On Using an Incorrect Value of ρ in Group Divisible Two-Associate PBIB Designs

Abstract

In the problem of recovery of inter-block information in incomplete block designs, the ratio ρ of the inter-block variance to the intra-block variance which plays a key role is usually unknown. This article studies the loss of information on treatment estimation and treatment contrast estimation due to using some value ρ* in place of ρ. It is seen that for the group divisible two-associate partially balanced incomplete block (PBIB) designs, this loss is generally small.