## Robustness of the linear mixed model to misspecified error distribution.(English)Zbl 1162.62319

Summary: A simulation study is performed to investigate the robustness of the maximum likelihood estimator of fixed effects from a linear mixed model when the error distribution is misspecified. Inference for the fixed effects under the assumption of independent normally distributed errors with constant variance is shown to be robust when the errors are either non-Gaussian or heteroscedastic, except when the error variance depends on a covariate included in the model with interaction with time. Inference is impaired when the errors are correlated. In the latter case, the model including a random slope in addition to the random intercept is more robust than the random intercept model. The use of Cholesky residuals and conditional residuals to evaluate the fit of a linear mixed model is also discussed.

### MSC:

 62F10 Point estimation 65C60 Computational problems in statistics (MSC2010) 62F35 Robustness and adaptive procedures (parametric inference) 62J05 Linear regression; mixed models

AS 181
Full Text:

### References:

 [1] Butler, S.M.; Louis, T.A., Random effects models with non parametric priors, Statist. med., 11, 1981-2000, (1992) [2] Hoaglin, D.C., Summarizing shape numerically: the $$g$$-and-$$h$$ distribution, (), 461-513 [3] Jacqmin-Gadda, H.; Fabrigoule, C.; Commenges, D.; Dartigues, J.F., A five-year longitudinal study of mini-mental state examination in normal aging, Amer. J. epidemiol., 145, 498-506, (1997) [4] James, F., A review of random number generators, Comp. phys. comm., 60, 329-344, (1990) · Zbl 0875.65021 [5] Laird, N.; Ware, J., Random-effects models for longitudinal data, Biometrics, 38, 963-974, (1982) · Zbl 0512.62107 [6] Lange, N.; Laird, N., The effect of covariance structure on variance estimation in balanced growth-curve models with random parameters, J. amer. statist. assoc., 84, 241-247, (1989) · Zbl 0678.62070 [7] Liang, K.; Zeger, S., Longitudinal data analysis using generalized linear models, Biometrika, 73, 13-22, (1986) · Zbl 0595.62110 [8] Marsaglia, G.; Tsang, W.W., A fast easily implemented method for sampling from decreasing or symmetric unimodal density functions, SIAM J. sci. statist. comput., 5, 349-359, (1984) · Zbl 0573.65116 [9] Mcneil, A.J.; Gore, S.M., Statistical analysis of zidovudine (AZT) effect on CD4 cell counts in HIV disease, Statist. med., 15, 1, 75-92, (1996) [10] Molina, J.M.; Chêne, G.; Ferchal, F.; Journot, V.; Pellegrin, I.; Sombardier, M.N., The ALBI trial: a randomized controlled trail comparing stavudine plus didanosine with zidovudine plus lamivudine and a regimen alternating both combination in previously untreated patients infected with human immunodeficiency virus, J. infect. dis., 180, 351-358, (1999) [11] Park, T.; Lee, S.Y., Model diagnostic plots for repeated measures data, Biometrical J., 46, 441-452, (2004) [12] Richardson, A.M.; Welsh, A.H., Robust restricted maximum likelihood in mixed linear models, Biometrics, 51, 1429-1439, (1995) · Zbl 0875.62313 [13] Royall, R.M., Model robust confidence intervals using maximum likelihood estimators, Internat. statist. rev., 2, 221-226, (1986) · Zbl 0596.62032 [14] Royston, P., A pocket-calculator algorithm for the shapiro – francia test for non-normality: an application to medicine, Statist. med., 12, 181-184, (1993) [15] Taylor, J.M.G.; Law, N., Does the covariance structure matter in longitudinal modelling for the prediction of future CD4 counts?, Statist. med., 17, 2381-2394, (1998) [16] Taylor, J.M.G.; Cumberland, W.G.; Sy, J.P., A stochastic model for analysis of longitudinal AIDS data, J. amer. statist. assoc., 89, 727-736, (1994) · Zbl 0806.92011 [17] Thiebaut, R., Chene, G., Jacqmin-Gadda, H., Morlat, P., Mercie, P., Dupon, M., Neau, D., Ramaroson, H., Dabis, F., Salamon, R., Groupe d’Epidemiologie Clinique du SIDA en Aquitaine, 2003. Time-updated $$\operatorname{CD} 4 + \operatorname{T}$$ lymphocyte count and HIV RNA as major markers of disease progression in naive HIV-1-infected patients treated with a highly active antiretroviral therapy: the Aquitaine cohort, 1996-2001. J. Acquir. Immune Deficiency Syndrome 33(3) 380-386. [18] Tsiatis, A.A.; Davidian, M., Joint modeling of longitudinal and time-to-event data: an overview, Statist. sinica, 14, 809-834, (2004) · Zbl 1073.62087 [19] Tsiatis, A.A.; DeGruttola, V.; Wulfsohn, M.S., Modelling the relationship of survival to longitudinal data measured with error, Application to survival and CD4 counts in patients with AIDS. J. amer. statist. assoc., 90, 27-37, (1995) · Zbl 0818.62102 [20] Verbeke, G.; Lesaffre, E., The effect of misspecifying the random effects distribution in linear mixed models for longitudinal data, Comput. statist. data anal., 23, 541-556, (1997) · Zbl 0900.62374 [21] Zhang, D.; Davidian, M., Linear mixed models with flexible distributions of random effects for longitudinal data, Biometrics, 57, 795-802, (2001) · Zbl 1209.62087
This reference list is based on information provided by the publisher or from digital mathematics libraries. Its items are heuristically matched to zbMATH identifiers and may contain data conversion errors. It attempts to reflect the references listed in the original paper as accurately as possible without claiming the completeness or perfect precision of the matching.