Causality in Political Networks

A 2011 paper in American Politics Research titled "Causality in Political Networks" discusses some important concerns about assigning causality to social network analysis studies, especially when it comes to political behavior and preferences. The basic definition of causality adopted by the authors is one put forward by Henry Brady in 2008 that basically says:

causal arguments are stronger to the extent that they demonstrate our elements: (a) “constant conjunction of causes and effects”; (b) No effect when the cause is absent in the most similar world to where the cause is present”; (c) “An effect after a cause is manipulated”; and (d) The identification of “Activities and processes [i.e., mechanisms] linking causes and effects”

The authors then go on to discuss, in five short essays, how causality may be assessed through social network analysis. From survey data, field experiments, socio-centric data to measurement errors and study-design, this paper is a pretty good place to start when thinking about how you can move past descriptive analysis to something closer to causal inference in social network analysis.

For the sake of disclosure: one of the authors of this paper is my former graduate advisor. Another was a professor during graduate school.

Citation: Causality in Political Networks. Fowler et al. American Politics Research 2011 39: 437.