Statistical theory for the parameters with student’s t-distribution

Consider the location model \bm{Y} = \theta \bm{1}_n + \bm{\epsilon} where \epsilon is distributed n-dimensional Student’s t-distribution. We study the properties of the estimators of \theta and variance \sigma^2_\bm{\epsilon} and compare them with normal theory estimators. In addition, we present some results which help to obtain risks of preliminary test related estimators.

Fulltext