一种小样本情况下的双参数IRT估计方法  

A Two-parameter IRT Estimation Method for Small Samples

作　　者：冯彦彰 李志军[1] 

机构地区：[1]北方工业大学电气与控制工程学院,北京100144

出　　处：《工业控制计算机》2021年第10期98-100,共3页Industrial Control Computer

摘　　要：研究了针对小样本的项目反应理论估计方法。首先,探究了CTT和IRT两种方法对题目难度参数估计结果的相关性及其形成原因,提出了在被试能力为非标准正态分布时CTT的难度参数估计方法,并将其作为IRT算法中参数b的初始值;然后,通过研究IRT算法中参数a对估计效果的影响,表明了参数a取固定值时,能够有效降低了算法的计算量,缩短计算时间,同时具有很好的估计效果;最后,结合上面对两个参数分析的结论,提出了一种适用于小样本的双参数IRT估计方法,仿真实验验证了其有效性。This paper researchs the estimation method of item response theory for small sample.First of all,this paper explores the correlation between CTT and IRT methods on the estimation results of difficulty parameters and the reasons for their formation,proposes the CTT difficulty parameter estimation method when the ability of the subjects is non-standard normal distribution,and takes it as the initial value of parameter B in the IRT algorithm.Then,by studying the influence of parameter A in IRT algorithm on the estimation effect,it is shown that when parameter A is fixed,the calculation amount of the algorithm can be effectively reduced,the calculation time can be shortened,and the estimation effect is good at the same time.Finally,a two-parameter IRT estimation method suitable for small samples is proposed based on the above two parameter analysis conclusions,and its effectiveness is verified by simulation experiments.

关 键 词：项目反应理论 参数估计 小样本 

分 类 号：TP301.6[自动化与计算机技术—计算机系统结构]