Recovery of Inter-Row and Inter-Column Information in Two-Way Designs

摘要

The analysis of two-way designs involves (rho sub r) and (rho sub c), the ratios of the inter-row and inter-column variances to the 'intra' variance. These are in general not known and their estimates are substituted. Several estimators with desirable properties are obtained as alternatives to the classical estimates. The effect of their use in the combined inter and intra estimates of treatment effects is studied and the increase in the variance of the estimates of treatment comparisons is obtained. Such designs are useful in practice, as heterogeneity is eliminated in two directions and thereby the precision of treatment estimates is increased. These designs are likely to be of use in the comparison of two or more factors associated with calibrating equipments. (Author)