Abstract
The conclusion of Smith and Clayton (1978) that United States migration streams contain ‘significant’ numbers of intransitivities—thus undermining the applicability to migration of spatial-interaction models—is contraindicated. Fewer intransitivities than could be expected under a null hypothesis of randomness are found in the solution of a linear assignment problem. The solution consists of a single entry from each row and column of a generalized tournament model of 1965–1970 interstate migration. Three cycles are formed by the entries. Three is less than 4–519, the amount expected in a random assignment. Under the constraint that the entries (fij) selected are greater than 0·5, indicative that more people move from i to j than from j to i, five intransitivities—still an insignificant number—are found. However, a highly significant number, fourteen, of intransitivities are found if the migration flows themselves—not a generalized tournament model of them—are studied.
Get full access to this article
View all access options for this article.