In many other countries it is possible to predict election results reasonably accurately. Sometimes this is done using public opinion polls or, in some countries, by drawing on national economic indicators.

In PNG these types of approaches are not possible. Public opinion polls are too expensive and the localised nature of electoral contests means national economic indicators cannot be used to predict elections either.

There is an alternative though, one identified by the Political Scientist David Hegarty in the 1980s. This is to look at candidate numbers in individual electorates. In recent years we have tested this approach in Papua New Guinea and neighbouring Solomon Islands.

In this presentation, Dr Wood will talk about whether it is possible to predict election outcomes for individual electorates using candidate numbers. He will discuss the most recent model and how well it would have predicted election outcomes in the 2017 and 2022 elections. He will explain how, even when it performed poorly, the model identified other interesting electoral patterns.

A discussion paper on the topic by Terence Wood and Maholopa Laveil is now available on the Development Policy Centre’s website.

The monthly ANU-UPNG seminar series is part of the partnership between the ANU Crawford School of Public Policy and the UPNG, supported by the PNG-Aus Partnership.