Likelihood analysis of multivariate probit models using a parameter expanded MCEM algorithm

Huiping Xu, Bruce A. Craig

Research output: Contribution to journalArticle

5 Citations (Scopus)

Abstract

Multivariate binary data arise in a variety of settings. In this article we propose a practical and efficient computational framework for maximum likelihood estimation of multivariate probit regression models. This approach uses theMonte Carlo expectation maximization (MCEM) algorithm, with parameter expansion to complete the M-step, to avoid the direct evaluation of the intractable multivariate normal orthant probabilities. The parameter expansion not only enables a closed-form solution in the M-step, but also improves efficiency. Using the simulation studies, we compare the performance of our approach with the MCEM algorithms developed by Chib and Greenberg (1998) and Song and Lee (2005), as well as the iterative approach proposed by Li and Schafer (2008). Our approach is further illustrated using a real-world example.

Original languageEnglish (US)
Pages (from-to)340-348
Number of pages9
JournalTechnometrics
Volume52
Issue number3
DOIs
StatePublished - Aug 2010
Externally publishedYes

Fingerprint

Probit Model
Multivariate Models
Likelihood
Probit Regression
Multivariate Regression
Maximum likelihood estimation
Binary Data
Multivariate Normal
Expectation-maximization Algorithm
Multivariate Data
Closed-form Solution
Maximum Likelihood Estimation
Regression Model
Simulation Study
Evaluation
Framework

Keywords

  • Correlated binary data
  • Gibbs sampler
  • Monte Carlo expectation maximization algorithm
  • Parameter expansion

ASJC Scopus subject areas

  • Modeling and Simulation
  • Statistics and Probability
  • Applied Mathematics

Cite this

Likelihood analysis of multivariate probit models using a parameter expanded MCEM algorithm. / Xu, Huiping; Craig, Bruce A.

In: Technometrics, Vol. 52, No. 3, 08.2010, p. 340-348.

Research output: Contribution to journalArticle

@article{9eb251476ad74c65aa73aba30161310c,
title = "Likelihood analysis of multivariate probit models using a parameter expanded MCEM algorithm",
abstract = "Multivariate binary data arise in a variety of settings. In this article we propose a practical and efficient computational framework for maximum likelihood estimation of multivariate probit regression models. This approach uses theMonte Carlo expectation maximization (MCEM) algorithm, with parameter expansion to complete the M-step, to avoid the direct evaluation of the intractable multivariate normal orthant probabilities. The parameter expansion not only enables a closed-form solution in the M-step, but also improves efficiency. Using the simulation studies, we compare the performance of our approach with the MCEM algorithms developed by Chib and Greenberg (1998) and Song and Lee (2005), as well as the iterative approach proposed by Li and Schafer (2008). Our approach is further illustrated using a real-world example.",
keywords = "Correlated binary data, Gibbs sampler, Monte Carlo expectation maximization algorithm, Parameter expansion",
author = "Huiping Xu and Craig, {Bruce A.}",
year = "2010",
month = "8",
doi = "10.1198/TECH.2010.09055",
language = "English (US)",
volume = "52",
pages = "340--348",
journal = "Technometrics",
issn = "0040-1706",
publisher = "American Statistical Association",
number = "3",

}

TY - JOUR

T1 - Likelihood analysis of multivariate probit models using a parameter expanded MCEM algorithm

AU - Xu, Huiping

AU - Craig, Bruce A.

PY - 2010/8

Y1 - 2010/8

N2 - Multivariate binary data arise in a variety of settings. In this article we propose a practical and efficient computational framework for maximum likelihood estimation of multivariate probit regression models. This approach uses theMonte Carlo expectation maximization (MCEM) algorithm, with parameter expansion to complete the M-step, to avoid the direct evaluation of the intractable multivariate normal orthant probabilities. The parameter expansion not only enables a closed-form solution in the M-step, but also improves efficiency. Using the simulation studies, we compare the performance of our approach with the MCEM algorithms developed by Chib and Greenberg (1998) and Song and Lee (2005), as well as the iterative approach proposed by Li and Schafer (2008). Our approach is further illustrated using a real-world example.

AB - Multivariate binary data arise in a variety of settings. In this article we propose a practical and efficient computational framework for maximum likelihood estimation of multivariate probit regression models. This approach uses theMonte Carlo expectation maximization (MCEM) algorithm, with parameter expansion to complete the M-step, to avoid the direct evaluation of the intractable multivariate normal orthant probabilities. The parameter expansion not only enables a closed-form solution in the M-step, but also improves efficiency. Using the simulation studies, we compare the performance of our approach with the MCEM algorithms developed by Chib and Greenberg (1998) and Song and Lee (2005), as well as the iterative approach proposed by Li and Schafer (2008). Our approach is further illustrated using a real-world example.

KW - Correlated binary data

KW - Gibbs sampler

KW - Monte Carlo expectation maximization algorithm

KW - Parameter expansion

UR - http://www.scopus.com/inward/record.url?scp=77955316042&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=77955316042&partnerID=8YFLogxK

U2 - 10.1198/TECH.2010.09055

DO - 10.1198/TECH.2010.09055

M3 - Article

AN - SCOPUS:77955316042

VL - 52

SP - 340

EP - 348

JO - Technometrics

JF - Technometrics

SN - 0040-1706

IS - 3

ER -