Causal description : moving beyond stamp collecting in political science

Abstract

The articles by Kees van Kersbergen and Daniele Caramani constitute an impressive joint plea in favour of descriptive analyses within comparative politics. They also warn, less convincingly, against an alleged obsession of the discipline with variation that, according to them, does not do justice to the similarity of many cases. This response demonstrates that limiting the analysis to similar cases creates the risk of engaging in the hapless exercise of explaining the constant. Augmenting the role of description without simultaneously advancing sound theoretical models furthermore leads to theory-free data mining exercises. I argue that all empirically oriented fields in political science could profit from what I call 'causal description' and hence the in-depth univariate analysis of the dependent variable.