Joint Seminar between the School of Mathematics & Statistics and the School of History, Philosophy, Political Science and International Relations

Recent decades have seen rapid progress in the development and application of models to predict election outcomes. Typically, these models seek to aggregate, or synthesize, information from multiple political polls in such a way that they can generate probabilistic predictions about election outcomes. In this talk, weâ€^TMll discuss the development and application of one such model, namely the NZ Herald – Motu Research poll-of-polls, which has been launched in the lead-up to the 2026 General Election. The talk will cover relevant aspects of the New Zealand context, the general structure of political polling data, our approach to modelling election outcomes, and highlight several useful insights provided by the model. To finish, we will discuss some potential limitations of the model and potential avenues for further research.