Using a spatial statistical model to explore voting behaviour in the UK 2019 general election

This paper examines aggregated voting behaviour in the 2019 UK general election using a hierarchical spatial autoregressive model (HSAM). Such a model aims to simultaneously measure two sources of spatial variation. The first is that due to a spatial unit’s position within a nested structure of administrative boundaries which are hypothesised as relevant a priori. The second is a type of spatial effect owing to the influence of neighbouring constituencies represented by an intrinsic conditional autoregressive (ICAR) component. The results show varying associations between census profile and voting behaviour at different levels and in different parts of England and Wales.