Tests of statistical hypotheses in the case of unequal or disproportionate numbers of observations in the subclasses