Missing Not at Random Models for Latent Growth Curve Analyses

Craig K. Enders

Research output: Contribution to journalArticle

124 Citations (Scopus)

Abstract

The past decade has seen a noticeable shift in missing data handling techniques that assume a missing at random (MAR) mechanism, where the propensity for missing data on an outcome is related to other analysis variables. Although MAR is often reasonable, there are situations where this assumption is unlikely to hold, leading to biased parameter estimates. One such example is a longitudinal study of substance use where participants with the highest frequency of use also have the highest likelihood of attrition, even after controlling for other correlates of missingness. There is a large body of literature on missing not at random (MNAR) analysis models for longitudinal data, particularly in the field of biostatistics. Because these methods allow for a relationship between the outcome variable and the propensity for missing data, they require a weaker assumption about the missing data mechanism. This article describes 2 classic MNAR modeling approaches for longitudinal data: the selection model and the pattern mixture model. To date, these models have been slow to migrate to the social sciences, in part because they required complicated custom computer programs. These models are now quite easy to estimate in popular structural equation modeling programs, particularly Mplus. The purpose of this article is to describe these MNAR modeling frameworks and to illustrate their application on a real data set. Despite their potential advantages, MNAR-based analyses are not without problems and also rely on untestable assumptions. This article offers practical advice for implementing and choosing among different longitudinal models.

Original languageEnglish (US)
Pages (from-to)1-16
Number of pages16
JournalPsychological Methods
Volume16
Issue number1
DOIs
StatePublished - Mar 2011

Fingerprint

Biostatistics
Social Sciences
Longitudinal Studies
Software
Growth
Datasets
Growth Curve

Keywords

  • Attrition
  • Missing data
  • Missing not at random
  • Pattern mixture model
  • Selection model

ASJC Scopus subject areas

  • Psychology (miscellaneous)
  • History and Philosophy of Science

Cite this

Missing Not at Random Models for Latent Growth Curve Analyses. / Enders, Craig K.

In: Psychological Methods, Vol. 16, No. 1, 03.2011, p. 1-16.

Research output: Contribution to journalArticle

Enders, Craig K. / Missing Not at Random Models for Latent Growth Curve Analyses. In: Psychological Methods. 2011 ; Vol. 16, No. 1. pp. 1-16.
@article{e4c2626352384acea0d3a60b65f754a4,
title = "Missing Not at Random Models for Latent Growth Curve Analyses",
abstract = "The past decade has seen a noticeable shift in missing data handling techniques that assume a missing at random (MAR) mechanism, where the propensity for missing data on an outcome is related to other analysis variables. Although MAR is often reasonable, there are situations where this assumption is unlikely to hold, leading to biased parameter estimates. One such example is a longitudinal study of substance use where participants with the highest frequency of use also have the highest likelihood of attrition, even after controlling for other correlates of missingness. There is a large body of literature on missing not at random (MNAR) analysis models for longitudinal data, particularly in the field of biostatistics. Because these methods allow for a relationship between the outcome variable and the propensity for missing data, they require a weaker assumption about the missing data mechanism. This article describes 2 classic MNAR modeling approaches for longitudinal data: the selection model and the pattern mixture model. To date, these models have been slow to migrate to the social sciences, in part because they required complicated custom computer programs. These models are now quite easy to estimate in popular structural equation modeling programs, particularly Mplus. The purpose of this article is to describe these MNAR modeling frameworks and to illustrate their application on a real data set. Despite their potential advantages, MNAR-based analyses are not without problems and also rely on untestable assumptions. This article offers practical advice for implementing and choosing among different longitudinal models.",
keywords = "Attrition, Missing data, Missing not at random, Pattern mixture model, Selection model",
author = "Enders, {Craig K.}",
year = "2011",
month = "3",
doi = "10.1037/a0022640",
language = "English (US)",
volume = "16",
pages = "1--16",
journal = "Psychological Methods",
issn = "1082-989X",
publisher = "American Psychological Association Inc.",
number = "1",

}

TY - JOUR

T1 - Missing Not at Random Models for Latent Growth Curve Analyses

AU - Enders, Craig K.

PY - 2011/3

Y1 - 2011/3

N2 - The past decade has seen a noticeable shift in missing data handling techniques that assume a missing at random (MAR) mechanism, where the propensity for missing data on an outcome is related to other analysis variables. Although MAR is often reasonable, there are situations where this assumption is unlikely to hold, leading to biased parameter estimates. One such example is a longitudinal study of substance use where participants with the highest frequency of use also have the highest likelihood of attrition, even after controlling for other correlates of missingness. There is a large body of literature on missing not at random (MNAR) analysis models for longitudinal data, particularly in the field of biostatistics. Because these methods allow for a relationship between the outcome variable and the propensity for missing data, they require a weaker assumption about the missing data mechanism. This article describes 2 classic MNAR modeling approaches for longitudinal data: the selection model and the pattern mixture model. To date, these models have been slow to migrate to the social sciences, in part because they required complicated custom computer programs. These models are now quite easy to estimate in popular structural equation modeling programs, particularly Mplus. The purpose of this article is to describe these MNAR modeling frameworks and to illustrate their application on a real data set. Despite their potential advantages, MNAR-based analyses are not without problems and also rely on untestable assumptions. This article offers practical advice for implementing and choosing among different longitudinal models.

AB - The past decade has seen a noticeable shift in missing data handling techniques that assume a missing at random (MAR) mechanism, where the propensity for missing data on an outcome is related to other analysis variables. Although MAR is often reasonable, there are situations where this assumption is unlikely to hold, leading to biased parameter estimates. One such example is a longitudinal study of substance use where participants with the highest frequency of use also have the highest likelihood of attrition, even after controlling for other correlates of missingness. There is a large body of literature on missing not at random (MNAR) analysis models for longitudinal data, particularly in the field of biostatistics. Because these methods allow for a relationship between the outcome variable and the propensity for missing data, they require a weaker assumption about the missing data mechanism. This article describes 2 classic MNAR modeling approaches for longitudinal data: the selection model and the pattern mixture model. To date, these models have been slow to migrate to the social sciences, in part because they required complicated custom computer programs. These models are now quite easy to estimate in popular structural equation modeling programs, particularly Mplus. The purpose of this article is to describe these MNAR modeling frameworks and to illustrate their application on a real data set. Despite their potential advantages, MNAR-based analyses are not without problems and also rely on untestable assumptions. This article offers practical advice for implementing and choosing among different longitudinal models.

KW - Attrition

KW - Missing data

KW - Missing not at random

KW - Pattern mixture model

KW - Selection model

UR - http://www.scopus.com/inward/record.url?scp=79952612356&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=79952612356&partnerID=8YFLogxK

U2 - 10.1037/a0022640

DO - 10.1037/a0022640

M3 - Article

C2 - 21381816

AN - SCOPUS:79952612356

VL - 16

SP - 1

EP - 16

JO - Psychological Methods

JF - Psychological Methods

SN - 1082-989X

IS - 1

ER -