Özet:

This study has studied the re-accessibility of individual parameters in the case of multi-category cognitive diagnosis and multi-dimensional substance reaction theory (CMT) models being used. For this purpose, data were produced for the multi-category cognitive diagnosis models for the polytomous generalized deterministic input noisy and gate (pG-DINA) and the fully-additive model (fA-M) and the CBMTK models for the compensatory 2PL model. The data, the substance differentiability index, the test ratio according to the structures of the substances, the test length and the inter-capacity correlation values are differentiated using a total of 54 conditions, produced with 25 repetitions. The results were obtained by the comparison of the individual parameters and the individual parameters used in data production. According to the findings, all types of data are made by the models in which they belong to the maximum accuracy of cutting. In the adjustment analysis, the fA-M cut the other two model data, while the CBMTK cut the fA-M data in a rate of accuracy close to the model cutting that the data belongs to. The other two model data of PG-DINA have been observed, while the CBMTK has low performance in cutting pG-DINA data. According to the conditions used in the study, the test length and substance differentiability were the strongest factors affecting the accuracy of the individual parameters. The substance structure ratio condition has been much more effective in CBMTK data analysis and adaptation. The influence of the existence of inter-capacity correlation on the accuracy of individual parameters has been clearer but still limited in the analysis of the CBMTK data.