None of these studies is based on a random sample; hence, application of the probability calculus to an analysis of the sample data and statements of statistical significance is again only justified in the same sense as was discussed for the sample data used in the regression studies.