关键词: error, generalized mean square error, admissibility, Bayes linear estimator

Abstract:  In response to some perceived deficiency with the BLUE of the regression parameter matrix in the growth curve model,this paper proposes a class of linear biased estimators,which is called a class of universal ridge ones(URE),based on the recent development theory of biased estimation on the classical regression model.By appropriate choices of the included biased parameters, many interesting and practical linear biased estimators maybe obtained.Some important properties are also discussed.With the UR estimators,the estimation methods used in the growth curve model will be greatly enriched.

Key words: error, generalized mean square error, admissibility, Bayes linear estimator