Optimal change-over designs when carry-over effects are proportional to direct effects of treatments

We investigate optimal change-over designs for a model in which treatment carry-over effects are proportional to their direct effects and the proportionality parameter lambda is unknown. A combination of analytical results and computer search is used to identify optimal designs for the estimation of direct effects. Neighbour-balanced designs are shown to have good optimality properties: designs with self-adjacencies perform better for positive lambda, while those without self-adjacencies are better when lambda is negative. The application of the designs to sensory studies is discussed.