Triallel-cross block designs that are robust: Against missing observations

Abstract

The robustness of triallel-cross block designs against missing data is investigated using connectedness and efficiency criteria. A sufficient condition for robustness of such a design is obtained for the loss of any m observations in a block. Designs robust against the loss of any m observations in a block are identified. For several patterns of loss of two observations the connectedness and efficiencies of the residual designs are evaluated for certain triallel-cross designs.