Summary
In longitudinal data analysis, as with cross-sectional regression, correction for heteroscedasticity is important for accurate hypothesis tests and confidence intervals. Using empirical standard errors to adjust for heteroscedasticity is common practice. In this article, we investigate alternative empirical standard errors that have better finite sample properties than those commonly used. We review the properties of an estimator that is based on the deletion, or jackknife, principle. We also consider an estimator based on standardized residuals that is unbiased in many circumstances. Our Monte Carlo simulations show that both alternatives have better small-sample properties than the usual empirical standard errors, regardless as to whether heteroscedasticity is present. Both alternatives behave well in terms of size and power even when the covariance matrix is completely misspecified. We recommend the use of the alternative empirical standard errors whenever there is a reason to suspect that heteroscedasticity is present.
Similar content being viewed by others
References
Bannerjee, M. and E.W. Frees (1997). Influence diagnostics for longitudinal models. Journal of the American Statistical Association 92, 999–1005.
Carroll, R. J. and Ruppert, D. (1988). Transformation and weighting in regression, Chapman-Hall.
Eicker, F. (1967). Limit theorems for regressions with unequal and dependent errors. Proceedings of the Fifth Berkeley Symposium on Mathematical Statistics and Probability 1, LeCam, L. M. and Neyman, J. editors, University of California Press, pp, 59–82.
Feng, Z., McLerran, D. and J. Grizzle (1996). A comparison of statistical methods for clustered data analysis with Gaussian error. Statistics in Medicine 15, 1793–1806.
Gourieroux, C., Monfort, A., and Trognon, A. (1984). Pseudo-maximum likelihood methods: theory. Econometrica, 52, 681–700.
Huber, P. J. (1967). The behaviour of maximum likelihood estimators under non-standard conditions. Proceedings of the Fifth Berkeley Symposium on Mathematical Statistics and Probability 1, LeCam, L.M. and Neyman, J. editors, University of California Press, pp, 221–33.
Kauermann, G. and R. J. Carroll (2001). A note on the efficiency of sandwich covariance matrix estimation. Journal of the American Statistical Association 96, 1387–1396.
Liang, K.-Y. and S.L. Zeger (1986). Longitudinal data analysis using generalized linear models. Biometrika 73, 12–22.
Long, J.S. and L.H. Ervin (2000). Using heteroscedasticity consistent standard errors in the linear regression model. American Statistician 54, 217–224.
MacKinnon, J.G. and H. White (1985). Some heteroskedasticity consistent covariance matrix estimators with improved finite sample properties. Journal of Econometrics 29, 53–57.
Manel, L.A. and T.A. DeRouen (2001). A covariance estimator for GEE with improved small-sample properties. Biometrics 57, 126–134.
Pan, W. (2001). On the robust variance estimator in generalized estimating equations. Biometrica 88, 901–906.
White, H. (1980). A heteroskedasticity-consistent covariance matrix estimator and a direct test for heteroskedasticity. Econometrica 48, 817–38.
Acknowledgements
We thank two anonymous reviewers for comments on the paper. The Fortis Health Insurance Professorship and the National Science Foundation Grant Number SES- 0095343 provided funding to support the research of Edward Frees.
Author information
Authors and Affiliations
Appendix
Appendix
From a standard result on the inverse of partitioned matrices, we have
With this and the definition of the delete i WLS estimator, we have
This establishes equation (4).
Rights and permissions
About this article
Cite this article
Frees, E.W., Jin, C. Empirical Standard Errors for Longitudinal Data Mixed Linear Models. CompStat 19, 455–475 (2004). https://doi.org/10.1007/BF03372107
Published:
Issue Date:
DOI: https://doi.org/10.1007/BF03372107