ABSTRACT
Some general criteria for evaluating the degree of relationship in order to attribute substantive significance will be offered as illustrative of the approach that must be taken. The substantive significance will be evaluated differently in the case in which only one or two items seem to contribute to the statistical significance from one in which there appear to be differences with respect to many items. In general, smaller differences based on relatively crude observations may justify attribution of substantive significance, while larger differences may be demanded from more refined observations in order to attribute substantive importance. An assessment of the magnitude of the association must be made in some terms other than that of a statistical test of significance. If a series of statistical tests has been performed, it will be instructive to compare in each case the size of differences on the dependent variable between the extremes of the different independent variables.
