Multilevel modelling of binary and ordinal dependent variables : exploring the effect of education on tolerance

Abstract

While the positive relationship between education and tolerance is well documented in the developed West, evidence from countries with different geographical, political, socio-economic, and cultural background is more limited. Moreover, based on preliminary empirical evidence, I expected that the liberalizing effect of education may not be as universal as often assumed and may vary across national contexts. Using an analysis of variations in aggregate levels of tolerance as well as in the educational effect as a practical example, this case study shows how to apply multilevel modelling techniques using multilevel logistic and multilevel ordinal regression models. At the micro-level, data reported in the analysis are derived from the World Values Survey third wave (1995–1998) across 24 countries. At the macro-level, data are derived from the Polity IV project’s dataset regarding countries’ political regimes. The case study reviews the basic concepts, terminology, and applications of multilevel modelling.