ABSTRACT
The inadmissibility of the MLE was established by showing that the JS had a uniformly smaller risk. The JS is also inadmissible because there are a number of different estimators obtained by truncating the JS that have uniformly smaller Bayes and frequentist risk. The methods of truncation consist of altering the form of the contraction factors when they are negative. The different ways of doing this produce the different positive part estimators. These positive part estimators are the subject of this chapter. Two cases will be considered:
estimating the mean of a multivariate normal distribution;
the single linear model.
