摘要
本研究基于项目反应理论,提出了一种检验力高且犯Ⅰ类错误率小的检测DIF的新方法:LP法(likelihood procedure),且以2PLM下对题目进行DIF检验为例介绍此法。本文通过与MH方法、Lord卡方检验法和Raju面积测量法三种常用的检验DIF的方法比较研究LP法的有效性,同时探讨样本容量、测验长度、目标组和参照组能力分布的差异、DIF值大小等相关因素对LP法有效性可能产生的影响。通过模拟研究,得到以下结论:(1)LP法比MH法及Lord卡方法更灵敏且更稳健;(2)LP法比Raju面积测量法更合理;(3)LP法的检验力随着被试样本容量或DIF值的增大而增大;(4)当参照组与目标组的能力无差异时,LP法在各种条件下的检验力比参照组与目标组的能力有差异时的检验力高;(5)LP法对一致性DIF和非一致性DIF都有良好的检验力,且LP法对一致性DIF的检验力比对非一致性DIF的检验力高。LP法可以简便的扩展并运用到多维度、多级评分项目上。
With the development of psychological metrology and the wide application of psychological and educational tests, the fairness of tests has been a matter of great concern to educators and psychologists, and more in-depth studies on the differential item functioning have taken place. Detection of differential item functioning（DIF） has been widely employed in the analysis of routine items, and a number of methods have been developed to detect DIF, such as Mantel-Hansel（MH） Procedure, Standardization（STND）, Simultaneous Item Bias Procedure（SIBTEST）, Likelihood Ration（LR） Test, Lord’s Chi-Square, Raju’s Area Measures, MIMIC Method, etc. In most of these methods, there exists either a low power of test or a high type I error rate. Therefore it is necessary to find out one more effective method to detect DIF. This paper proposed the LP（Likelihood Procedure） method for detecting DIF, which is an IRT-based method with item detection under the condition of two parameter logistic model（2PLM） as a representative.The performance of LP was compared with that of the MH method, Lord chi-squared and Raju Area Measurement. DIF size, test length, sample size, the difference distribution of abilities between the focal group and reference group were also considered. Three levels of DIF size were.3,.5 and.8. Two levels of test length were 40 and 100. Three levels of sample size were 500, 1000 and 2000 examinees. There were two distributions of abilities between the focal group and the reference group： one fitted in with standard normal distribution individually, the other with the distribution of abilities in the reference group fitting in with standard normal distribution, while the distribution in the focal group fitting in with normal distribution in which the mean was-1 and the standard deviation was 1. In this simulation study, data was generated using two parameter logistic models. The difficulty value of the DIF item in the study corresponding to that in the focal group, or the discrimination value, was greater than the one in the reference group. There were totally six DIF items in each group under the conditions of uniform DIF and non-uniform DIF, including the corresponding ones of the three true value DIF items. The simulation research indicates the following results：（1） LP has a high power of test and low and stable type I error rate.（2） As a whole, the power of LP is higher than that of the Lord chi-squared method and much higher than that of the Mantel-Hansel（MH） method. The type I error rate of LP is lower than that of the Lord chi-squared method when the test length is 100. The type I error rate of the MH method is far beyond the stability scope.（3） LP is no better than Raju Area Measurement method in test power, but the type I error rate of the latter is so high that it is above.1 and far beyond the stability scope under a variety of conditions.Generally speaking, LP has the following advantages：（1） LP is more sensitive and stable compared with MH.（2） LP is more reasonable to be used for checking DIF, compared with Raju Area Measurement.（3） The test power of LP increases with the sample size of participants or true DIF value. 4） Compared with the condition of the same ability, the test power of LP is lower when the focal group and the reference group behave different abilities. 5） The test power of LP is high for both uniform DIF and non-uniform DIF, with a higher power for the former. To conclude, LP is not only applicable to the two-parameter logistic model, but also to the single-parameter and three-parameter logistic models. In addition, LP is easy to be applied extensively to multi-dimensional and multi-category scoring items
出处
《心理科学》
CSSCI
CSCD
北大核心
2016年第3期720-726,共7页
Journal of Psychological Science
关键词
项目功能差异
项目反应理论
LP法
MH法
Lord卡方检验法
Raju面积测量法
differential item functioning（DIF）
item response theory（IRT）
LP（Likelihood Procedure）
the Mantel-Hansel（MH） method
Lord’s ChiSquare
Raju’s area measures
