组的不同而有明显变化,实践证明能显著提高试题的有效性。Usually the management systems of examination item bank all select the sample group according to classical test theory for measurement ; then according to statistical analysis results of the measurement, the system modifies or selects proper test items to complete the con- struction of test item bank. Such method partly depends on the ability of the test group, and the correction parameters derived will deviate as well. Aiming at the problem of classical test theory, in this paper we propose that taking item response theory as the basis, through multiple iterations, and using maximum likelihood estimation method to estimate the ability parameters of the test group, the item parameters are generated based on this. These item parameters will be employed to correct the corresponding initial value of the test items, and the test paper will be composed according to these parameter indexes. Experimental results show that the item parameters estimated by this method are relatively stable, and do not significantly alter with different test groups, the practice proves that the method can remarkably improves the effectiveness of the test items.

关 键 词：项目反应理论 最大似然估计 难度分析 区分度分析 

分 类 号：TP391.1[自动化与计算机技术—计算机应用技术]