[1]李 佳,丁树良.计算机化自适应测验中能力估计新方法[J].江西师范大学学报(自然科学版),2019,(02):142-146.[doi:10.16357/j.cnki.issn1000-5862.2019.02.05]
　LI Jia,DING Shuliang.The New Method of Ability Estimation in CAT[J].Journal of Jiangxi Normal University:Natural Science Edition,2019,(02):142-146.[doi:10.16357/j.cnki.issn1000-5862.2019.02.05]
点击复制

计算机化自适应测验中能力估计新方法()
分享到：

《江西师范大学学报》（自然科学版）[ISSN:1006-6977/CN:61-1281/TN]

卷:
期数:
2019年02期
页码:
142-146
栏目:
心理与教育测量
出版日期:
2019-04-10

文章信息/Info

Title:
The New Method of Ability Estimation in CAT
文章编号:
1000-5862(2019)02-0142-05
作者:
李 佳; 丁树良
江西师范大学计算机信息工程学院,江西 南昌 330022
Author(s):
LI Jia; DING Shuliang
College of Computer Information Engineering Jiangxi Normal University,Nanchang Jiangxi 330022,China
关键词:
贝叶斯众数估计方法;  期望后验估计方法;  改进的极大似然估计方法;  能力估计效率
Keywords:
MAP;  EAP;  NMLE;  ability estimation efficiency
分类号:
B 841
DOI:
10.16357/j.cnki.issn1000-5862.2019.02.05
文献标志码:
A
摘要:
能力估计的极大似然估计方法(MLE)不能处理全0或全1的被试反应模式,若事先设置好能力估计的上下界,则会导致能力估计的有效范围缩小的后果; 而贝叶斯估计方法需要选择先验分布,先验分布的选择必须很慎重.在原有似然函数的基础上,构建2个新的项目,提出了改进的MLE方法(NMLE).NMLE既不需要能力先验分布,也不会缩小能力估计范围,而且可以处理各种反应模式.蒙特卡洛实验结果表明新方法表现良好.
Abstract:
The maximum likelihood estimation method(MLE)of the ability estimation does not work with special response patterns,such as all elements of the response patter are 0s or all 1s.If setting lower and upper bounds of ability estimation,the ability estimation scale will shorten.Bayesian-based estimation methods need a prior distribution,the choice of prior distribution must be careful.A new ability estimated method(NMLE)is introduced,adding two new items to establish a new likelihood function based on the existing item bank.New method not only need not ability prior distribution,but also does not shorten the ability estimation scale,and can deal with all kinds of response patterns.New method has better performance through the Monte Carlo simulation method on 3PLM.

参考文献/References:

[1] 漆书青,戴海崎,丁树良.现代教育与心理测量学原理[M].北京:高等教育出版社,2002.
[2] 张心,涂冬波.计算机化自适应测验中几种常用能力估计方法的特性与评价[J].中国考试,2014(5):18-25.
[3] Lord F M,Novick M R.Statistical theories of mental test scores[M].New Jersey:Addison-Wesley,1968:392-449.
[4] Samejima F.Estimation of latent ability using a response pattern of graded scores[J].Psychometrika,1969,34(1):1-97.
[5] Bock R,Mislevy R.Adaptive EAP estimation of ability in a microcomputer environment[J].Applied Psychological Measurement,1982,6(4):431-444.
[6] Hambleton R K,Swaminathan H.Item response theory:Principles and application[M].Boston:Kluwe-Nijhoff,1985.
[7] Wang Tianyou,Walter P Vispoel.Properties of ability estimation methods in computerized adaptive testing[J].Journal of Educational Measurement,1998,35(3):109-135.
[8] Warm T A.Weighted likelihood estimation of ability in term response theory[J].Psychmetrika,1989,54(3):427-450.
[9] 李佳,丁树良.多种分层方法在CAT校准误差中的应用研究[J]. 江西师范大学学报:自然科学版,2016,39(1):69-72.
[10] 李佳,丁树良,方剑英.基于平均数形式的选题策略比较[J].江西师范大学学报:自然科学版,2015,39(1):69-72.
[11] 孟祥斌,陶剑,陈莎莉.四参数Logistic模型潜在特质参数的Warm加权极大似然估计[J].心理学报,2016,48(8):1047-1056.
[12] Baker F B,Kim S H.Item response theory:parameter estimation techniques[M].New York:Marcel Dekker,2004.
[13] Magis D A.Accuracy of asymptotic standard errors of the maximum and weighted likelihood estimators of proficiency levels with short tests[J].Applied Psychology Measurement,2014,38(2):105-121.
[14] 毛秀珍,辛涛.多维计算机化自适应测验:模型、技术和方法[J].心理科学进展,2015,23(8):907-918.
[15] 韩雨婷,涂冬波,王潇濛,等.多维计算机化自适应测验选题策略的开发及比较[J].心理学报,2017,40(4):997-1004.

备注/Memo

备注/Memo:
收稿日期:2018-07-19 基金项目:国家自然科学基金(31500909,31360237,31160203,30860084,11401271)和江西省教育厅科学技术(GJJ170212)资助项目. 作者简介:李 佳(1979-),女,江西南昌人,讲师,主要从事计算机辅助教学和心理测量方面的研究.E-mail:1276676143@qq.com
更新日期/Last Update: 2019-04-10