2024 Site Review

Bring Back Universal Analytics Again!

As in my 2023 review, I again open this annual site review by expressing my disgust at Google stuffing with the continuity of my previous site stats by replacing Universal Analytics with GA4, which is not only apparently worse and different but also very hard for me to understand.   Anyway, again, on with the show as best I can ...

The dotted top line of the graph below shows activity in 2024 in the form of user numbers per week.  There is a big spike on the left caused by the Tasmanian election buildup and post-count, and a substantial spike on the right for the overlapping ACT and Queensland elections.  Other spikes correspond to the Tasmanian Legislative Council counts in May and the NT election in late August.  

Site engagement as measured in total events was up 84% on 2023, making 2024 as best I can determine the second-biggest year for the site after 2022.  That said, the new Analytics version seems to show high levels of site activity relative to activity on specific posts when compared to the old one, so I am not sure that is reliable.

In 2024 I released 79 articles, up 13 on 2023.  This included 27 articles about the snap Tasmanian election. There were also six articles about the Queensland election, five for the 2024 Tasmanian Legislative Council election and four federal poll roundups.  

Twitter Vs Bluesky And The Psephosphere

Last year's introduction will do just as well for this one.  It's an almost annual tradition on this site to release something every Christmas Day.  Click the Xmas tag for previous random examples.  Why do I do this?  Partly it's a present for those who like Christmas, which in a very low-key non-religious fashion includes your host, but it's also a present for those who don't want to deal with this particular Christmas or even generally cannot stand Christmas and just wish it was a normal day when normal things happened.  And what could be more normal than niche meta-psephology?  Therefore, the campaign against compulsory Christmasing brings you ... whatever this is.  

As the owner of X, which I still call Twitter, continues to run it into the ground both by making almost invariably bad changes to the website and by generally being himself, interest in alternative microblogging websites has grown.  In June 2023, with Twitter appearing in some danger of permanent technical collapse, I signed up accounts on Bluesky, Mastodon and Threads to cater for those fleeing the mothership, and as an insurance in case Twitter died or downgraded its features to a point where I wanted to leave it.  Microblogging sites are very important to my work, both as a place to obtain news about electoral matters but also as a place to make people aware of newly released posts on here (given how irregularly I post them.)

Until recently my activity on Bluesky, Mastodon and Threads has been mostly a secondary service with links to articles on this site posted; sometimes I will crosspost a debunking of some particularly bad poll-shaped object, and usually I will reply to questions when I can.  However this changed in the wake of the 2024 US election, when there was a mass exodus of Twitter users to BlueSky as a result of Elon Musk's active support for the return of Donald Trump.  Over the coming weeks my total Twitter following dropped from about 16900 to 16400 as some of the refugees deleted their Twitter accounts.  At the same time my Bluesky following rose within a week from around 950 to over 2000, even though I was not initially posting that much there, and it has now reached nearly 4000.  Clearly plenty of account holders have become active on Bluesky without deleting Twitter, though some may have stopped posting actively on the latter.

2024 Federal Polling Year In Review

 2PP Aggregate Average For 2024: 50.9 To Labor (-3.9)
Labor lost aggregate lead late in the year

It's the time of the year when most busy pollsters take a few weeks off and I bring out an annual feature, a review of the year in federal polling.  Click here for last year's edition and for articles back to 2014 click on the "annual poll review" tab at the bottom of this one.  As usual if any late polls come out I will edit this article to update the relevant numbers.

2024 was another strong year in results terms for the Australian polling industry.  Pollsters came out in good numbers for the early 2024 Tasmanian election and did pretty well in a very hard to poll election, although the lobby group(s) that commissioned two Freshwater polls contemptibly failed to ensure the release of definitive results of either, leaving poll-watchers to play jigsaw puzzles with incomplete media reports. Polling for the Queensland election was mostly excellent though no one pollster nailed the result, and a mini-cluster of close-ish polls at the end led to some misreads of what in the end was not a close election.   Despite this there are storm clouds about in federal polling in the form of inadequate transparency from several pollsters and a somewhat suspicious level of clustering of results, especially at a stage in the cycle where that doesn't usually happen.  (More on the latter later).  US polling this year wasn't as bad as 2020, but didn't quite get the real story in part because of the latter issue.

Media Fail Again Over MRP Reporting

Redbridge/Accent MRP model projects Coalition would be likely to have the most seats if the federal election was held now.  

It does not say Coalition would be likely to have a majority.

It does not predict the result of the election.

---------------------------------------------------------

I thought an article was in order to unpack some of what is going on with the recent Redbridge/Accent MRP model and the woeful reporting of it by pretty much every outlet that has so far mentioned it.

MRP models (stands for Multilevel Regression with Poststratification) are not the easiest to explain to lay readers at the best of times, but what we have seen from several media sources reporting on this one goes beyond understandable confusion and into the realms of reckless innumerate false reporting.

What a MRP model does is to build a picture of how certain types of seats are likely to vote based on small samples of all 150 electorates.  Although each seat's sample is uselessly tiny by itself, by assuming that seats that resemble each other in ways that affect voting intention will vote similarly, one can smooth out a lot of the rough edges in the sampling, and samples of a few dozen voters per electorate can build a model that's about as good on a seat-by-seat basis as if those samples were actually a few hundred.  That still isn't very good on a seat-by-seat basis, but on a nationwide basis, the model could capture some general trends about the kinds of seats where each party is likely to be doing well or badly, and about how a party might be going in converting vote share to expected seats.