A powerful test method for analyzing unreplicated factorials

Abstract

In this paper, a new test method for analyzing unreplicated factorial designs is proposed. The proposed method is illustrated by some examples. An extensive simulation with the standard 16-run designs was carried out to compare the proposed method with three another existing methods. Besides the usual power criterion, another three versions of power, Power I–III, were also used to evaluate the performance of the compared methods. The simulation study shows that the proposed method has higher ability than the remaining three compared methods to identify all active effects without misidentifying any inactive effects as active.