Assessing the goodness-of-fit of the laird and ware model : An example : The jimma infant survival differential longitudinal study

The Jimma Infant Survival Differential Longitudinal Study is an Ethiopian study, set up to establish risk factors affecting infant survival and to investigate socio-economic, maternal and infant-rearing factors that contribute most to the child's early survival. Here, a subgroup of about 1500 children born in Jimma town is examined for their first year's weight gain. Of special interest is the impact of certain cultural practices like uvulectomy, milk teeth extraction and butter swallowing, on child's weight gain; these have never been thoroughly investigated in any study. In this context, the linear mixed model (Laird and Ware) is employed. The purpose of this paper is to illustrate the practical issues when constructing the longitudinal model. Recently developed diagnostics will be used herefor. Finally, special attention will be paid to the two-stage interpretation of the linear mixed model.

[1]  E Lesaffre,et al.  Local influence in linear mixed models. , 1998, Biometrics.

[2]  James H. Ware,et al.  A simulation study of estimators for rates of change in longitudinal studies with attrition. , 1991, Statistics in medicine.

[3]  Christopher H. Morrell,et al.  Linear Transformations of Linear Mixed-Effects Models , 1997 .

[4]  H. Weisberg,et al.  Empirical Bayes estimation of individual growth-curve parameters and their relationship to covariates. , 1983, Biometrics.

[5]  R. Einterz,et al.  Traditional uvulectomy in northern Cameroon , 1994, The Lancet.

[6]  G. Verbeke,et al.  The effect of misspecifying the random-effects distribution in linear mixed models for longitudinal data , 1997 .

[7]  An Ethiopian birth cohort study. , 1996, Paediatric and perinatal epidemiology.

[8]  P. Diggle.  An approach to the analysis of repeated measurements. , 1988, Biometrics.

[9]  J. Ware,et al.  Random-effects models for longitudinal data. , 1982, Biometrics.

[10]  D. Rubin,et al.  Statistical Analysis with Missing Data. , 1989 .

[11]  H Goldstein,et al.  Multilevel time series models with applications to repeated measures data. , 1994, Statistics in medicine.

[12]  G. Reinsel,et al.  Models for Longitudinal Data with Random Effects and AR(1) Errors , 1989 .

[13]  E Lesaffre,et al.  The detection of residual serial correlation in linear mixed models. , 1998, Statistics in medicine.

[14]  R. Cook.  Assessment of Local Influence , 1986 .

[15]  P. Royston,et al.  Regression using fractional polynomials of continuous covariates: parsimonious parametric modelling. , 1994 .

[16]  Yuedong Wang.  Smoothing Spline Models with Correlated Random Errors , 1998 .