Comment on minority status and family size: a rejoinder to Professor Burch.

The authors replying to Professor Burch believe that he misrepresents the logic of their statistical analysis. Professor Burch refers only to "zero order" differences. Instead Marshall and Jiobu argue that groups can have identical family sizes but for different reasons and the statistical analysis they used permits an assessment of the following: 1) mean levels on such variables as education age at first marriage; 2) differences in the degree to which these variables affect family size; and 3) the effect of minority membership per se. There is also a fourth factor the error term but it has no substantive interpretation. Even if 2 groups have the same family sizes the underlying causal structures explaining the means need not be the same for both groups. For example while the mean difference between Japanese-American and native whites is 0.11 the difference due to minority status alone is 0.99; that is if other forces affecting Japanese-American and native white fertility did not offset the effect of minority status Japanese-American families would be almost 1 person smaller than native white families. The authors submit that this difference is quite substantial theoretically interesting and not at all evident from the analysis that Professor Burch presents in his letter. The significance tests which he computes are not germane to the problem since they are zero-order statistics and cannot be used to tease out underlying causal structures.