EcoSta 2019: Start Registration
View Submission - EcoSta2019
Title: Marginalized maximum a posteriori estimation for the 4-parameter logistic model under a mixture modeling framework Authors:  Xiangbin Meng - Northeast Normal University (China) [presenting]
Gongjun Xu - University of Michigan (United States)
Abstract: The 4-parameter logistic model (4PLM) has recently gained great interests in various applications. The 4PLM is reexpresed to be a mixture model with two levels of latent variables and further develop a marginalized maximum a posteriori (MMAP) estimation with an Expectation-Maximization (EM) algorithm. The mixture modeling framework of the 4PLM not only makes the proposed EM algorithm more easily to be implemented in practice, but also provides a natural connection with the popular cognitive diagnosis models. Simulation studies were constructed to show the good performance of the proposed estimation method and to investigate the impact of the additional upper asymptote parameter on the estimation of other parameters. Moreover, a real data set was analyzed by the 4PLM to show its outperformance over the 3-parameter logistic model (3PLM).