Abstract
Robustness aspects of block designs for diallel crosses against missing data have been investigated using connectedness and efficiency criteria. Sufficient condition for robustness of a design has been obtained for the loss of any q observations, using connectedness criterion. Designs robust against the loss of any two observations, loss of any q observations in a block and loss of any two blocks have been identified. Efficiencies of the residual designs have been evaluated for certain designs for diallel crosses for several pattern of loss of two observations. The efficiencies of the residual designs have also been calculated when any q observations in a block and any two blocks (not necessarily disjoint) are lost.
Acknowledgments
Authors are grateful to the referees for their constructive criticisms and suggestions that helped in improving the presentation of the results.