As yet there is no comprehensive migration theory available for the analysis of the complexity of migratory processes. Thus, middle-range theories ought to integrate insights from various social sciences into a unifying analytical framework. This framework needs to go beyond the individual and take into account the political economy of migration in order to analytically connect micro level factors, with political, economic and social factors at various levels. We suggest the theory of New Institutionalism as a middle-range theory for the investigation of migration decision making. New Institutionalism connects different levels of analysis, and the theory is based at the junction of political science, sociology and economics. It incorporates the structure-agency dimension while describing individual behavior in structures. Thus, the complex web of factors influencing migration decisions can be examined within a single coherent analytical framework. Methodologically, this calls for an analytical instrument to model factors at different levels simultaneously. Thus, Bayesian Networks is used as an innovative instrument to capture the systemic complexity of migration decision making. In doing so, the depth and breadth of cross-linkages existing within the migration decision making process are revealed. Moreover, it potentially gives new insights into how migration processes can be influenced. Regionally, the proposal focuses on Kazakhstan. Kazakhstan is the economic motor in central Asia and displays fascinating migration patterns (i.e. immigration, emigration, return-migration and internal migration), which have been partly politically motivated. Finally, this proposal centers on internal migration movements, which have been neglected in favor of international migration research during the last three decades and which are also an under-researched area in Kazakhstan. By applying a new analytical framework and methodology in a migratory-dynamic region we are advancing basic migration research at a theoretical, methodological and empirical level.