Examining high-stakes language tests for migrants: effectiveness, validity and bias
Over the past decade it has become standard practice in European countries to use language tests as part of migration and integration policy. The fact that the empirical evidence supporting the use of language tests as a proxy for integration, is rather thin has not stopped these policies from gaining traction worldwide. This research project holds that policy against the light in terms of validity, effectiveness and fairness. In doing so, we will examine (1) the policy makersU+2019 rationale for relying on language testing policies in migration and integration policies, and (2) the fairness of the test-based policy, from the perspective of low-educated adults. The first part of the project charts the dynamics that shape migration policy processes. Relying on Q methodology we will (1a) uncover which stakeholders and dynamics influence how migration policy takes shape in Flanders and in Europe. Using policy network analysis, we will also (1b) determine the impact of societal, political and economic actors on policy-making, and how are their interests are weighed against each other. The second part of the study focuses on the impact of educational background on test performance. Using Many-Facet Rasch analysis to detect differential item functioning, we will (2a) examine to what extent a candidateU+2019s educational background and literacy level impact the performance on the Flemish language test for immigration and citizenship purposes. Relying on eye tracking we will investigate (2b) to what extent adult migrants with diverging levels of educational background interact differently with a computer- or paper-based language test.