Modelling conditional covariance in the linear mixed model

Abstract

We provide a data-driven method for modelling the conditional, within-subject covariance matrix arising in linear mixed models (Laird and Ware, 1982). Given an agreed structure for the between-subject covariance matrix we use a regression equation approach to model the within-subject covariance matrix. Using an EM algorithm we estimate all of the parameters in the model simultaneously and obtain analytical expressions for the standard errors. By re-analyzing Kenward's (1987) cattle data, we compare our new model with classical menu-selection–based modelling techniques, demonstrating its superiority using the Bayesian Information Criterion. We also conduct a simulation study, which confirms our observational findings. The paper extends our previous covariance modeling work (Pan and MacKenzie, 2003, 2006) to the conditional covariance space of the linear mixed model (LMM).

Keywords

Cholesky decomposition conditional covariance EM algorithm joint mean–covariance models linear mixed models longitudinal data

Get full access to this article

View all access options for this article.

References

Anderson TW (1973) Asymptotically efficient estimation of covariance matrices with linear structure. ratio . Annals of Statistics, 25, 573–78 .

Chiu TYM , Leonard T and Tsui KW (1996) The matrix–logarithm covariance model . Journal of the American Statistical Association, 91, 198–210 .

Daniels MJ and Pourahmadi M (2002a) Bayesian analysis of covariance matrices and dynamic models for longitudinal data . Biometrika 89, 553–66 .

Daniels MJ and Zhao YD (2003) Modelling the random effects covariance matrix in longitudinal data . Statistics in Medicine, 22, 1631–47 .

Diggle PJ , Liang KY and Zeger SL (1994) Analysis of longitudinal data. Oxford: Oxford University Press .

Diggle PJ and Verbyla AP (1998) Nonparametric estimation of covariance structure in longitudinal data . Biometrics, 54, 403–15 .

Goldstein H , Healy MJR and Rasbash J (1994) Multilevel time series models with applications to repeated measures data . Statistics in Medicine, 13, 1643–55 .

Harville DA (1977) Maximum likelihood approaches to variance component estimation and to related problems . Journal of the American Statistical Association, 72, 320–38 .

Kenward MG (1987) A method for comparing profiles of repeated measurements . Applied Statistics, 36, 296–308 .

10.

Laird NM and Ware JJ (1982) Random–effects models for longitudinal data . Biometrics, 38, 963–74 .

11.

Lindstrom MJ and Bates DM (1990) Newton–Raphson and EM algorithms for linear mixed–effects models for repeated–measures data . Journal of the American Statistical Association, 83, 1014–22 .

12.

Louis TA (1982) Finding observed information using the EM algorithm . Journal of the Royal Statistical Society, B, 44, 98–130 .

13.

MacKenzie G and Pan JX (2001) On modeling marginal covariance structures in linear mixed model. Proceedings of the 16th International Workshop of Statistical Modelling, Odense, Denmark, July, 275–82.

14.

Meng XL and van Dyk D (1998) Fast EM–type implementations for mixed effects models . Journal of the Royal Statistical Society, B, 60, 559–78 .

15.

Meng XL and Rubin DB (1991) Using EM to obtain asymptotic variance–covariance matrices: the SEM algorithm . Journal of the American Statistical Association, 86, 899–909 .

16.

Pan JX and Fang KT (2002) Growth curve models and statistical diagnostics. New York: Springer–Verlag .

17.

Pan JX and MacKenzie G (2003) On modeling mean–covariance structures in longitudinal studies . Biometrika, 90, 239–44 .

18.

Pan JX and MacKenzie G (2006) Regression models for covariance structures in longitudinal studies . Statistical Modelling, 1, 43–57 .

19.

Pinheiro JC and Bates DM (1996) Unconstrained parameterizations for variance covariance matrices . Statistical Computing 6, 289–96 .

20.

Pourahmadi M (1999) Joint mean–covariance models with applications to longitudinal data: unconstrained parameterisation . Biometrika 86, 677–90 .

21.

Pourahmadi M (2000) Maximum likelihood estimation of generalised linear models for multivariate normal covariance matrix . Biometrika 87, 425–35 .

22.

Pourahmadi M and Daniels MJ (2002b) Dynamic conditionally linear mixed models for longitudinal data . Biometrics 58, 225–31 .

23.

Vu HTV and Knuiman MW (2002) A hybrid ML–EM algorithm for calculation of maximum likelihood estimates in semiparametric shared frailty models . Computational Statistics and Data Analysis, 40, 173–87 .

24.

Wu CFJ (1983) On the convergence properties of the EM algorithm . Annals of Statistics, 11, 95–103 .

25.

Zimmerman DL and Núñez–Antän V (2001) Parametric modelling of growth curve data: an overview . Test, 10, 1–73 .