Protecting against nonrandomly missing data in longitudinal studies

C. H. Brown

Research output: Contribution to journalArticle

66 Citations (Scopus)

Abstract

Nonrandomly missing data can pose serious problems in longitudinal studies. We generally have little knowledge about how missingness is related to the data values, and longitudinal studies are often far from complete. Two approaches that have been used to handle missing data-use of maximum likelihood with an ignorable mechanism and direct modeling of the missing data mechanism-have the disadvantage of not giving consistent estimates under important classes of nonrandom mechanisms. We introduce two protective estimators, that is, estimators that retain their consistency over a wide range of nonrandom mechanisms. We compare these protective estimators using longitudinal data from a mental health panel study. We also investigate their robustness to certain departures from normality.

Original languageEnglish
Pages (from-to)143-155
Number of pages13
JournalBiometrics
Volume46
Issue number1
DOIs
StatePublished - Jul 25 1990

Fingerprint

Longitudinal Study
longitudinal studies
Missing Data
Maximum likelihood
Longitudinal Studies
Health
Estimator
mental health
Missing Data Mechanism
Mental Health
Consistent Estimates
Longitudinal Data
Normality
Maximum Likelihood
Robustness
Modeling
Range of data

ASJC Scopus subject areas

  • Agricultural and Biological Sciences(all)
  • Public Health, Environmental and Occupational Health
  • Agricultural and Biological Sciences (miscellaneous)
  • Applied Mathematics
  • Statistics and Probability

Cite this

Protecting against nonrandomly missing data in longitudinal studies. / Brown, C. H.

In: Biometrics, Vol. 46, No. 1, 25.07.1990, p. 143-155.

Research output: Contribution to journalArticle

Brown, C. H. / Protecting against nonrandomly missing data in longitudinal studies. In: Biometrics. 1990 ; Vol. 46, No. 1. pp. 143-155.
@article{80a7e0b516854db8aa1278eb69c9a640,
title = "Protecting against nonrandomly missing data in longitudinal studies",
abstract = "Nonrandomly missing data can pose serious problems in longitudinal studies. We generally have little knowledge about how missingness is related to the data values, and longitudinal studies are often far from complete. Two approaches that have been used to handle missing data-use of maximum likelihood with an ignorable mechanism and direct modeling of the missing data mechanism-have the disadvantage of not giving consistent estimates under important classes of nonrandom mechanisms. We introduce two protective estimators, that is, estimators that retain their consistency over a wide range of nonrandom mechanisms. We compare these protective estimators using longitudinal data from a mental health panel study. We also investigate their robustness to certain departures from normality.",
author = "Brown, {C. H.}",
year = "1990",
month = "7",
day = "25",
doi = "10.2307/2531637",
language = "English",
volume = "46",
pages = "143--155",
journal = "Biometrics",
issn = "0006-341X",
publisher = "Wiley-Blackwell",
number = "1",

}

TY - JOUR

T1 - Protecting against nonrandomly missing data in longitudinal studies

AU - Brown, C. H.

PY - 1990/7/25

Y1 - 1990/7/25

N2 - Nonrandomly missing data can pose serious problems in longitudinal studies. We generally have little knowledge about how missingness is related to the data values, and longitudinal studies are often far from complete. Two approaches that have been used to handle missing data-use of maximum likelihood with an ignorable mechanism and direct modeling of the missing data mechanism-have the disadvantage of not giving consistent estimates under important classes of nonrandom mechanisms. We introduce two protective estimators, that is, estimators that retain their consistency over a wide range of nonrandom mechanisms. We compare these protective estimators using longitudinal data from a mental health panel study. We also investigate their robustness to certain departures from normality.

AB - Nonrandomly missing data can pose serious problems in longitudinal studies. We generally have little knowledge about how missingness is related to the data values, and longitudinal studies are often far from complete. Two approaches that have been used to handle missing data-use of maximum likelihood with an ignorable mechanism and direct modeling of the missing data mechanism-have the disadvantage of not giving consistent estimates under important classes of nonrandom mechanisms. We introduce two protective estimators, that is, estimators that retain their consistency over a wide range of nonrandom mechanisms. We compare these protective estimators using longitudinal data from a mental health panel study. We also investigate their robustness to certain departures from normality.

UR - http://www.scopus.com/inward/record.url?scp=0025071087&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=0025071087&partnerID=8YFLogxK

U2 - 10.2307/2531637

DO - 10.2307/2531637

M3 - Article

C2 - 2350568

AN - SCOPUS:0025071087

VL - 46

SP - 143

EP - 155

JO - Biometrics

JF - Biometrics

SN - 0006-341X

IS - 1

ER -