An estimating equations approach for the liscomp model

Beth A. Reboussin, Kung Yee Liang

Research output: Contribution to journalArticle

Abstract

Maximum likelihood estimation is computationally infeasible for latent variable models involving multivariate categorical responses, in particular for the LISCOMP model. A three-stage generalized least squares approach introduced by Muthén (1983, 1984) can experience problems of instability, bias, non-convergence, and non-positive definiteness of weight matrices in situations of low prevalence, small sample size and large numbers of observed indicator variables. We propose a quadratic estimating equations approach that only requires specification of the first two moments. By performing simultaneous estimation of parameters, this method does not encounter the problems mentioned above and experiences gains in efficiency. Methods are compared through a numerical study and an application to a study of life-events and neurotic illness.

Original languageEnglish (US)
Pages (from-to)165-182
Number of pages18
JournalPsychometrika
Volume63
Issue number2
StatePublished - Jun 1998
Externally publishedYes

Fingerprint

Estimating Equation
Simultaneous Estimation
Latent Variable Models
Generalized Least Squares
Quadratic equation
Maximum likelihood estimation
Small Sample Size
Least-Squares Analysis
Maximum Likelihood Estimation
Categorical
Sample Size
Numerical Study
experience
illness
Specification
Moment
Specifications
Weights and Measures
efficiency
event

Keywords

  • Estimating equations
  • Latent variable
  • LISCOMP model
  • Multivariate binary data

ASJC Scopus subject areas

  • Mathematics (miscellaneous)
  • Psychology(all)
  • Psychology (miscellaneous)
  • Social Sciences (miscellaneous)

Cite this

Reboussin, B. A., & Liang, K. Y. (1998). An estimating equations approach for the liscomp model. Psychometrika, 63(2), 165-182.

An estimating equations approach for the liscomp model. / Reboussin, Beth A.; Liang, Kung Yee.

In: Psychometrika, Vol. 63, No. 2, 06.1998, p. 165-182.

Research output: Contribution to journalArticle

Reboussin, BA & Liang, KY 1998, 'An estimating equations approach for the liscomp model', Psychometrika, vol. 63, no. 2, pp. 165-182.
Reboussin BA, Liang KY. An estimating equations approach for the liscomp model. Psychometrika. 1998 Jun;63(2):165-182.
Reboussin, Beth A. ; Liang, Kung Yee. / An estimating equations approach for the liscomp model. In: Psychometrika. 1998 ; Vol. 63, No. 2. pp. 165-182.
@article{7c619f95f4534fb89a3e5829411dfc7c,
title = "An estimating equations approach for the liscomp model",
abstract = "Maximum likelihood estimation is computationally infeasible for latent variable models involving multivariate categorical responses, in particular for the LISCOMP model. A three-stage generalized least squares approach introduced by Muth{\'e}n (1983, 1984) can experience problems of instability, bias, non-convergence, and non-positive definiteness of weight matrices in situations of low prevalence, small sample size and large numbers of observed indicator variables. We propose a quadratic estimating equations approach that only requires specification of the first two moments. By performing simultaneous estimation of parameters, this method does not encounter the problems mentioned above and experiences gains in efficiency. Methods are compared through a numerical study and an application to a study of life-events and neurotic illness.",
keywords = "Estimating equations, Latent variable, LISCOMP model, Multivariate binary data",
author = "Reboussin, {Beth A.} and Liang, {Kung Yee}",
year = "1998",
month = "6",
language = "English (US)",
volume = "63",
pages = "165--182",
journal = "Psychometrika",
issn = "0033-3123",
publisher = "Springer New York",
number = "2",

}

TY - JOUR

T1 - An estimating equations approach for the liscomp model

AU - Reboussin, Beth A.

AU - Liang, Kung Yee

PY - 1998/6

Y1 - 1998/6

N2 - Maximum likelihood estimation is computationally infeasible for latent variable models involving multivariate categorical responses, in particular for the LISCOMP model. A three-stage generalized least squares approach introduced by Muthén (1983, 1984) can experience problems of instability, bias, non-convergence, and non-positive definiteness of weight matrices in situations of low prevalence, small sample size and large numbers of observed indicator variables. We propose a quadratic estimating equations approach that only requires specification of the first two moments. By performing simultaneous estimation of parameters, this method does not encounter the problems mentioned above and experiences gains in efficiency. Methods are compared through a numerical study and an application to a study of life-events and neurotic illness.

AB - Maximum likelihood estimation is computationally infeasible for latent variable models involving multivariate categorical responses, in particular for the LISCOMP model. A three-stage generalized least squares approach introduced by Muthén (1983, 1984) can experience problems of instability, bias, non-convergence, and non-positive definiteness of weight matrices in situations of low prevalence, small sample size and large numbers of observed indicator variables. We propose a quadratic estimating equations approach that only requires specification of the first two moments. By performing simultaneous estimation of parameters, this method does not encounter the problems mentioned above and experiences gains in efficiency. Methods are compared through a numerical study and an application to a study of life-events and neurotic illness.

KW - Estimating equations

KW - Latent variable

KW - LISCOMP model

KW - Multivariate binary data

UR - http://www.scopus.com/inward/record.url?scp=0032391470&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=0032391470&partnerID=8YFLogxK

M3 - Article

AN - SCOPUS:0032391470

VL - 63

SP - 165

EP - 182

JO - Psychometrika

JF - Psychometrika

SN - 0033-3123

IS - 2

ER -