Xin Tao, Zhang Jiahui
Beijing Normal University, China.
Michigan State University, East Lansing, USA.
Appl Psychol Meas. 2015 Jan;39(1):44-61. doi: 10.1177/0146621614542427. Epub 2014 Jul 23.
A local equating method of cognitively diagnostic modeled observed score was introduced to address the problem of equating without anchor test as well as the satisfaction of equating criteria. The criterion of test construct was redefined on attribute level. Then, a true family of equating transformations given attribute mastery pattern was derived from the criterion of equity, which is the main concern of local equating. The results of simulation studies showed a substantial bias for the traditional equipercentile equating under the equivalent groups design but much smaller bias for the local method. The proposed method also showed satisfactory performances under the nonequivalent groups design even without anchor test. The local method appeared to be quite robust with respect to changes in the sample size, but its biases became larger with increasing model misfit.
引入了一种认知诊断模型观测分数的局部等值方法，以解决无锚定测验等值问题以及满足等值标准的问题。在属性层面重新定义了测验结构标准。然后，从公平标准（这是局部等值的主要关注点）推导出给定属性掌握模式的真正等值变换族。模拟研究结果表明，在等效组设计下，传统的等百分位等值存在较大偏差，而局部方法的偏差要小得多。即使在无锚定测验的非等效组设计下，所提出的方法也表现出令人满意的性能。局部方法在样本量变化方面似乎相当稳健，但其偏差会随着模型拟合度的增加而变大。
