By-election Archive

All votes have now been counted for last weekend’s Fisher by-election for the state seat in Adelaide’s southern suburbs.

Earlier this week I wrote about how, after Labor won a solid lead on election night, a strong Liberal lead in the declaration votes brought the race close to a tie. As of yesterday afternoon, all two-party-preferred votes have been counted and Labor leads the Liberal candidate by 24 votes.

As long as Labor remains in the top two when the official preference distribution concludes today, you would expect Labor to win by roughly 24 votes.

However, there is a scenario where independent candidate Dan Woodyatt could overtake Labor and win the seat on Labor preferences, in a similar way to how the Greens won Prahran earlier this week.

On primary votes, Labor’s Nat Cook is leading independent Dan Woodyatt by 711 votes, and there are an additional 2864 votes cast for five other minor candidates.

For Woodyatt to win, he needs to gain a lead over Labor of 25% of those minor candidates’ votes. That could be a scenario where 50% goes to Woodyatt and 25% each goes to Labor and Liberal.

This is definitely a possible outcome, but there is not as much information about how those candidates’ preferences are flowing, as there was in Prahran earlier this week.

The preference distribution is taking place today, so we should find out this afternoon.

While we were all distracted by Prahran yesterday afternoon, quite strange things were happening in the count for the South Australian state seat of Fisher following Saturday’s by-election.

We all expected Labor to win the Fisher by-election after an error on Sunday revealed that Labor was holding 52% of the vote on election-day votes.

The Electoral Commission of South Australia (ECSA) added a whole bunch of ‘declaration votes’ yesterday, and with these votes the Liberal Party took a 17-vote lead. Just before this blog post went up, another batch was added and this resulted in Labor’s Nat Cook regaining the lead by 21 votes. At the time of writing, this update hasn’t made it to the ECSA website.

When I say ‘declaration votes’, I’m referring to all votes other than those cast at a local polling place on election day. This includes, prepoll votes, postal votes, absentee votes, and a few other small categories. There are no absentee votes because this is a sole by-election, so you would expect most of that category to be postal votes and prepoll votes.

While there was a swing of about 9% away from the Liberal Party on ordinary votes, the current sample of declaration votes suggests only a slight drop in the Liberal two-party-preferred vote, down to 55.6%. That seems quite unlikely.

Sadly, ECSA does not break down declaration votes by type, so it’s hard to know whether those votes counted are all postal or prepoll, and we can’t separately compare them to similar votes from March.

It’s also unclear whether all votes have been counted. However, when you compare total votes counted, it seems that not as many votes have been counted. In March, 24,087 votes were counted in Fisher, including informal votes. So far, only 21,175 votes have been counted. It’s likely that there has been some drop in turnout, but it’s also possible that some declaration votes are yet to be counted.

On Twitter, Nine News reporter Tom Richardson reported that most declaration votes have been counted, with a small batch to come.

The Liberal Party was only winning by the slimmest of margins thanks to a very high declaration vote, considering their election day vote. If there are more votes, and they don’t favour the Liberals by the same margin, then you would expect Labor to win. Labor has already taken a slim lead.

Update: Sunday’s recheck did indeed discover an error at the Aberfoyle Park booth which means the booth was won by Labor with 54%, not the Liberal Party with 54%. This puts the Labor two-party-preferred vote over 52%, and pretty much locks down the result.

Original post: Yesterday’s by-election in the southern Adelaide state seat of Fisher produced an unclear result, but one major party should be very happy with the result, while the other should be disappointed.

Fisher is a traditional Liberal seat – the seat was held by the Liberal Party from 1989 until Bob Such left the party in 2000. At the March 2014 general election, the Liberal Party won 57.2% of the two-party-preferred (Liberal vs Labor) vote in Fisher, which suggests that the party would have won the seat without Such’s candidacy.

In comparison, Labor won 50.9% of the two-party-preferred vote in the nine polling places used on election day. Labor won four out of nine polling places. In March, the Liberal Party won all ten polling places, with a vote ranging from 52.3% to 68%.

Independent candidate Dan Woodyatt ran explicitly as a successor to Bob Such, but polled much less than Such. Woodyatt is now polling 22.5% of the primary vote, compared to 38.4% for Bob Such in March.

While around 16% of Bob Such’s voters moved away from Woodyatt, the Liberal Party completely failed to gain those votes. The Liberal vote is currently down 0.05% compared to March, while Labor’s primary vote is up by over 10%.

I don’t have exact figures, but it appears that the number of declaration votes (which includes postal and prepoll votes) has increased compared to March. Antony Green estimates that the Liberal Party will need to do 3% better than they did on election day, which is greater than the gap in March.

If Labor holds on to their lead and wins the seat, the Labor government will regain a majority in the House of Assembly, with 24 seats as well as the support of two independents, compared to 21 seats for the Liberal Party.

For my analysis, I have split up booths into the same three areas used in the by-election guide: East, Central and West.

The ALP won 54.7% of the two-party-preferred vote in the east, and 49% of the 2PP vote in the Central area. The Liberal Party won a large 63.5% majority in the west, but the population is much smaller in that area.

One other thing worth noting: Antony Green has identified a possible error in counting that may have underestimated the Labor two-party-preferred vote.

In seven out of nine polling places, the proportion of minor party and independent preferences flowing to the Liberal Party was clustered from 30% to 35%, but the Liberal Party received 45% of preferences in Clarendon, and over 50% of preferences in Aberfoyle Park.

If you look at the following map showing the swings after preferences, you’ll also see that the swing to Labor is much lower at those two booths. Every other booth has a swing to Labor of 7-11%, but those two booths have a swing of 2-3%. This discrepancy does not exist on Labor’s primary vote – Labor gained a swing of 9.6% on primary votes in Aberfoyle Park, although the Labor primary vote swing in Clarendon was substantially lower at 4.2%.

There’ll be a recheck of votes tomorrow, so if there is a problem it should be discovered, and this would make the challenge even more serious for the Liberal candidate.

Two-party-preferred votes at the 2014 Fisher by-election.

Two-party-preferred swings at the 2014 Fisher by-election.

Liberal primary votes at the 2014 Fisher by-election.

Labor primary votes at the 2014 Fisher by-election.

Primary votes for independent candidate Dan Woodyatt at the 2014 Fisher by-election.

Fisher by-election primary votes – all booths reporting, no declaration votes reported

Fisher by-election two-party-preferred votes – all booths reporting, no declaration votes reported

9:01pm – With no more results expected tonight, I would have to say Labor is the favourite to win in an extremely close race.

8:59pm – Labor has won 62.44% of preferences distributed tonight, with the Liberal Party winning 37.56%. If you apply the swings we saw on the booths to the March declaration vote, and then distribute preferences from Woodyatt, the Greens and other minor candidates in the same proportions, then Labor wins the declaration vote by 50.55%. Without knowing how big that vote is, that would see Labor win with a slightly reduced margin. Having said that, this assumes that the declaration vote is similar to what it was in March. We can’t assume that.

8:53pm – And we now have the final 2PP figures for election day – the Liberal Party narrowly winning Aberfoyle Park North. This leaves Labor leading by 269 votes before the inclusion of declaration votes. According to Antony Green there has been a big increase in the number of prepoll votes which makes it hard to predict how they will break.

8:21pm – Using the same model as before (which actually underestimated Labor’s preference flow slightly), I expect the Liberals to win Aberfoyle Park North but not by enough to offset Labor’s current lead. Then it will all come down to declaration votes. In March, there was 5865 declaration votes in Fisher. If Labor holds on to a lead of 227 votes after Aberfoyle Park North reports, then the Liberal Party would need 51.9% of the 2PP to win.

8:17pm – We just got the primary votes for Aberfoyle Park North, and the 2PP figures for Reynella East and Happy Valley West. Those 2PP figures have pushed Labor into first place by more than I predicted – they’re leading by 371 votes. Aberfoyle Park North also saw a solid pro-Labor swing but not as big as some other booths.

8:12pm – If you look at the three Aberfoyle Park booths that have reported votes, the Liberal Party is leading 2020-1989 – a 31-vote lead. If something similar happens at Aberfoyle Park North – a reasonably large booth – then Labor may end up ahead by about 100 votes before declaration votes are counted.

8:09pm – So far there are six booths reporting 2PP figures, and in those booths the minor party vote is splitting roughly 60% to Labor and 40% to the Liberal Party. If you extrapolate that to Reynella East and Happy Valley West, then Labor turns a 386 vote deficit into a 157-vote lead. However Labor is not expected to do quite as well at the ninth booth, Aberfoyle Park North, from which we’ve heard nothing.

8:04pm – We now have eight out of nine booths reporting and the ALP is well ahead of Woodyatt. There’s roughly 14% of the vote with other candidates but it’s hard to see Woodyatt overtaking Labor – quite a lot of that will flow to Labor or Liberal.

7:54pm – Labor has also topped the primary votes in Happy Valley West, which has pushed Cook ahead of Woodyatt on primary votes. The projection model is holding steady.

7:45pm – We’re now missing primary votes from Aberfoyle Park North, Happy Valley West and Reynella East. Happy Valley West, Reynella East and the abolished booth of Woodcroft were the only booths where Labor polled over 20% in March 2014. They were also the three worst booths for Bob Such. This is why my model is suggesting Labor will overtake Woodyatt, but that certainly could be wrong.

7:42pm – Vote after preferences now reported from Happy Valley and the Liberal Party won – just. This narrows the Liberal lead from 52.8% to 52.2%.

7:37pm – Now have the 2PP figures for Aberfoyle Park Central and the primary votes for Aberfoyle Park, and they are both good for Labor. Labor has now won the vote in the two Aberfoyle Park booths to report 2PP figures. In the two booths reporting primary votes but no 2PP, Labor is up over 10% and the Liberal Party is down 2%. Overall, the Liberal Party is only leading by 2.8% after preferences off a sample biased towards them.

7:25pm – We’ve just added another Aberfoyle Park booth – four out of nine booths are in Aberfoyle Park – and we saw a 0.3% swing against the Liberal Party and a swing of over 11% to the ALP. We haven’t yet gotten 2PP figures for either Aberfoyle Park booth.

7:22pm – I’ve also now added in the two-party-preferred count between Liberal and Labor. There are no swings or projections possible on these figures, but it’s worth noting that Labor narrowly won in Aberfoyle Park South. We haven’t gotten results from Happy Valley, but there was a primary vote swing of over 10% to Labor and 2% away from the Liberal Party there, so that should increase Labor’s 2PP vote.

7:18pm – An even bigger booth, Happy Valley, has about as many voters as the first three booths combined. The Liberal vote has actually gone down at Happy Valley by 2%.

7:14pm – The first large booth – Aberfoyle Park South – has reported, and the Liberal vote has dropped to 41%. The projection is roughly the same – Liberal on 41%, Labor 25%, Woodyatt on 17.5%.

7:09pm – With two booths reporting, the Liberal Party is on 50% of the primary votes, with independent Dan Woodyatt second on 21% and Nat Cook on 14%. However my projection suggests that these booths are strong for the Liberal Party and particularly bad for Labor, and if the trends continue (comparing Woodyatt to Such’s vote in March), then Labor will overtake Woodyatt and the Liberal vote will drop to 42%. Heidi Harris should still win on that vote.

6:00pm – Polls have just closed in the by-election for the state seat of Fisher in the southern suburbs of Adelaide. I’ll be back with results as soon as they are available.