Μέθοδοι ανάλυσης παραγοντικών σχεδιασμών χωρίς επαναλαμβανόμενες παρατηρήσεις (Unreplicated factorial designs)
Methods for analysing unreplicated factorial designs
In certain cases the time or money required to obtain the output response of an experiment makes the replication of the experiment prohibitive. As a result there is no independent estimate of error which makes the judgment of the significance of effects impossible. The purpose of this dissertation is to develop quantitative methods in order to successfully analyse unreplicated factorial designs as well as the application of these methods on real or simulated data. The first chapter is an introduction to factorial designs at two levels as well as a description of the problem that arises when trying to analyse unreplicated designs. The second chapter is an overview of five popular quantitative methods proposed for the previous designs. A simulation study to compare these methods is described in the final chapter.