Institutionally-induced pendulum? An assessment of state-level influences on postwar Australian federal voting

The extent of state-level influences on federal voting behaviour in Australia has been debated for many years. In this paper, I extend and improve on previous research by using the most advanced techniques (multilevel modelling) available for analysing survey data in this context, and present results based on an extensive investigation of postwar Australian elections.The results conclusively show that the overall extent of measurable state-level effects on federal voting over the period is relatively very small, despite the institutional significance of the states. In an attempt to reconcile these two facts, I therefore propose an institutional explanation of the small extent of state-level effects. I argue that the representation entitlements of the states in the House of Representatives would tend to result in the overall extent of state-level effects being relatively small because they introduce a self-correcting, 'pendulum' element into party competition. I investigate some hypotheses that follow from this argument, demonstrate that the evidence from the multilevel modelling is consistent with these hypotheses and conclude that this institutional element is a significant part of the explanation of the apparent unimportance of state-level influences