The analysis of the influence of birth order and other factors in multiple birth data
Authors:
- 8 December 2003
- journal article
- research article
- Published by Wiley in Statistics in Medicine
- Vol. 22 (24) , 3739-3753
Abstract
We compare three methods which can be used to analyse theinfluence of birth order and other factors on health outcomes inmultiple birth data. We consider marginal models based ongeneralized estimating equations (GEE) and two kinds ofconditional models; conditional logistic regression (CLR) andmixed effects models (MEM). Although the models may be writtensimilarly, there are differences in both the interpretation andthe numerical values assigned to the parameters. Our mainconclusion is that GEE and MEM are preferable to CLR since theyprovide more flexibility in dealing with missing values andcovariates. The choice between GEE and MEM is less obvious anddepends on the data, the parameter of interest and statisticalpower. Copyright © 2003 John Wiley & Sons, Ltd.Keywords
This publication has 11 references indexed in Scilit:
- Cross‐over Trials In Clinical Research2002
- Sample Size and Power Calculations with Correlated Binary DataControlled Clinical Trials, 2001
- Respiratory distress syndrome and birth order in premature twinsArchives of Disease in Childhood - Fetal and Neonatal Edition, 2001
- Between- and Within-Cluster Covariate Effects in the Analysis of Clustered Data1998
- On the appropriateness of marginal models for repeated measurements in clinical trialsStatistics in Medicine, 1998
- Estimation efficiency in a binary mixed-effects model settingBiometrika, 1996
- A Random Effects Model for Binary Data from Crossover Clinical TrialsJournal of the Royal Statistical Society Series C: Applied Statistics, 1992
- Models for Longitudinal Data: A Generalized Estimating Equation ApproachBiometrics, 1988
- Longitudinal data analysis using generalized linear modelsBiometrika, 1986
- Hyaline membrane disease in twinsAmerican Journal of Obstetrics and Gynecology, 1976