Quick Analysis Methods for Random Balance, Screening Experiments

A short expository account of random balance ia given, in which some different types of sampling are distinguished. As a quick significance test of effectiveness of single factors, a simple analysis of variance method is recommended. For the sake of sensitivity, it is suggested that the number of levels of quantitative factors should preferably be less than five. The degree of unbalance of a random balance design is studied, largely through an example, and a desirable upper bound is suggested for the number of levels of any factor, namely one eighth of the total number of observations.