Abstract
Some simple criteria for comparing the spatial systems that are used in different applications of doubly constrained interaction models are described and their usefulness is demonstrated by a comparative analysis of four models drawn from British, Dutch, and German sources. The influence that spatial representation has on the performance of the models is also examined and considerable differences are found between the four case studies. These effect the separate interpretability of the endogenously predicted variables and the rate at which the convergence criteria are satisfied.
Get full access to this article
View all access options for this article.
