Abstract
In a situation where the given set of parameters (b, k andv) precludes the existence of any known optimal block designs, but an optimal block design is known to exist with parametersb, k andv*>v, a new design is shown to be useful. This (b, k, v) design is obtained from the (b, k, v*) optimal design by collapsing the classes of a suitable paritition of the treatment set (of the latter design) to treatments (of the former). We call the new design a quotient of the original design. Although the quotient is non binary and unequally replicated, it turns out to beE-optimal within the class of all proper and connected designs withb, k andv, provided the replication number of the optimal design we start with is not too large.
Similar content being viewed by others
References
Constantine GM (1981) SomeE-optimal block designs. Ann Statist 9:866
Jacroux M (1980) On the determination and construction ofE-optimal block designs with unequal number of replicates. Biometrika 67:661
Jacroux M (1982) On theE-optimality of block designs for the one-way and two-way elimination of heterogeneity. J Roy Statist Soc B 44:253
Jacroux M (1984) SomeE-optimal block designs. Sankhya B 45:351–361
Sinha Bikas K, Sinha Bimal K (1969) Comparisons of relative efficiencies of some classes of designs. Cal Stat Assoc Bull 18:99
Takeuchi K (1961) On the optimality of certain type of PBIB designs. Rep Stat Appl Res Un Japan Sei Engrs 8(3):140
Author information
Authors and Affiliations
Rights and permissions
About this article
Cite this article
Bagchi, S. A class of non binary unequally replicatedE-optimal designs. Metrika 35, 1–12 (1988). https://doi.org/10.1007/BF02613280
Received:
Issue Date:
DOI: https://doi.org/10.1007/BF02613280