MSE dominance of the pre-test iterative variance estimator over the iterative variance estimator in regression

In this paper, we examine the small sample properties of the pre-test iterative variance estimator in regression. The explicit formula of MSE is derived, and it is shown that the pre-test iterative variance estimator with an appropriate critical value dominates the iterative variance estimator without pre-testing in terms of MSE. We also compare the MSE performances of the pre-test iterative variance estimators using the Stein-rule, minimum mean squared error, and adjusted minimum mean squared error estimators by numerical evaluations. (C) 2001 Elsevier Science B.V. All rights reserved.