基于等级反应模型的广义距离判别法

The Generalized Distance Discrimination Based on Graded Response Model

受0-1评分广义距离判别法的启发,给出一种新的多级评分认知诊断方法.蒙特卡洛模拟实验结果显示:基于等级反应模型的广义距离判别法(GRM-GDD)比相应的属性层级模型(GRM-AHM)有更高的模式判准率,尽管随着被试作答失误率的提高,模式归准率均有所下降.

It is motivated by generalized distance discrimination（GDD） for 0-1 scoring proposed by Sun and her colleagues,the GDD for polytomous scoring based on the graded response model（GRM-GDD）is proposed.The results of Monte Carlo simulation study show that the behavior of GRM-GDD is better than that of polytomous attribute hierarchy method based on GRM（GRM-AHM） although the pattern classification ratio decreasing with the slipping of the responses increasing.