Lumpers and Splitters on the United Kingdom Supreme Court

Abstract

I model the probability that two judges on the Supreme Court of the United Kingdom (UKSC) will indicate their express agreement with each other. I do so using a latent space model for social network data, using as my unit of analysis the undirected judge-judge dyad in each case. I test for the effects of homophily (same legal background, similar age, same inn of court, similar length of service), case complexity and texture (days of oral argument, presence of intervenors, human rights issues raised), and interactions between judge and case characteristics (whether a judge in the dyad writes the leading opinion or not; whether a judge in the dyad is a President or Deputy President of the court). I also allow for judges to vary in (a) their propensity to indicate express agreement with any judge (i.e., their tendency to be a ‘lumper’ rather than a ‘splitter’), and in (b) their position in a one-dimensional latent space. Whilst case characteristics are the most important predictors of whether judges will agree with each other, judges’ propensity to ‘lump together’ with their colleagues is the most important judge characteristic.