Multiple comparisons with and selection of the best treatment in (incomplete) block designs

Publisher Website
Google Scholar
Abstract
Suppose we want to select the (ubknown) best treatment of t treatments by means of an experiment. In order to minimize the experimental error, many such experiments involve the use of blocks and we have to take that into consideration.In the framework a general block design we will give a procedure that selects a subset of the treatments, such that the probability the unknown best treatment is included, is at least a prespecified value P *.Simulantaneously confidence bounds of the distance of each treatment with that same confidence level P * are derived. Also a confidence lower bound is given for the probability of correct selection, i.e. the probability that the best treatment is selected.An example illustrates all these concepts.Many results can be generalized to fixed-effects linear models.

This publication has 13 references indexed in Scilit: