One or More Approaches to Social Sciences? Different Perspectives on Democracy Support

The question if one or more than one type of social science methodology exists has been dominating the methodological discussion of the last two decades. A special contribution to this discussion has been made by the American social scientist Charles C. Ragin who proposed Qualitative Comparative Analysis (QCA) as a strategy for moving beyond the qualitative and quantitative distinction. Numerous attempts have been made since then to assess the applicability of Ragin’s proposals, contrasting the results with more quantitatively oriented forms of analysis. However, this overlooks that the two ways of analysis do not exclude each other. Rather, they start from fundamentally different assumptions of how scholars arrive at valid results. Therefore, the interest of the paper is not to outweigh the two approaches against each other, but to demonstrate how different basic assumptions of the analytical process influence (or do not influence) the results we obtain. We demonstrate this with an example from the research area concerned with the health of democracy and citizens’ support. This seems to be a suitable case, since numerous current studies describe the concept of democracy, and various theoretical perspectives, deriving from them, have tried to formalize this concept in different ways. Our focus is on 24 consolidated European democracies. We assess with both regression methods and QCA the impact of economic aspects and of education on two measures of democracy support, namely, confidence in institutions and democratic principles. The data used are Eurostat (1993; 1999; 2000) and the European World Value Survey Integrated Data File (1999-2002). The results show that both methods perform better when we analyze ‘confidence in institutions’. In either case, it becomes clear that regression analysis looks at the contribution of the independent variables for the explanation of the dependent variable, whereas the QCA analysis renders us concerned if such a limited set of causal factors could not fall short to explain the outcome as well as we would wish. We also conclude that the two methodologies are not infinitely exchangeable. They rather have to be related to very distinct processes of doing research.