Abstract
This paper focuses on rank-based (R) estimation of parameters in a linear model with cluster correlated errors. The clusters are assumed to be independent, however, within a cluster the responses are allowed to be dependent. The method is applicable to general within cluster error structure. Application of a model which assumes the within cluster errors which follow an AR(1) process is developed. Discussion of an estimate of the AR(1) parameter is included. The algorithm first estimates the correlation structure by obtaining a robust rank-based estimate of the AR(1) parameter. The responses are then transformed to working independence and the model parameters are fit using ordinary rank regression. Estimates of standard errors—which utilize a sandwich estimate—are provided. An example and simulation results are discussed.
Keywords
This is a preview of subscription content, log in via an institution.
Buying options
Tax calculation will be finalised at checkout
Purchases are for personal use only
Learn about institutional subscriptionsNotes
- 1.
Having bounded influence function and 29 % breakdown point.
- 2.
We do not need the variance to exist. Only that a linear transformation exists which has the goal of reducing the dependence in the response variables. The variance notation is adopted for convenience.
- 3.
As the intercept was fit in the previous step, the residuals should have location zero.
References
Bilgic, Y. K., McKean, J. W., Kloke, J. D., & Abebe, A. (2015, submitted). Iteratively reweighted generalized rank-based methods for hierarchical mixed models.
Dixon, S. L., & McKean, J. W. (1996). Journal of the American Statistical Association, 91, 699 (1996).
Hettmansperger, T. P., & McKean, J. W. (2011). Robust nonparametric statistical methods (2nd ed.). Boca Raton, FL: Chapman & Hall/CRC Press.
Jaeckel, L. A. (1972). Estimating regression coefficients by minimizing the dispersion of residuals, 43, 1449.
Jureǎková, J. (1971). Nonparametric estimate of regression coefficients, 42, 1328.
Kloke, J. D., & McKean, J. W. (2011). In D. R. Hunter, D. P. Richards, & J.L. Rosenberger (Eds.), Nonparametric statistics and mixture models: A festschrift in honor of Thomas P. Hettmansperger (pp. 183–203). Hackensack, NJ: World Scientific Publishing Co. Pte. Ltd.
Kloke, J. D., & McKean, J. W. (2012). The R Journal, 4(2), 57.
Kloke, J. D., McKean, J. W., & Rashid, M. (2009). Journal of the American Statistical Association, 104, 384.
Kloke, J., & McKean, J. W. (2015). Nonparametric statistical methods using R. Boca Raton, FL: Chapman & Hall/CRC Press.
Koul, H. L., Sievers, G. L., & McKean, J. W. (1987). Scandinavian Journal of Statistics, 14, 131.
Pinheiro, J., & Bates, D. (2006). Mixed-effects models in S and S-PLUS. Springer Science & Business Media.
R Development Core Team (2010). R: A Language and Environment for Statistical Computing. Vienna: R Foundation for Statistical Computing. http://www.R-project.org. ISBN:3-900051-07-0.
Rashid, M. M., McKean, J. W., & Kloke, J. D. (2012). Statistics in Biopharmaceutical Research, 4(1), 37.
Rashid, M. M., McKean, J. W., & Kloke, J. D. (2013). Journal of Biopharmaceutical Statistics, 23(6), 1207.
Terpstra, J., McKean, J. W., & Naranjo, J. D. (2000). Statistics, 35, 45.
Terpstra, J., McKean, J. W., & Naranjo, J. D. (2001). Statistics and Probability Letters, 51, 165.
Author information
Authors and Affiliations
Corresponding author
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 2016 Springer International Publishing Switzerland
About this paper
Cite this paper
Kloke, J. (2016). Generalized Rank-Based Estimates for Linear Models with Cluster Correlated Data. In: Liu, R., McKean, J. (eds) Robust Rank-Based and Nonparametric Methods. Springer Proceedings in Mathematics & Statistics, vol 168. Springer, Cham. https://doi.org/10.1007/978-3-319-39065-9_3
Download citation
DOI: https://doi.org/10.1007/978-3-319-39065-9_3
Published:
Publisher Name: Springer, Cham
Print ISBN: 978-3-319-39063-5
Online ISBN: 978-3-319-39065-9
eBook Packages: Mathematics and StatisticsMathematics and Statistics (R0)