In a footnote, the authors of this paper acknowledge the assistance of William B. Ware, PhD, of the School of Education of Northwestern University, who performed the statistical evaluation of their data. A question by a reviewer of the original manuscript led to an exchange of letters with the senior author. Because a problem was involved which in the past has been and in the future will be encountered by many others in the statistical treatment of observational data and because Mr. Ware was helpfully lucid in clarifying the problem, a comment by the editor is justified. After perusing the manuscript and the query of the reviewer, your editor wrote to the senior author as follows: "You speak of using the analysis of variance as modified by Winer. Does this mean that unaccounted-for random variability (variance) was determined as the difference between total variance and the sum of the variances