This research introduces an alternative conceptual framework for the rational-economic spatial interaction model (SIM) that geographers and others have worked with for almost thirty years. SIMs use numerical interpolation to 'calibrate' a negative-exponential parameter on rational cost (friction of distance). The parameter allows researchers to infer characteristics of the population. An alternative interpretation is called the bandwagon effect (BWE) because it seeks an analogy with the tendency of crowds to attract more members. The BWE assumes that the raw quantity of information about a place supersedes rational cost in the mind of a spatial decision-maker (SDM). A surrogate for all possible information emanating from a place is the rate of out-migration observed there in units of the out-migration at the SDM's origin. The SIM is respecified with information in place of rational cost, and interpreted to mean that places that are losing fewer residents are more attractive. Two interregional migration datasets (Canada 1976-1981, and U.S.A. 1965-1970) are used to compare the methods. Information is found to be uncorrelated with distance in both data sets. The BWE is observed to substantially exceed the performance of the SIM with Canadian data, and to match it with U.S. data.

In this concise, clear introduction, the authors describe the theory of spatial diffusion, its method of measurement and many of its applications. The seminal work of Torsten Hagerstrand, who introduced the original spatial model of diffusion, is outlined. The authors then summarise the developments that have been made to Hagerstrand's formulation, and make suggestions for future research.

Gravity models describe, and hence help predict, spatial flows of commuters, air-travelers, migrants, commodities and even messages. They are one of the oldest and most widely used of all social science models. This book presents an up-to-date, consistent and unified approach to the theory, methods and application of the gravity model - which spans from the axiomatic foundations of such models all the way to practical hints for their use. "I have found no better general method for use in applied research dealing with spatial interaction... It is against this background that the present book by Sen and Smith is most welcomed." Walter Isard

In this book, the author's strong commitment to the multi-disciplinary field of regional science emerges to provide a unifying framework between spatial modelling traditions from quantitative geography and those from spatial economics, whereby each is enhanced. Starting with a detailed discussion of each field illustrated with numerical examples, the two traditions are brought together by either making the economic models probabilistic or transforming the objectives of the geographic models to reflect both utility theory and production theory. The ideas are applied to develop urban models of activity analysis, face-to-face contacts and housing supply, as well as regional models in the areas of input-output analysis, imperfect competition and interregional migration.

This book has grown out of a desire to explore the possibilities of using optimizing models in transportation planning. This approach has been followed throughout. Models which combine descriptive and optimizing elements are not treated. The gravity model is here studied as the solution to an optimizing model. In spite of this approach, much of the material shoula be of general interest. Algorithms are not discussed. The author has benefited from discussions with many colleagues. M. Florian suggested the term "interacti vi ty". N. F. Stewart and P. Smeds gave many valu able comments on a first draft. M. Beckmann made me think once more about the final chapters. R. Grubbstrem and K. Jornsten helped clarifYing some things in the same chapters. Remaining insufficiencies are due to the author. Gun Mannervik typed with great patience. Linkoping in October 1979 Sven Erlander ABSTRACT The book proposes extended use of optimizing models in transportation plann ing. An entropy constrained linear program for the trip distribution problem is formulated and shown to have the ordinarJ doubly constrained gravity model as its solution. Entropy is here used as a measure of interactivity, which is constrained to be at a prescribed level. In this way the variation present in the reference trip matrix is preserved. (The properties of entropy as a dispersion measure are shortly discussed. ) The detailed mathematics of the optimal solutions as well as of sensitivity and duality are given.