| The application of computer technology in the field of psychological andeducational measurement, contributes the birth and rapid development ofcomputerized adaptive testing (CAT).CAT has attracted wide respected for itsadvantage of building the "tailored" test for every examinees. Computerized adaptivetesting for cognitive diagnosis (CD-CAT), as the extension of CAT, not only has acognitive diagnostic function to provide detailed measurement of psychologicalcharacteristics for test-takers, but also employ the CAT as the test form to improve thediagnostic accuracy of measurement and efficiency. Therefore, in recent years,CD-CAT has received much attention in the field of educational measurement.Generally cognitive diagnosis test is considered to be low-stake. Thus, intraditional CD-CAT, high accuracy is the goal of studying item selection methods.However, the efficiency of item selection and the security of item pool are more orless neglected. So far, discussions of item exposure control methods in CD-CAT havebeen nearly absent from the current literature. In order to address this problem, thispaper proposes a new item selection strategy called p-STR, meanwhile briefly studiesunfixed-length CD-CAT. The first study compared six CD-CAT item selectionstrategies (random method (RD), KL information method (KL), Shannon Entropymethod (SHE), posterior weighted KL method (PWKL) hybrid KL method (HKL)and p-STR method) under the new item bank design, and evaluated theirperformances primarily in terms of the uniformity of item pool usage and the testoverlap rate and other criteria. The four types of the attribute hierarchies of linear,convergent, divergent and unstructured structure are considered and each typecorresponds to an item pool and tests are fixed-length. The results of Monte Carlosimulations indicate that compared to other methods, the new strategy is effective inimprovement of the security of item pool, meanwhile with high measurementaccuracy. |
PDF Full Text Request