Department of Marketing , University of Cincinnati , Cincinnati, OH, 45221
Abstract:
A sample size selection procedure for paired comparisons of means is presented which controls the half width of the confidence intervals while allowing for unequal variances of treatment means.