What happened on the weekend to rev up the week ahead.

There’s an hilarious cartoon in the Australian Financial Review today, with a story from me below it. My op ed piece runs through the elections and by-elections we’ve had since the last Federal election on May 21 2022. There’s some good news in there for the Government I think, from the evidence and explanations of why it’s occurring.

🔗 https://www.afr.com/politics/state-polls-position-albanese-for-a-second-term-20240318-p5fd6g

Opinion polls are useful indicators to track public opinion, but by-elections and elections (including the big Brisbane Council ballot) are even more useful. It’s one thing for voters to tell someone on the phone how they are thinking of voting at the time, but altogether different when they actually front up to the booth and do it. And there’s more of them of course at an election.

Personal votes need to be taken into account here. The swing against Governments at by-elections tends to occur when it’s a Government sitting members personal vote being lost and if the Government is unpopular at the time, then that can get added to it. So they can get pretty big and bad news for a Government, if it is on the nose at the time, as we saw during the Whitlam Government in Bass.

But if a very popular Opposition MP retires, and the Government is travelling reasonably well with its constituents, we need to consider that the personal vote for the Opposition MP retiring was taken from the Government’s local candidate in the first place. The return of this vote to the Government candidate at the by-election isn’t a swing, just a reset. That’s why Aston’s figures looked good for the Albanese Government and the Dunstan figures looked pretty good for the Malinauskas SA State Government last weekend.

What demographic modelling shows us is an approximation of this personal vote and thus we can take it into account. I started researching personal votes and donkey votes 50 years ago when working for Don Dunstan, the then SA Premier, after whom the SA seat was named. We had a rare occurrence at the time with simultaneous Upper and Lower House elections and a decent set of rolls and, from memory, few minor parties to cloud the major party vote in the upper house. (You can do the same sort of thing with Senate and Reps votes, but it’s a lot harder these days with so many minor parties and more strategic voting.) I was able to isolate the donkey votes, get to the personal votes for the sitting members and the personal vote estimates was very close to demographic residuals for models we were doing at the time. So a strong demographic model, possible then in SA due to a range of demographic, economic and social factors, provided a good estimate of the vote the party could get at the relevant election and a pointer to the personal vote of the candidates.

We’ve been tracking personal votes and party votes ever since for most Federal and some State polls and the evidence tends to hold up pretty well when we look at the outcome for by-elections.

It’s interesting in that the personal vote is just that: personal. An MP in a city seat with a big population turnover tends to have a small personal vote, as the voter who’s been met in electoral office is often voting in another seat the next election and voters meet at social or sporting functions often live in other seats. But a country MP in a stable seat talks to voters who tend to live and play and work and vote in the same seat. This means when the MP meets someone it’s a local voter. And the MP builds up friends and personal contacts. Personal votes can also be negative, for cases where an MP antagonises his or her local constituents on a regular basis. I tend not to mention personal votes in election reports unless they are a couple standard errors of estimate above or below predicted votes. And, for the record, the best performing MPs in the current Parliament are female Labor representatives of regional and rural seats. And long-standing MPs for the Coalition also do well. If you’re ever running a campaign, these factors are taken into account. Or should be.

The other factor taken into account in the article is the voters’ desire for balance between parties at the state and federal levels, especially in Queensland. This is why winning State elections is not necessarily a good thing for Federal Governments of the same party. Anyway, check out the article. I hope its useful.

🔗 https://www.afr.com/politics/state-polls-position-albanese-for-a-second-term-20240318-p5fd6g