Politics and Elections

With Christopher Campbell, I recently published A computational model of spatial politics: Hotelling-downs model as statistical physics in PLOS One. The paper extends the classic Hotelling-Downs picture of electoral competition to multidimensional opinion spaces.

What the model shows

"Opinion space" is a concept that each voter can be represented by their position is a "space", the axes of which represent opinions on different issues. Traditionally, there is the economic left-right axis, but increasingly there are conservative-liberal views on social issure. The model treats political parties as moving through an opinion space in search of votes. By including turnout effects, activist pressure and competition between more than two parties, it can produce either convergence on the political centre or strong polarisation.

We also show how it is possible for voters to have a consensus on all issues (single peak distribution) but still get polarisation (multiple peaks) in a 2D opinion space. This is due to correlations, which may be fostered by position of politial parties.

One counterintuitive result is that two-party governments are more representative than muti-party ones. This is because both parties are forced towards moderate positions, while multi-party systems can drive parties toward the peaks of the multi-peaked voter distribution. The work also shows how lower turnout and activist influence can amplify polarisation even when the electorate as a whole is comparatively moderate.

Interactive versions

Related interactive pages on this site: Hotelling model and opinion dynamics.