Long-term multivariate prediction of migration patterns

In- and out-migration rates of 56 Israeli cities with a population of over 5000 were predicted for a five-year period (1977–1981), on the basis of four 1976 social indicators: crime rate, percent unemployment, population size and distance from a major metropolitan center. Rather than employ independent correlation coefficients, the four indicators were entered as predictors into regression equations with in- and out-migration rates serving as the dependent variable. The analytic methods are compared and the clear advantages of the regression method emerge. None of the indicators reliably predicted in-migration. Crime consistently predicted out-migration. Distance and unemployment each entered into three of the prediction models.