What went wrong with the Polls during this Australian Federal Election?

Now that it has been a week following elections here in Australia, it may be time to take stock and unpack why the major polls showed a trend opposite to what actually happened.

A caveat first, while polling data consistently showed the Labor Party leading on a two party preferred basis, it was not always clearly stated whether the differences between the parties fell within each polls margin of error.

This lack of clarity as emphasised in this recent article in the conversation highlighting the lack of transparent documentation for polling methods used by the big four polling companies here in Australia. Read the article here

What this article alludes to is an increasingly problematic approach to polling worldwide, that appears to be veering away from methodological rigour and transparency, into the realm of loosely informed speculation due to a lack of attention paid to ensuring both valid and reliable sampling.

This could be due to the following: a lack of adequate randomisation protocols to obtain a distributive sample across key demographic targets, response bias due to how the information was captured (online versus robocols versus face to face interviews) and flawed analytical methodology which has not been reviewed under scrutiny.

These factors combine to create distorted sampling which may simultaneously under-represent some voters at the expense of over representing others, resulting in a sample with poor general validity.

However, until polling companies hold up their methods to scientific scrutiny and provide transparent documentation, the drivers of these consistent failures will be the realm of speculation in opinion pieces.

Sunny Naicker, PhD

Partner

Monocle