Appropriate Sample Sizes for Selecting a Population With the Largest Correlation Coefficient from Among k Bivariate Normal Populations

Abstract

Bechhofer's (1954) procedure for selecting the population with the largest mean from among k normal populations with known variances is discussed as it applies to the problem of selecting the bivariate normal population with the largest correlation coefficient. Tables of appropriate sample sizes are provided and an example is discussed concerning their use in which one might wish to select the correct population at a specified probability level.