Sankhya: The Indian Journal of Statistics

1993, Volume 55, Series B, Pt. 2, 249--261

OPTIMAL CROSS-OVER EXPERIMENTAL DESIGNS

By

D. COLLOMBIER,

And

I. MERCHERMEK, University de Pau

SUMMARY. This paper deals with multiperiod cross-over trials. We consider experiments where carry-over effects are suspected to persist not only for one, but also for two or more periods after treatment application. Conditions of adjusted orthogonality of a design are defined. Within large classes of cross-over plans with or without preperiod, designs fulfilling these conditions are proved to be universally optimal for the estimation of direct effects and/or of residual effects. Thus we extend some results of kunert (1984). Strongly (or completely) balanced designs introduced by Sharma (1977) or by Cheng and Wu (1979) are examples of such optimal plans when first and second residual effects are considered. Other examples of such optimal cross-over designs are given.

AMS subject classification. 62K05

Key words and phrases. Cross-over designs, change-over designs, carry-over effects, universal optimality, strongly balanced cross-over designs

FULL PAPER.

This article in Mathematical Reviews.