四参数Logistic模型研究进展及其评析  

The Advances of Four - Parameter Logistic Model and Its Comments

作　　者：简小珠[1,2] 张敏强[1] 彭春妹[2] 

机构地区：[1]华南师范大学心理应用研究中心,广州510631 [2]井冈山大学教育学院,吉安343009

出　　处：《心理学探新》2010年第3期69-73,共5页Psychological Exploration

基　　金：教育部省部共建人文社会科学重点研究基地项目(2009JJDXLX006);广东省自然科学基金项目(9151063101000002);江西省社会科学规划"十一五"学科共建项目(09JY226)

摘　　要：在测验中存在着低能力被试答对高难度试题的猜测现象,和高能力被试答错容易试题的睡眠现象,此时可以使用四参数模型来分析测验数据。Barton和Lord认为应用四参数模型的实践意义不大,但结论的依据不充分。近年来研究者从测验项目拟合,改善被试能力估计等方面进行了分析,认为在四参数模型下可以有效纠正被试能力高估或低估现象,认为单、两、三参数模型是四参数模型的特例,建议使用四参数模型。In the paper - pencil test and CAT,there exist the phenomena that the high - ability examinee makes wrong response on the easy item（ guessing） ,and that the low - ability examinee makes correct response on the difficult item（ sleeping） . Many researches have been made on the topic of the guessing or ceiling phenomenon under four - parameter Logistic model（ 4PM） . Barton Lord didn’t agree to urge the use of 4PM. In recent years,however,many researchers have made the researches and proposed the use of 4PM. It has been demonstrated that： （ 1） 4PM can improve the good - fitness of mode - data; （ 2） 4PM can rectify the underestimation phenomenon when there exist the sleeping phenomenon,and 4PM can rectify the overestimation phenomenon when there exist the guessing phenomenon; （ 3） one - parameter,two - parameter and three - parameter Logistic model are the special case of 4PM. The new program WINSTEPS has published. The user can estimate the item parameter under 4PM using the WINSTEPS. WINSTEPS will help to popularize the 4PM.

关 键 词：IRT 四参数模型 猜测现象 睡眠现象 

分 类 号：B841.2[哲学宗教—基础心理学]