Abstract: | "Asymmetric square tables, such as those arising from interregional migration, can be analysed by separating the skew-symmetric and symmetric components. A least-squares analysis of the skew-symmetric part can indicate the degree of complexity of model that is consistent with data and this can be combined with some suitable model for the symmetric part. The joint model may then be fitted by maximum likelihood based on suitable distributional assumptions. This approach is used for an analysis of Australian interstate migration for l960-l966 and indicates a model with independent in-migration and out-migration rates proportional to a symmetric function of population sizes and interstate distance." |