2022 Federal Polling Year In Review

2PP average for year 54.9 to Labor.
Labor led on 2PP in all 87 polls released this year.

At the end of each year I release an annual review of federal polling. See the 2021 edition here and/or click the annual poll review tab for articles back to 2014.

I'll start by saying this was a great year for polling generally with a good industry result in the federal election (after a famous failure in 2019), a very good result for final polls in the Victorian state election, and YouGov/Newspoll performing very well in South Australia.

How many polls?

Once again the business of counting how many polls there have been is complicated by Morgan's habit of often releasing only the 2PP from a poll sample.  In all I count 87 readings, the most since 2017, from what I consider to be mainline pollsters: 

* 12 Newspolls before the election and four after.  The dramatic slowing in release of Newspoll post-election makes me suspect YouGov had a contract for sixteen (the same number as 2021).

* Six Resolves before the election and five after, however only two of the pre-election polls and none since had a pollster-derived 2PP (I've calculated last-election preferences for those that didn't)

* 14 Morgans before the election and 30 Morgan readings post-election, however nearly all the Morgans since the election have been 2PP only.

* Nine Essential readings before the election and two afterwards.

* Four Ipsos polls before the election

* One Freshwater Strategy poll this month

There may be more pre-election Morgan readings that I missed. I have not included Dynata (lobby group poll) or anything from KORE (panel survey with numerous red flags) or ANUPoll (wildly inaccurate with incomplete data).

2PP Voting Intention

The story of published 2PPs is a simple one: Labor won them all.  It also won every 2PP that I derived from published primaries, noting that the Resolve polls are mostly missing from the former data set and the Morgan polls are mostly missing from the latter.

Morgan made a significant method-shift two polls out from the election, ditching respondent preferencing (which skews to Labor) in favour of last-election preferencing.  I have not always replicated these so it would be good to see the full series of primary figures.

Prior to the election the closest Labor leads in published 2PPs were a couple of 1-point leads in Essential's "2PP+" measure in February and April (effectively 50.5-49.5).  The widest was a 58 from Morgan in March.  I got the same April Essential as the closest by last-election preferences (51-49) and the same March Morgan as least close (57.1). It turned out that last-election preferences were actually underestimating Labor by about 1%.

During the pre-election phase Newspoll averaged 54.0, Resolve (by last-election preferences where no 2PP published) 53.3, Essential 51.8, Morgan 55.9, Ipsos 54.9.  Essential's low leads for Labor could have been a result of use of Party ID weighting (past weak supporters of parties that are on the nose at a given time may be less likely to identify as such, meaning strong supporters might get over-weighted), while Morgan's large leads were probably partly caused by using respondent preferences.  

An estimated aggregate of published 2PPs pre-election (weighting the average for each pollster to weight those that polled in more months higher) is 53.9 to Labor. (The election result was 52.13.)

Post-election the averages are 55.8 for Newspoll, 59.1 (!) for Resolve by last-election preferences, 54 for Essential, 54.1 for Morgan (the use of last-election preferences making a big difference) and 54 in the single Freshwater. The lowest and highest released 2PPs were a 52 from Morgan in late August and a perhaps tinsel-distorted 58.5 from Morgan in its last December reading. However I get three of the Resolve polls even higher than that, peaking at 61.3 in August. Similarly high Resolve implied 2PPs have been seen in state polling, albeit less consistently than federal. 

Slightly lower 2PPs can be obtained by lumping Resolve's "independent" and "other" readings (there is some case for doing this since Resolve's overestimate of the "Independent" vote would fall mainly in seats where the preferences of independents would probably break more weakly to Labor than elsewhere if they did run.)

As a weighted average I get 55.6 for the post-election period, and all up 54.9 to Labor for the year. That's the best year for either side since Labor averaged 56.0 in 2009 (not that that did them much good), and indeed I place it fifth in the Newspoll era with only the Coalition in 1996 and Labor in 2007-9 higher (Coalition 2011 is very close.)

Here's an aggregation of pollster-released 2PPs per month, adjusted for apparent house effects including a shifting house effect for Morgan when its 2PP method changed:

The December reading may be inflated by the house effect correction for Morgan (for all I know said pollster may have changed some of its methods again) but Bludgertrack is also showing a (albeit milder) uptick in non-Morgan polls in December.

Leaderships

In the first five months Scott Morrison's Newspoll net satisfaction rates were generally lower than even the -8 he finished last year on. He averaged net -12.6 with a range of -7 (late April) to -19 (late Jan). Anthony Albanese averaged net -5.3 with a range of -14 (mid-April) to +2 (mid-March). In mid to late April Morrison was at times the better rating leader but it didn't last. 

On the skewed (to incumbents) Better Prime Minister indicator Morrison led by an insufficient average of 3.3 points, with ten leads and two ties.

Since the election we've seen lopsided leader polling with Albanese averaging net +31.5, Dutton net -7 and Albanese having an average 33.8 point Better Prime Minister lead. This is only based on four Newspolls but very similar patterns have been seen from other polls that canvass leadership ratings.

Betting

I mention betting odds in these annual roundups for interest although they are not reliably predictive. There is currently a sizeable (6.4%) arbitrage between two major bookies, one of whom has an implied 68% chance of Labor winning the next election and the other 79%.

Can Labor Avoid The First-Term Curse?

Recently first term federal government have reliably been re-elected, but their results have been underwhelming.  A history of swings against such government has some Opposition-boosters hopeful that an average-sized swing would at least see Labor lose the 2PP.  However, I don't trust this reasoning. 2PPs tend not to correlate from one election to the next so it should not be expected that the swing from a 52.13% 2PP would be as big on average as the swings from more lopsided wins. Also, just because there hasn't been a swing to a new government for many decades doesn't mean such a thing cannot happen. (Changes of government are rare, so we don't know that much about their properties.)

While the Albanese Labor honeymoon has been smaller than the Rudd one, it has so far been enduring (unlike that for the Abbott government which lasted only a few months.) Predictively, that doesn't mean a thing, but several of the recent governments that have had large honeymoons have complicated their first re-election attempts. Labor in 2010 removed their own Prime Minister for reasons they still struggle to explain, the Coalition in 1998 had to carry the burden of the "never ever" GST, and Hawke in 1984 was disadvantaged by a long and personally rocky and underwhelming campaign and perhaps informal voting. The last sophomore PM to get it all more or less right was Malcolm Fraser, who won by lots, albeit with a swing against his government.

It is easy to write off oppositions at times like this. Next year we'll see if the Coalition can make the inroads that it couldn't make in 2022.