Regression mixture models

Does modeling the covariance between independent variables and latent classes improve the results?

A.E. Lamont, J.K. Vermunt, M.L. Van Horn

Research output: Contribution to journalArticleScientificpeer-review

Abstract

Regression mixture models are increasingly used as an exploratory approach to identify heterogeneity in the effects of a predictor on an outcome. In this simulation study, we tested the effects of violating an implicit assumption often made in these models; that is, independent variables in the model are not directly related to latent classes. Results indicate that the major risk of failing to model the relationship between predictor and latent class was an increase in the probability of selecting additional latent classes and biased class proportions. In addition, we tested whether regression mixture models can detect a piecewise relationship between a predictor and outcome. Results suggest that these models are able to detect piecewise relations but only when the relationship between the latent class and the predictor is included in model estimation. We illustrate the implications of making this assumption through a reanalysis of applied data examining heterogeneity in the effects of family resources on academic achievement. We compare previous results (which assumed no relation between independent variables and latent class) to the model where this assumption is lifted. Implications and analytic suggestions for conducting regression mixture based on these findings are noted.

Original languageEnglish
Pages (from-to)35-52
JournalMultivariate Behavioral Research
Volume51
Issue number1
DOIs
Publication statusPublished - 2016

Keywords

  • Regression mixture models
  • finite mixture models
  • simulation
  • latent class regression
  • FINITE MIXTURE
  • FAMILY RESOURCES
  • SWITCHING REGRESSIONS
  • SEROTONERGIC GENES
  • HETEROGENEITY
  • DISTRIBUTIONS
  • OFFENDERS

Cite this

@article{6e82dcf95e5349529bb3de4edbfcfdee,
title = "Regression mixture models: Does modeling the covariance between independent variables and latent classes improve the results?",
abstract = "Regression mixture models are increasingly used as an exploratory approach to identify heterogeneity in the effects of a predictor on an outcome. In this simulation study, we tested the effects of violating an implicit assumption often made in these models; that is, independent variables in the model are not directly related to latent classes. Results indicate that the major risk of failing to model the relationship between predictor and latent class was an increase in the probability of selecting additional latent classes and biased class proportions. In addition, we tested whether regression mixture models can detect a piecewise relationship between a predictor and outcome. Results suggest that these models are able to detect piecewise relations but only when the relationship between the latent class and the predictor is included in model estimation. We illustrate the implications of making this assumption through a reanalysis of applied data examining heterogeneity in the effects of family resources on academic achievement. We compare previous results (which assumed no relation between independent variables and latent class) to the model where this assumption is lifted. Implications and analytic suggestions for conducting regression mixture based on these findings are noted.",
keywords = "Regression mixture models, finite mixture models, simulation, latent class regression, FINITE MIXTURE, FAMILY RESOURCES, SWITCHING REGRESSIONS, SEROTONERGIC GENES, HETEROGENEITY, DISTRIBUTIONS, OFFENDERS",
author = "A.E. Lamont and J.K. Vermunt and {Van Horn}, M.L.",
year = "2016",
doi = "10.1080/00273171.2015.1095063",
language = "English",
volume = "51",
pages = "35--52",
journal = "Multivariate Behavioral Research",
issn = "0027-3171",
publisher = "ROUTLEDGE JOURNALS, TAYLOR & FRANCIS LTD",
number = "1",

}

Regression mixture models : Does modeling the covariance between independent variables and latent classes improve the results? / Lamont, A.E.; Vermunt, J.K.; Van Horn, M.L.

In: Multivariate Behavioral Research, Vol. 51, No. 1, 2016, p. 35-52.

Research output: Contribution to journalArticleScientificpeer-review

TY - JOUR

T1 - Regression mixture models

T2 - Does modeling the covariance between independent variables and latent classes improve the results?

AU - Lamont, A.E.

AU - Vermunt, J.K.

AU - Van Horn, M.L.

PY - 2016

Y1 - 2016

N2 - Regression mixture models are increasingly used as an exploratory approach to identify heterogeneity in the effects of a predictor on an outcome. In this simulation study, we tested the effects of violating an implicit assumption often made in these models; that is, independent variables in the model are not directly related to latent classes. Results indicate that the major risk of failing to model the relationship between predictor and latent class was an increase in the probability of selecting additional latent classes and biased class proportions. In addition, we tested whether regression mixture models can detect a piecewise relationship between a predictor and outcome. Results suggest that these models are able to detect piecewise relations but only when the relationship between the latent class and the predictor is included in model estimation. We illustrate the implications of making this assumption through a reanalysis of applied data examining heterogeneity in the effects of family resources on academic achievement. We compare previous results (which assumed no relation between independent variables and latent class) to the model where this assumption is lifted. Implications and analytic suggestions for conducting regression mixture based on these findings are noted.

AB - Regression mixture models are increasingly used as an exploratory approach to identify heterogeneity in the effects of a predictor on an outcome. In this simulation study, we tested the effects of violating an implicit assumption often made in these models; that is, independent variables in the model are not directly related to latent classes. Results indicate that the major risk of failing to model the relationship between predictor and latent class was an increase in the probability of selecting additional latent classes and biased class proportions. In addition, we tested whether regression mixture models can detect a piecewise relationship between a predictor and outcome. Results suggest that these models are able to detect piecewise relations but only when the relationship between the latent class and the predictor is included in model estimation. We illustrate the implications of making this assumption through a reanalysis of applied data examining heterogeneity in the effects of family resources on academic achievement. We compare previous results (which assumed no relation between independent variables and latent class) to the model where this assumption is lifted. Implications and analytic suggestions for conducting regression mixture based on these findings are noted.

KW - Regression mixture models

KW - finite mixture models

KW - simulation

KW - latent class regression

KW - FINITE MIXTURE

KW - FAMILY RESOURCES

KW - SWITCHING REGRESSIONS

KW - SEROTONERGIC GENES

KW - HETEROGENEITY

KW - DISTRIBUTIONS

KW - OFFENDERS

U2 - 10.1080/00273171.2015.1095063

DO - 10.1080/00273171.2015.1095063

M3 - Article

VL - 51

SP - 35

EP - 52

JO - Multivariate Behavioral Research

JF - Multivariate Behavioral Research

SN - 0027-3171

IS - 1

ER -