Based On Anchor Attribute Non-equivalent Group Design's Cognitive Diagnosis Equivalent Research:The Method Of Attribute Characteristic Curve Equivalent

Test equating refers to measuring the same psychological traits of multiple test formal of test score(or latent trait level) or item parameters unit system conversion, to correspond with each other indicators comparable process. Test equating has become an important issue in educational measurement research and application, but also as an education management attention. However, the current equivalent study focused on the framework of the classical measurement theory(CTT) and the item response theory(IRT), and the equivalent of a small number of studies in cognitive diagnosis(CD) did not break through the traditional method of equivalence, achieve equivalent of attribute and ability in cognitive diagnosis. Therefore, this research developed a new equivalent method in the cognitive diagnostic framework to achieve the equivalent of attribute and ability in the cognitive diagnostic theory. In HO-DINA model, this study use anchor attribute non-equivalent group design, and attribute characteristic curve equating method and using Monte Carlo simulations verify the equivalent method and equivalent scientific and rationality.Research indicates:1. Based on the HO-DINA model, the new developed equivalent method: attribute characteristic curve equating under the anchor attribute non-equivalent group design, it has ideal equivalent accuracy. The bias of equivalent coefficients A and B is 0.03 and 0.05, respectively, and the effect is ideal.2. When a value of the equivalent coefficient A in [0.9,1.6] range, the equivalent coefficient B in the [-0.4,0.4] range, equivalent coefficients A of bias' s index is small and the equivalent error is also small, so the equivalent accuracy is high. When a value of the equivalent coefficient A in [0.9,1.6] range, the equivalent coefficient B in the [-0.4,0.4] range, equivalent coefficient B of bias' s index is small and the equivalent error is also small, so the equivalent accuracy is high. When a value of the equivalent coefficient A in [0.9,1.6] range, the equivalent coefficient B in the [-0.4,0.4] range,equivalent coefficient A of RMSD's index is small and the equivalent error is also small, so the equivalent accuracy is high. When a value of the equivalent coefficient A in [0.9,1.5] range, the equivalent coefficient B in the [-0.4,0.4] range, equivalent coefficient B of RMSD's index is little difference, but they are relatively small, so the equivalent accuracy is high.3. Increasing the number of anchor attributes can effectively improve the accuracy of the new method of equivalent. Increasing the number of subjects also can effectively improve the accuracy of the new method of equivalent.4. Using MCMC algorithm under the HO-DINA model can achieve parameter estimation and equivalent, parameter estimation precision and equivalent's accuracy is ideal. Attribute discrimination parameters, attribute difficulty parameters, the bias of equivalent coefficient A and coefficient B are also ideal.