Parameter Estimation And Application Of 2PL-Constrained Model In Raven's Advanced Progressive Matrix Test

Within the framework of psychology, this paper begins with a brief review of the ongoing study and the developing tendencies concerning the IRT at home and abroad. On the description of the basic models in IRT and on the basis of comparing the advantages and disadvantages of its new models, prospects are presented for research and development over the 2PL-Constrained Model.Proposed by Embretson in 1999 and based on the new cognitive theory, the 2PL-Constrained Model reads as a new IRT model that includes the impact on item discrimination as well as difficulty. By using the joint maximum likelihood estimation and the presently popular MATLAB language, the program of parameter estimation is composed in this thesis. Furthermore, by Mont Color simulation, a number of simulation tests are designed to analyze and discuss the accuracies of parameter estimation. Consequently it is revealed that when ABS and RMSD are used as indexes of repair capacity of estimated value to true value, the estimated repair capacity and flexibility of the program's parameters work well.Also empirical studies have been conducted to testify the feasibility of the 2PL-Constrained model in cognitive model of Raven's Advanced Progressive Matrix Test. That is, an integration of the 2PL-Constrained Model with the cognitive theory will not only contribute to explaining the test results, but also contribute to the on-line item generation.In the ending, the thesis presents the conclusion and discussion for the research, and suggests prospects for future studies of the 2PL-Constrained Model in the field of psychology.