Addressing methodological assumptions of correspondence tests when measuring discrimination

Abstract

In this study, we introduce a critical assessment of methodological assumptions made by researchers when using correspondence tests to measure ethnic discrimination. We aim to investigate whether (1) the order in which the applications are sent out, between the test and control person, (2) the conducted matched triad tests, in comparison to matched duo tests, and(3) the timing of application within the week, have an effect on discrimination rates. For this purpose, we made use of 2.984 matched correspondence tests across Flanders, the Dutchspeaking part of Belgium. Our findings suggest that the correspondence tests are rather robust for the variability in methodology. While these different methodological choices do affect the general invitation rates, they do not affect discrimination rates.