艾森克人格问卷项目质量的项目反应理论分析

使用MULULOG7．03软件的边际极大似然估计法，选取双参数kgistic模型对某大学2011级新生的EPQ测试数据进行分析，针对项目区分度、项目难度、信息量等指标对各项目及各分量表进行深入探讨．结果显示：数据符合项目反应理论的基本假设；多数项目的区分度、难度达到理论要求．但E、P、N三个分测验在划界分数点上得到的信息量有限，难以对被试做出良好的区分；三个分测验各自的总信息量未达到理论要求．

AN ITEM ANALYSIS OF EPQ ON THE ITEM RESPONSE THEORY

It is 30 years since Eysenck＇ s personality （EPQ） theory was first introduced to China, and the latest norm was published in 2000. Exposed too much in the past 10 years, its applicability needs to be tested empirically again with present samples. The aim of this paper is to analyze EPQ＇ s items＇ properties under IRT, focusing on the measurement accuracy of the items. The program MULTILOG 7.03 was employed as the tool for parameter estimation. In the estimating procedure, maximum likelihood estimation and two parameter logistic model were utilized. The parameters of difficulty, discrimination and information curve were analyzed in detail. The results indicated that the data accorded with the basic assumptions in IRT： unidimensionality, monotone increasing and invariance of parameter estimation etc. The difficulty and discrimination of most of the EPQ＇ s items met the theoretical requirements, which demonstrated the revision of the questionnaire was quite successful ; many erroneous judgments, nevertheless, should be aroused if it is applied to make assessments or interventions to the subject due to its limited amount of information obtained from the subject. As an essential complement to CTT to analyze items, IRT will be widely used in psychological and education test studies in the future.