The Simple Virtues of Descriptive Modeling

Responds to a critique by Ge Lin & Yu Xie (1998) of the authors' application (1997) of loglinear models to the study of interstate mobility & holding power, focusing on the value of purely descriptive models. It is argued that, by separating out economic factors, which are central to total holding power, Lin & Xie pursue a conventional causal approach rather than the descriptive one used in the original analysis, & assume the applicability of causal models to descriptive problems. This is seen in their discussion of the problems arising from using overlapping persistence parameters & their proposed models that include higher-order contiguity, which are not applicable to the generalization of contiguity effects in the original analysis. Similar problems are found in their hypotheses regarding the ordering of contiguity effects, discussion of dispersistence effects, & potential explanatory models for the influence of economic push & pull factors on migration. It is concluded that the rules of causal model building cannot be so directly applied to answering sociological questions descriptively, a distinction that Lin & Xie overlook. 1 Table, 1 Figure, 3 References. T. Arnold