Small sample estimation properties of longitudinal count models

Melanie L Bell, Gary K. Grunwald

Research output: Contribution to journalArticle

4 Citations (Scopus)

Abstract

Motivated by the need to validly analyse data from small longitudinal designs with count outcomes, we carried out a simulation study with a two-group longitudinal design with as few as five subjects per group and three measurements per subject. Various correlation structures related to auto regression and/or subject heterogeneity were used to simulate, most of which did not correspond to any of the methods being assessed and so induced some model misspecification. We evaluated validity (Type I error rate) and efficiency (power) for the interaction effect using several methods based on generalized linear mixed models (GLMM) or generalized estimating equations (GEE). Conclusions included that Type I error rates were too high for GEE using sandwich standard errors, and too low for GLMM. Generally, Type I error rate was improved with more subjects but not with more measurements per subjects, and both improved power. Many of our results differ from corresponding results in other cases (e.g. logistic models), emphasizing a diversity of behaviour for various non-normal outcomes.

Original languageEnglish (US)
Pages (from-to)1067-1079
Number of pages13
JournalJournal of Statistical Computation and Simulation
Volume81
Issue number9
DOIs
StatePublished - Sep 2011
Externally publishedYes

Fingerprint

Type I Error Rate
Small Sample
Generalized Linear Mixed Model
Count
Generalized Estimating Equations
Model Misspecification
Autoregression
Interaction Effects
Logistic Model
Sandwich
Correlation Structure
Standard error
Simulation Study
Model
Logistics
Small sample
Type I error
Count models
Design
Generalized estimating equations

Keywords

  • Bias
  • Correlated data
  • GEE
  • GLMM
  • Interaction
  • Mancl and derouen correction
  • Poisson
  • REPL
  • Sandwich standard errors
  • Type I error rate

ASJC Scopus subject areas

  • Applied Mathematics
  • Statistics and Probability
  • Modeling and Simulation
  • Statistics, Probability and Uncertainty

Cite this

Small sample estimation properties of longitudinal count models. / Bell, Melanie L; Grunwald, Gary K.

In: Journal of Statistical Computation and Simulation, Vol. 81, No. 9, 09.2011, p. 1067-1079.

Research output: Contribution to journalArticle

@article{b53d4fcf049541eb8d037087850eb279,
title = "Small sample estimation properties of longitudinal count models",
abstract = "Motivated by the need to validly analyse data from small longitudinal designs with count outcomes, we carried out a simulation study with a two-group longitudinal design with as few as five subjects per group and three measurements per subject. Various correlation structures related to auto regression and/or subject heterogeneity were used to simulate, most of which did not correspond to any of the methods being assessed and so induced some model misspecification. We evaluated validity (Type I error rate) and efficiency (power) for the interaction effect using several methods based on generalized linear mixed models (GLMM) or generalized estimating equations (GEE). Conclusions included that Type I error rates were too high for GEE using sandwich standard errors, and too low for GLMM. Generally, Type I error rate was improved with more subjects but not with more measurements per subjects, and both improved power. Many of our results differ from corresponding results in other cases (e.g. logistic models), emphasizing a diversity of behaviour for various non-normal outcomes.",
keywords = "Bias, Correlated data, GEE, GLMM, Interaction, Mancl and derouen correction, Poisson, REPL, Sandwich standard errors, Type I error rate",
author = "Bell, {Melanie L} and Grunwald, {Gary K.}",
year = "2011",
month = "9",
doi = "10.1080/00949651003674144",
language = "English (US)",
volume = "81",
pages = "1067--1079",
journal = "Journal of Statistical Computation and Simulation",
issn = "0094-9655",
publisher = "Taylor and Francis Ltd.",
number = "9",

}

TY - JOUR

T1 - Small sample estimation properties of longitudinal count models

AU - Bell, Melanie L

AU - Grunwald, Gary K.

PY - 2011/9

Y1 - 2011/9

N2 - Motivated by the need to validly analyse data from small longitudinal designs with count outcomes, we carried out a simulation study with a two-group longitudinal design with as few as five subjects per group and three measurements per subject. Various correlation structures related to auto regression and/or subject heterogeneity were used to simulate, most of which did not correspond to any of the methods being assessed and so induced some model misspecification. We evaluated validity (Type I error rate) and efficiency (power) for the interaction effect using several methods based on generalized linear mixed models (GLMM) or generalized estimating equations (GEE). Conclusions included that Type I error rates were too high for GEE using sandwich standard errors, and too low for GLMM. Generally, Type I error rate was improved with more subjects but not with more measurements per subjects, and both improved power. Many of our results differ from corresponding results in other cases (e.g. logistic models), emphasizing a diversity of behaviour for various non-normal outcomes.

AB - Motivated by the need to validly analyse data from small longitudinal designs with count outcomes, we carried out a simulation study with a two-group longitudinal design with as few as five subjects per group and three measurements per subject. Various correlation structures related to auto regression and/or subject heterogeneity were used to simulate, most of which did not correspond to any of the methods being assessed and so induced some model misspecification. We evaluated validity (Type I error rate) and efficiency (power) for the interaction effect using several methods based on generalized linear mixed models (GLMM) or generalized estimating equations (GEE). Conclusions included that Type I error rates were too high for GEE using sandwich standard errors, and too low for GLMM. Generally, Type I error rate was improved with more subjects but not with more measurements per subjects, and both improved power. Many of our results differ from corresponding results in other cases (e.g. logistic models), emphasizing a diversity of behaviour for various non-normal outcomes.

KW - Bias

KW - Correlated data

KW - GEE

KW - GLMM

KW - Interaction

KW - Mancl and derouen correction

KW - Poisson

KW - REPL

KW - Sandwich standard errors

KW - Type I error rate

UR - http://www.scopus.com/inward/record.url?scp=80051704560&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=80051704560&partnerID=8YFLogxK

U2 - 10.1080/00949651003674144

DO - 10.1080/00949651003674144

M3 - Article

AN - SCOPUS:80051704560

VL - 81

SP - 1067

EP - 1079

JO - Journal of Statistical Computation and Simulation

JF - Journal of Statistical Computation and Simulation

SN - 0094-9655

IS - 9

ER -