基于曼哈顿距离构建非参数Q矩阵修正方法

将被试得分、理想反应距离和被试得分异常原理相结合,并加入属性计分下的被试得分特性,开发了用于多级评分情境下属性计分曼哈顿距离法(SA-MD),在不同条件下验证了SA-MD的稳定性和适宜性.通过模拟研究和实证研究表明:(i)从逻辑推导出SA-MD用于多级评分情境下Q矩阵修正更合理;(ii)在多种条件中,SA-MD 均有较优的修正效果,适用范围更广,稳定性更高;(iii)当小样本测验进行Q矩阵修正时,使用SA-MD方法可获得更优的效果.

Constructing a Non-Parametric Q-Matrix Correction Method Based on Manhattan Distance

Combining the subjects' score characteristics with the ideal response distance and the abnormal principle of subjects' scoring,and adding the subject score characteristics under attribute scoring, the Manhattan distance method for attribute scoring(SA-MD)is developed in multistage scoring situations.The stability and suitability of SA-MD are verified under different conditions.Through simulation and empirical research,the following conclusions are drawn.(i)The rationality of SA-MD for Q-matrix correction in multistage scoring situation is obtained by logical derivation.(ii)In a variety of conditions,SA-MD has a better correction effect,a wide range of application,high stability.(iii)Sa-MD method can obtain better effect when Q matrix is modified for small sample test.