Online Program Home
  My Program

Abstract Details

Activity Number: 423 - SPEED: Biopharmaceutical Statistics, Medical Devices, and Mental Health
Type: Contributed
Date/Time: Tuesday, August 1, 2017 : 2:00 PM to 2:45 PM
Sponsor: Mental Health Statistics Section
Abstract #325295
Title: Identifying Latent Structures in Restricted Latent Class Models
Author(s): Zhuoran Shang* and Gongjun Xu
Companies: University of Minnesota and University of Michigan
Keywords: Restricted latent class models ; Cognitive Diagnosis ; Q-matrix
Abstract:

This study focuses on a family of restricted latent structure models, where the model parameters are restricted via a latent structure matrix to reflect pre-specified assumptions on the latent attributes. In application such a latent structure matrix is often provided by experts and assumed to be correct upon construction, yet it may be subjective and Mis-specified. Recognizing this problem, researchers have been developing methods to estimate the structure matrix from data. The first goal is to establish identifiability conditions that ensure the estimability of the structure matrix. The results provide theoretical justification for the existing estimation methods as well as a guideline for the related experimental designs. The second part proposes an information-based model selection method to estimate the latent structure. We consider two important cases in practice: (1) misspecification detection in confirmatory analysis where a pre-specified matrix is provided by experts; and (2) estimation of the whole latent matrix in exploratory analysis. Simulation and case studies show that the proposed method outperforms the existing approaches.


Authors who are presenting talks have a * after their name.

Back to the full JSM 2017 program

 
 
Copyright © American Statistical Association