Journal of Statistics: Advances in Theory and ApplicationsAuthor's: Kenichi Satoh, Hirokazu Yanagihara and Ken-Ichi Kamo
Pages: [113] - [124]
Received Date: April 21, 2010
Submitted by: 113
Repeated measurements at several time points can be described by a growth curve model. However, there might exist a few individual curves that are located far away from the mean trend or have large variances. These individual curves can be candidates for outliers and robust estimators of the regression coefficients are desired. In this paper, an estimation method is proposed that uses weighted least square estimators, where the weights are optimized from squared residuals following the gamma distribution.
growth curve model, longitudinal data, repeated measurements, robust estimation.
