Simulation of educational trajectories, taking latent decision-making processes into account - in the framework of the research group “Multi-sectoral Regional Microsimulation Model” (MikroSim)

This project will create a methodological framework for embedding a transition model with free (unknown) parameters in a dynamic, small-scale microsimulation. A transition model of this kind will also enable latent and/or only partially observed processes to be represented in a microsimulation. The key element for embedding a transition model with free parameters is a metamodel (or statistical emulator) for the dynamic, small-scale microsimulation model.

As an example application, the model will be used to simulate educational decisions at the threshold between elementary and secondary school. Here, the focus will be on educational inequalities between Germany’s autochthonous and migrant population that are already evident at this early life course transition.

One of the main goals of the project is to find out which metamodeling approach is best suited to application in a dynamic microsimulation (with a view to the “elementary school - secondary school” transition model). Another goal is to develop efficient methods for calibrating a simulation model and validation strategies for testing the usefulness of the transition model. In addition, as part of the metamodeling, the project will create a methodological toolbox (in R) that implements the approaches to statistical emulation and calibration that are best suited to application in the microsimulation data structure of the MikroSim project.