The gravity model has proven remarkably accurate, even with its overly simple assumptions. That a simple model can be constructed of people's spatial distribution is not surprising; that the model has proven so robust and been successfully applied in so many settings is noteworthy.
The model used in the spreadsheet is a simplified constrained gravity model. It does not explicitly calculate exponents for the measures of "population" or "opportunity." The constant, K, is not explicitly entered. Instead, all of those adjustments are combined in the adjustment matrix. What the spreadsheet gains in ease of use it loses is specificity and sophistication. The experienced user can, of course, modify the spreadsheet to include the exponents as fixed values in the analysis.
A final warning about using the constrained gravity model: The system treats each site and each neighborhood as discrete, self-contained units. They are not. If the level of measurement is fairly crude, this source of error will not be significant. However, as the model is refined more and more, as the matrices are adjusted to more and more closely model actual events, then the user will find a nagging, elusive error running through the results. That error is unavoidable; it is due to the inherent inability to completely specify either the neighborhood or the site.
© 1996 A.J.Filipovitch
Revised 11 March 2005