ABSTRACT
In all cases in which such a discrepancy between the two samples exists, we have at least noted the fact. Furthermore, in any subsequent analysis we have kept the two samples separate. Not infrequently, when we have done this some further trend still shows up quite clearly in both samples, and we can then give a final overall estimate of its significance. Thus in the case of the use of dummies there are consistent social class differences found in both the samples, and we therefore feel justified in going on to discuss these, even though the two samples differed initially.
