Federal Election 2022: Pollster Performance Review

In 2019 the Australian polling industry had a disaster after decades of reliability - every final poll had the wrong two-party winner, the final polls were bizarrely clustered around the same wrong result, many final polls were individually wrong by more than their claimed margins and the lack of transparency in the industry was such that it was difficult to understand just why it had happened.

Fortunately 2022 has not been a repeat.  The fallout from 2019 saw a great increase in polling transparency, especially via the formation of the Australian Polling Council (though unfortunately not all pollsters have been on board with that) and also more diversity in polling approaches.  No one poll has ended up nailing the remarkable results of this year's election, but collectively, federal polling has bounced back and done well.   This is especially so on two-party-preferred results, where a simple average of the 2PP figures released in the final polls is pretty much a bullseye. The primary vote results were a little less impressive.

Here I discuss polling in several categories.  Overall YouGov (which does Newspoll) made the most useful contributions to forecasting the result, Redbridge's performance in publicly known niche polling during the campaign was very good, and Resolve Strategic's final poll was a useful counterpoint to Newspoll.  The other major polls were so-so on the whole, and many minor pollsters were wildly inaccurate.

Poll Roundup: 2022 Seat Poll Sludgefest

This is the second part of what was at first intended as a single roundup, and deals with seat polls and seat betting.  Any new seat polls will continue to be updated on this page until election day while a seat betting update will probably be added to a final roundup tomorrow night.

As a general comment, 2022 has seen a major change in the election seat polling landscape.  At previous recent elections, published seat polls were dominated (in public attention if not always in numbers) by major national players - YouGov, ReachTEL, the old Newspoll and so on, or at least by specialised pollsters who were distant from campaigns (JWS).  Often the seat polls still sucked.  In 2013 they skewed to Coalition on average (some firms more than others), in 2016 they were under-dispersed and in 2019 they did well at picking winners but badly when they sat on the fence (and they also skewed to Labor).  But at least they were neutral attempts by pollsters with skin in the national game.  

At this election YouGov has so far given seat polling the flick and instead switched to its MRP model (discussed too briefly in a previous edition).  The seat poll landscape has been dominated by uComms (a union-connected pollster with simplistic weightings and an ordinary recent track record), Redbridge (another campaign-focused pollster with often weird methods decisions and a remarkable ability to detect UAP voters), and to a lesser extent Utting Research and KJC/Telereach, neither of whom have had much public testing and the first of which is not an APC member, with publication of details rarely exceeding a single media article.  The overwhelming method of seat polling has been robopolling of often already saturated seats (one voter in Swan this week told me they'd been polled seven times).  Moreover, the two most commonly seen pollsters have been mostly conducting internal and campaign-adjacent polls rather than media-commissioned polls.   The seat polling landscape has been dominated by strategic or incidental releases of polling for campaigning purposes - mostly fed to journalists to get publicity and (in the case of teal independents in some seats) try to exploit strategic voting arguments.

South Australia 2022: Final Newspoll And Other Comments

Time for a short post ahead of tonight's South Australian state election, which I will be covering live on here from the close of polling and, as time permits, through the postcount over coming weeks. 

This has been another state election to have seen remarkably little polling.  The only statewide public polls have been two Newspolls by YouGov and one self-branded YouGov poll (the two versions employing different weightings).  The first Newspoll had a 53-47 Labor 2PP lead, the final Newspoll had 54-46 (41 Labor 38 Liberal 9 Green 12 Other), and the YouGov in between had 56-44.  There is very little sign of life for the Marshall government in all this, but it's still within the reach of polling error that it might lose the 2PP more narrowly (or if it's a big error, get around 50-50) and then, perhaps, get lucky enough with the distribution of votes to hang on in a hung parliament.

Using the 54-46 Newspoll as an input, my conditional probability model (see previous article) suggests that Labor would win the 2PP in around 26 seats to 20 for the Liberals (Mount Gambier is excluded from the model).  However, there is a widespread expectation (including some remarkably short betting odds) that the ex-Liberal independents will all be re-elected despite the clouds hanging over all of them except Dan Cregan.  That would end up with a 26-17-4 parliament.  The seat model can't reliably predict which of the seats on 6-10% margins would fall, other than that on a 6% swing enough of them would be close enough to going that Labor should pick up a couple just by luck.  A similar outlook is seen in betting odds, which continue to have Labor favourites in the low-hanging fruit of Adelaide, Newland, Elder and King, but not in any other Liberal seats.