Abstract
Sobel, Hout, and Duncan criticize the linear programming approach to the analysis of mobility tables on the basis of its four alleged deficiencies. The objections of our critics are that (a) we misconceive the object of mobility research by focusing on observed frequencies; (b) our solution neglects the relationship between a sample and a population; (c) within our approach the intercountry differences cannot be assessed; and (d) we arbitrarily define structural mobility. Upon careful consideration these objections are shown to be invalid. First, we argue that the observed frequencies are fundamental and demonstrate that our approach is oriented toward theoretical problems. Second, we provide a method for computing confidence intervals of structural and circulation mobility. Third, we determine the statistical significance of the intercountry difference in the proportion of circulation mobility. Fourth, we explicitly derive our operationalization of structural mobility from one of the traditional notions. We either show that the objections are incompatible with the canons of empirical science or reject them by demonstrating that our critics' assertions are false.
Get full access to this article
View all access options for this article.
