Investigating Factorial Invariance of Latent Variables Across Populations When Manifest Variables Are Missing Completely

Keith F. Widaman, Kevin Grimm, Dawnté R. Early, Richard W. Robins, Rand D. Conger

Research output: Contribution to journalArticle

10 Citations (Scopus)

Abstract

Difficulties arise in multiple-group evaluations of factorial invariance if particular manifest variables are missing completely in certain groups. Ad hoc analytic alternatives can be used in such situations (e.g., deleting manifest variables), but some common approaches, such as multiple imputation, are not viable. At least 3 solutions to this problem are viable: analyzing differing sets of variables across groups, using pattern mixture approaches, and a new method using random number generation. The latter solution, proposed in this article, is to generate pseudo-random normal deviates for all observations for manifest variables that are missing completely in a given sample and then to specify multiple-group models in a way that respects the random nature of these values. An empirical example is presented in detail comparing the 3 approaches. The proposed solution can enable quantitative comparisons at the latent variable level between groups using programs that require the same number of manifest variables in each group.

Original languageEnglish (US)
Pages (from-to)384-408
Number of pages25
JournalStructural Equation Modeling
Volume20
Issue number3
DOIs
StatePublished - Jul 2013
Externally publishedYes

Fingerprint

Latent Variables
Factorial
Invariance
Random number generation
Group
Random number Generation
Multiple Imputation
Latent variables
Alternatives
Evaluation
evaluation
Values

Keywords

  • confirmatory factor analysis
  • factorial invariance
  • missing data
  • structural equation modeling

ASJC Scopus subject areas

  • Modeling and Simulation
  • Decision Sciences(all)
  • Economics, Econometrics and Finance(all)
  • Sociology and Political Science

Cite this

Investigating Factorial Invariance of Latent Variables Across Populations When Manifest Variables Are Missing Completely. / Widaman, Keith F.; Grimm, Kevin; Early, Dawnté R.; Robins, Richard W.; Conger, Rand D.

In: Structural Equation Modeling, Vol. 20, No. 3, 07.2013, p. 384-408.

Research output: Contribution to journalArticle

Widaman, Keith F. ; Grimm, Kevin ; Early, Dawnté R. ; Robins, Richard W. ; Conger, Rand D. / Investigating Factorial Invariance of Latent Variables Across Populations When Manifest Variables Are Missing Completely. In: Structural Equation Modeling. 2013 ; Vol. 20, No. 3. pp. 384-408.
@article{4ffc94c0023947fb877258473f97cd75,
title = "Investigating Factorial Invariance of Latent Variables Across Populations When Manifest Variables Are Missing Completely",
abstract = "Difficulties arise in multiple-group evaluations of factorial invariance if particular manifest variables are missing completely in certain groups. Ad hoc analytic alternatives can be used in such situations (e.g., deleting manifest variables), but some common approaches, such as multiple imputation, are not viable. At least 3 solutions to this problem are viable: analyzing differing sets of variables across groups, using pattern mixture approaches, and a new method using random number generation. The latter solution, proposed in this article, is to generate pseudo-random normal deviates for all observations for manifest variables that are missing completely in a given sample and then to specify multiple-group models in a way that respects the random nature of these values. An empirical example is presented in detail comparing the 3 approaches. The proposed solution can enable quantitative comparisons at the latent variable level between groups using programs that require the same number of manifest variables in each group.",
keywords = "confirmatory factor analysis, factorial invariance, missing data, structural equation modeling",
author = "Widaman, {Keith F.} and Kevin Grimm and Early, {Dawnt{\'e} R.} and Robins, {Richard W.} and Conger, {Rand D.}",
year = "2013",
month = "7",
doi = "10.1080/10705511.2013.797819",
language = "English (US)",
volume = "20",
pages = "384--408",
journal = "Structural Equation Modeling",
issn = "1070-5511",
publisher = "Psychology Press Ltd",
number = "3",

}

TY - JOUR

T1 - Investigating Factorial Invariance of Latent Variables Across Populations When Manifest Variables Are Missing Completely

AU - Widaman, Keith F.

AU - Grimm, Kevin

AU - Early, Dawnté R.

AU - Robins, Richard W.

AU - Conger, Rand D.

PY - 2013/7

Y1 - 2013/7

N2 - Difficulties arise in multiple-group evaluations of factorial invariance if particular manifest variables are missing completely in certain groups. Ad hoc analytic alternatives can be used in such situations (e.g., deleting manifest variables), but some common approaches, such as multiple imputation, are not viable. At least 3 solutions to this problem are viable: analyzing differing sets of variables across groups, using pattern mixture approaches, and a new method using random number generation. The latter solution, proposed in this article, is to generate pseudo-random normal deviates for all observations for manifest variables that are missing completely in a given sample and then to specify multiple-group models in a way that respects the random nature of these values. An empirical example is presented in detail comparing the 3 approaches. The proposed solution can enable quantitative comparisons at the latent variable level between groups using programs that require the same number of manifest variables in each group.

AB - Difficulties arise in multiple-group evaluations of factorial invariance if particular manifest variables are missing completely in certain groups. Ad hoc analytic alternatives can be used in such situations (e.g., deleting manifest variables), but some common approaches, such as multiple imputation, are not viable. At least 3 solutions to this problem are viable: analyzing differing sets of variables across groups, using pattern mixture approaches, and a new method using random number generation. The latter solution, proposed in this article, is to generate pseudo-random normal deviates for all observations for manifest variables that are missing completely in a given sample and then to specify multiple-group models in a way that respects the random nature of these values. An empirical example is presented in detail comparing the 3 approaches. The proposed solution can enable quantitative comparisons at the latent variable level between groups using programs that require the same number of manifest variables in each group.

KW - confirmatory factor analysis

KW - factorial invariance

KW - missing data

KW - structural equation modeling

UR - http://www.scopus.com/inward/record.url?scp=84880930965&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84880930965&partnerID=8YFLogxK

U2 - 10.1080/10705511.2013.797819

DO - 10.1080/10705511.2013.797819

M3 - Article

AN - SCOPUS:84880930965

VL - 20

SP - 384

EP - 408

JO - Structural Equation Modeling

JF - Structural Equation Modeling

SN - 1070-5511

IS - 3

ER -