Categorical, Multivariate Analysis of Intra-urban Migration

Abstract

A two-step paradigm is employed to analyse intra-urban migration patterns within Israeli towns. In step 1 of the paradigm, census-derived predictor macro-variables descriptive of towns are classified according to the intersection of two nominal dimensions. The three levels of dimension 1 correspond to the three factors emergent from a factor analysis of the predictor variables. These are labelled as representing either economic level, demographic composition or social features of the towns. Dimension 2 was a priori defined as being dichotomous, as it represented each variable's enhancing (a positive β-coefficient) or inhibiting (a negative β-coefficient) effect on intra-urban migration rates. Six cells result from the intersection of the two dimensions. Multivariate regression analysis is carried out within each level of the two dimensions. Regression analysis within the inhibitor level of dimension 1 reveals that four variables inhibited urban residential change: home-ownership, relatively high income, religiosity level of the town and median age of the town's population. These variables are distributed among the three levels of dimension 2. These results partially parallel findings concerning inmigration to Israeli towns that economic variables are crucial to locational decisions. However, with respect to intra-urban moves, economic considerations are less salient. Implications concerning basic urban policy decisions are discussed.