The Regression Model of Regionalism: A Critique

Abstract

No matter how one interprets Professor Irvine's model (X-form or Z-form), its validity is dubious. This derives not only from his misuse of regression analysis but also from his tendency to generalize beyond his data and his model. For instance, I cannot see how he deduces from his model that “regionalism can occur in three guises: (1) persistence of historical effects but response to a common stimulus in different regions; (2) different degrees of response to the same predominant stimulus or cleavage in each region; (3) response to quite different stimuli in each region” (p. 24). What are the stimuli? What is the “predominant” one? What are the “responses”? What does he mean by “degrees of response”? From one dependent variable (“bilingualism”) and three independent variables (integration, income, population) he comes up with a totally different set of S-R variables at the end of his analysis. With some reservation, I can see only one stimulus – bilingualism – which is the dependent variable arbitrarily chosen to study regional differences.