An improved stochastic EM algorithm for large-scale full-information item factor analysis

Zhang, Siliang, Chen, Yunxiao ORCID: 0000-0002-7215-2324 and Liu, Yang (2018) An improved stochastic EM algorithm for large-scale full-information item factor analysis. British Journal of Mathematical and Statistical Psychology. ISSN 0007-1102

Text - Accepted Version Download (779kB)

• Author

Abstract

In this paper, we explore the use of the stochastic EM algorithm (Celeux & Diebolt, 1985) for large-scale full-information item factor analysis. Innovations have been made on its implementation, including (1) an adaptive-rejection-based Gibbs sampler for the stochastic E step, (2) a proximal gradient descent algorithm for the optimization in the M step, and (3) diagnostic procedures for determining the burn-in size and the stopping of the algorithm. These developments are based on the theoretical results of Nielsen (2000), as well as advanced sampling and optimization techniques. The proposed algorithm is computationally efficient and virtually tuning-free, making it scalable to large-scale data with many latent traits (e.g. more than five latent traits) and easy to use for practitioners. Standard errors of parameter estimation are also obtained based on the missing information identity (Louis, 1982). The performance of the algorithm is evaluated through simulation studies and an application to the analysis of the IPIP-NEO personality inventory. Extensions of the proposed algorithm to other latent variable models are discussed.

Item Type:	Article
Official URL:	https://onlinelibrary.wiley.com/journal/20448317
Additional Information:	© 2018 The British Psychological Society
Divisions:	Statistics
Subjects:	Q Science > QA Mathematics
Date Deposited:	06 Dec 2018 10:15
Last Modified:	17 Nov 2024 02:30
URI:	http://eprints.lse.ac.uk/id/eprint/91027

Actions (login required)

View Item