Contextual effects and other influences: using multilevel modelling to study the extent and causes of spatial variations in post-war Australian federal voting

Abstract

Although the techniques of multilevel modelling have been applied to survey data for some individual elections in a small number of countries (including Australia), with the objective of investigating some aspects of the extent and causes of spatial variations in voting behaviour, there are currently no published general studies that cover lengthy periods for any country. In this paper I will provide such a detailed discussion for major party voting at Australian elections for the House of Representatives during most of the post-WWII period.I begin by outlining the overall extent of spatial variations at the level of states and electoral divisions and show how these are influenced by institutional factors and contemporary political events. Following this, I will present results of analyses that examine how much of the localised variation at the level of electoral divisions can be straightforwardly explained by social compositional differences that result from the spatially heterogeneous distribution of important individual level characteristics such as occupation and religion. I then outline some results on the extent to which the other main potential sources of spatial variation, contextual effects, have influenced voting behaviour. Finally, I shall give a detailed discussion of how the significance of particular contextual effects has altered over a long period and will relate these variations to political events and social and economic changes.