Specification of Variance-Covariance Structure in Bivariate Mixed Model for Unequally Time-Spaced Longitudinal Data

Authors

  • Melike Bahçecitapar Department of Statistics, Faculty of Science, Hacettepe University, Ankara, Turkey
  • Özge Karadag Department of Statistics, Faculty of Science, Hacettepe University, Ankara, Turkey
  • Serpil Aktas Department of Statistics, Faculty of Science, Hacettepe University, Ankara, Turkey

DOI:

https://doi.org/10.6000/1929-6029.2015.04.04.6

Keywords:

Multivariate longitudinal data, mixed models, covariance structures

Abstract

In medical studies, the longitudinal data sets obtained from more than one response variables and covariates are mostly analyzed to investigate the change in repeated measurements of each subject at different time points. In this study, the usability of multivariate models in the analysis of these kind of data sets is investigated, because it provides the joint analysis of multiple response variables over time and enables researchers to examine both the correlations of response variables and autocorrelation between measurements from each response variable over time. It has been shown that different parameter estimation methods affect the results in the analysis of multivariate unbalanced longitudinal data. We investigated that autocorrelation structure over time between measurements from same response variable should be truly specified. We also illustrated and compared the simpler, more standard models for fixed effects with multivariate models provided by SAS on a real-life data set in the joint analysis of two response variables. Results show that misspecification of autocorrelation structures has a negative impact on the parameter estimates and parameter estimation method should become of interest.

References

Harvey, DJ, Beckett LA, Mungas DM. Multivariate modelling of two associated cognitive outcomes in a longitudinal study. J Alzheimer’s Dis 2003; 5(5): 357-365. DOI: https://doi.org/10.3233/JAD-2003-5502

Hoekstra T, Twisk JWR. The analysis of individual health trajectories across the life couse: latent class growth models versus mixed models. The Series Life Course Research and Social Policies 2015; Vol. 4: pp. 179-195. DOI: https://doi.org/10.1007/978-3-319-20484-0_9

Galecki AT. General class of covariance structures for two or more repeated factors in longitudinal data analysis. Communications in Statistics - Theory and Methods 1994; 23: 3105-3119. http://dx.doi.org/10.1080/03610929408831436 DOI: https://doi.org/10.1080/03610929408831436

Littell RC, Milliken GA, Stroup WW, Wolfinger RD. SAS System for Mixed Models, Cary, NC: SAS Institute Inc. 1996.

Hatcher L. A step-by step approach to using the SAS System for factor analysis and structural equation modeling, Cary, NC: SAS Institute Inc. 1998.

Gao F, Xiong C, Thompson P, Miller JP. Analyzing multivariate longitudinal data using SAS, SUGI 31 Proceedings 2006.

Sy JP, Taylor JM,Cumberland WG. A stochastic model for the analysis of bivariate longitudinal AIDS data. Biometrics 1997; 53: 542-55. http://dx.doi.org/10.2307/2533956 DOI: https://doi.org/10.2307/2533956

Laird NM, Ware JH. Random-effects models for longitudinal data, Biometrics 1982; 38: 963-74. http://dx.doi.org/10.2307/2529876 DOI: https://doi.org/10.2307/2529876

Hedeker D, Gibbons RD. Longitudinal data analysis. Wiley. 2006.

Harville DA. Maximum Likelihood approaches to variance component estimation and to related problems. J Am Stat Assoc 1977; 72: 320-340. http://dx.doi.org/10.1080/01621459.1977.10480998 DOI: https://doi.org/10.1080/01621459.1977.10480998

Rao CR. Estimation of variance and covariance components-MINQUE theory. Journal of Multivariate Analysis 1971; 1: 257-275. http://dx.doi.org/10.1016/0047-259X(71)90001-7 DOI: https://doi.org/10.1016/0047-259X(71)90001-7

Wolfinger RD, Tobias RD, Sall J. Computing Gaussian Likelihoods and Their Derivatives for General Linear Mixed Models. SIAM Journal on Scientific Computing 1994; 15(6): 1294-1310. http://dx.doi.org/10.1137/0915079 DOI: https://doi.org/10.1137/0915079

Dempster AB, Laird NM, Rubin DB. Max. Likelihood from incomplete data via the EM algorithm.Journal of the Royal Statistical Society B 1977; 39: 1-38. DOI: https://doi.org/10.1111/j.2517-6161.1977.tb01600.x

Lindstrom MJ, Bates DM. Newton-Raphson and EM Algorithm for linear mixed-effects models for repeated-measures data. JASA 1988; 83: 1014-22. DOI: https://doi.org/10.1080/01621459.1988.10478693

SAS/IML Software: usage and Reference. Version 6., (SAS Institute Inc., Carry, NC), 1990.

Downloads

Published

2015-11-02

How to Cite

Bahçecitapar, M., Karadag, Özge ., & Aktas, S. (2015). Specification of Variance-Covariance Structure in Bivariate Mixed Model for Unequally Time-Spaced Longitudinal Data. International Journal of Statistics in Medical Research, 4(4), 370–377. https://doi.org/10.6000/1929-6029.2015.04.04.6

Issue

Section

General Articles