Shrunken Estimators of Weibull Shape Parameter in Censored Samples

This paper suggests some shrunken estimators of the shape parameter of the Weibull distribution for failure-censored samples. They are compared with the approximately unbiased estimators derived from Bain & Engelhardt´s estimator. Two estimators are s-efficient if the guess value is near the true value. Of the two 2-stage shrunken estimators discussed here, one is based on a 1-sided preliminary test while the other is based on a 2-sided preliminary test.