Modelling Migration: Decisions, Processes and Outcomes

Jakub Bijak , University of Southampton
Philip Higham, University of Southampton
Jason Hilton, University of Southampton
Martin Hinsch, University of Southampton
Sarah Nurse, University of Southampton
Toby Prike, University of Southampton
Oliver Reinhardt, University of Rostock
Peter W. F. Smith, University of Southampton
Adelinde Uhrmacher, University of Rostock

Human migration is uncertain and complex, and some of its distinct features, such as migration routes or responses of flows to the underlying determinants, can emerge in a very short time. Agency of various actors is one of the key drivers of this complexity and a reason why migration typically eludes attempts at its theoretical description, explanation and prediction, which typically remain scattered across various disciplines. To address migration challenges, what is needed is a unified, interdisciplinary approach for describing micro-level migration decisions and macro-level processes in a coherent way. In this talk, we will present four building blocks of such an approach: construction of an agent-based simulation model of migration route formation; formal quality evaluation of the data used to inform the modelling process; psychological experiments aimed at eliciting decisions and preferences under uncertainty; and the choice of an appropriate programming language and modelling formalisms. These four elements are brought together in a unified analysis of the modelled migration processes and their outcomes. Here, we apply the methods of Bayesian meta-modelling based on the Gaussian Process assumptions and utilising the principles of statistical experimental design. In particular, the analysis covers the uncertainty of model outputs, sensitivity to a range of input parameters, as well as calibration of the model to the extent allowed by the available data. By framing the analysis in such formalised way, we hope to strengthen the case for model-based demography, intended to fill the widely acknowledged theoretical gaps in population studies.

See extended abstract

 Presented in Session 33. Methods for Migration Research