MEASURING LOCATION EFFECTS FROM FACTORIAL-EXPERIMENTS WITH A DIRECTIONAL RESPONSE

Because much of the data motivating previous research on directional statistics comes from observational sources, a number of unresolved issues remain when studying location effects from a multi-way design, In this paper, we review and clarify possible definitions of factor effects on a directional response, and discuss their associations with some existing methods by Watson & Williams (1956), Harrison, Kanji & Gadsden (1986, 1988) and Fisher & Lee (1992), Problems with trying to extend these methods to study multi-way designs are presented, and a new method based on the likelihood ratio test statistic is proposed, Simulation results demonstrate the relative performance of the methods, and an example using industrial data illustrates the technique.