I do like an election analysis! After having some fun predicting the result in Indi for the last federal election, I thought I’d look at my home seat of Brunswick in the 2014 Victorian state election.
Jane Garrett, the sitting member for Labor, won Brunswick in 2010 with a 2-candidate preferred (2CP) vote of 19545 votes against Cyndi Dawes, the Greens candidate, on 16974; Garrett won with a margin of 3.5%.
In 2014, the Greens candidate is Tim Read. With 54.08% of the vote counted, the 2CP count gives Garrett 53.9% of the 2CP vote and a lead of 1897 – on the face of it a slight swing to her. Garrett might be re-elected quite safely. However, I’m not so sure it is clear cut just yet.
In 2010, Garrett won 56.3% of the ordinary votes from the 16 booths in Brunswick plus the postal votes. If these are the votes counted to date in the 2CP (it seems they are), then Garrett has actually suffered a swing of 2.4% against her.
In 2010, Garrett won 45.6% of early, provisional and absent votes; she didn’t get a majority of any of those vote types on a 2CP basis, and these are the votes yet to be counted. The question is whether Read, the Greens candidate in this election, can claw back enough votes to overcome the current deficit of almost 2000 votes. It might seem unlikely, but let’s look at whether it is possible.
From 9407 2CP early, provisional and absent votes in 2010, the Greens won by 827 votes. If we add the -2.4% swing seen in the 16 booths and postal votes to 45.6%, we might predict Garrett would get 43.2% of these remaining votes. If there were 10,000 remaining votes, Garrett might get 4320 votes, and Read would then get 5680 – not enough to overcome Read’s current deficit of 1897.
However, there were many more postal and early votes in this election. Antony Green reports that almost 30% of Brunswick’s electors cast postals or early votes in 2014. With an electoral role of 46954, that is almost 14000 votes. Take out the currently listed postals that have already been counted, and we’d expect 12000 early votes, almost 10000 more than in 2010.
In 2010, these early votes favoured the Greens, but it is very difficult to predict where they might fall in 2014 – many more people have taken advantage of the early voting in 2014, so it is hard to guess their preferences.
But add on almost 7000 absent and provisional votes (there were 6773 formal absent votes in 2010; only 400 provisionals), and we might expect 19000 further votes in the election for the seat of Brunswick. If Read wins 55% of these, he will claw back 1900 votes from Garrett – enough for him to win. That seems within the realm of possibilities if we see a swing of 2.4% against Garrett in these votes, noting that the Greens won 52% of the early votes and 55% of the absent votes in 2010.
The election in Brunswick is not over yet. It comes down to where the early and absent votes fall. If they move away from Garrett, Read might spring a surprise.
Edit (3:00 p.m., 30 November): Kevin Bonham doesn’t expect the early votes to be quite as different from the ordinary votes this year as compared to 2010. I’d tend to agree with his assessment. It is hard to see such a large swag of early votes being radically different from those cast on the day. Still, it is worth keeping an eye on this one.
Edits: I corrected some typos in the original version where I typed the wrong figures for the percentage of early and absent votes.