ABSTRACT
This chapter presents the empirical test employing a statistical framework which will permit conclusions about family, community and interaction effects. Logistic regression is used to estimate the parameters of the multi-level model of migration, because the dependent variable is dichotomous. The interactive contextual model of migration calls for a curvilinear relation between migration and family class, due to the nonlinear pattern of risk-aversion found among families ranked by socio-economic class. This nonlinear pattern was documented in the bivariate comparison of families with and without migrants. The sharply contrasting patterns of migration associated with different levels of social infrastructure. The relation between class and migration is affected by the size of the community’s migrant reference group, and the relation between development level and migration is affected by community accessibility.
