Poll Bludger’s Griffth byelection live blog

Sunday

While Terri Butler’s 2.3% buffer at the end of the night is enough to secure her victory, Bill Glasson can at least claim the uncommon feat of delivering a by-election swing to the party in government. The current margin represents a 0.7% two-party swing to the Liberal National Party compared with the September election result, which is likely to widen a little further on postals.

Commentators around the place have been scrambling to place the result into historical context, mostly with reference to the long record of federal by-elections. A general paucity of swings to governments is easy to spot, but closer examination shows how much swings can vary according to the circumstances of the by-election, and how unreliable a guide they can be to a government’s future electoral performance. The last pro-government swing federally was achieved when Carmen Lawrence moved from state to federal politics in Fremantle in 1994, and it was followed by a shellacking for the Keating government at the next general election two years later. The most recent state example I can think of is the Peel by-election in Western Australia in February 2007, when Alan Carpenter’s Labor government boosted its margin 18 months before being dumped from office.

Given the array of circumstances that can bring by-elections about, an effort should be made to compare like with like. Griffith is part of a long tradition of by-elections held when a member of a defeated government decides opposition isn’t for them. Unfortunately, those involved tend to be senior figures representing safe seats which the opposing party doesn’t bother to contest. During its first term, Rudd Labor only took the field when Peter McGauran departed in the seemingly winnable seat of Gippsland, only to cop a bloody nose for its trouble. Few were surprised Labor stayed out of the fray in Higgins (Peter Costello), Bradfield (Brendan Nelson), Mayo (Alexander Downer) and Lyne (Mark Vaile). Labor likewise went undisturbed during John Howard’s first term at by-elections to replace Paul Keating in Blaxland and John Langmore in Fraser.

The one by-election held during the parliament elected in December 1975 was occasioned by the death of Rex Connor, with the remainder of Labor’s diminished caucus staying put. It was a different story early in the life of the Hawke government, as Malcolm Fraser (Wannon), Doug Anthony (Richmond), Billy Snedden (Bruce), Jim Killen (Moreton) and Tony Street (Corangamite) headed for the exit at a time when forfeiting a by-election was still thought poor form. The last useable example in anything resembling modern history is the Parramatta by-election of 1973, which brought Philip Ruddock to parliament.

From this field of seven, the only result to match Griffith is Richmond in 1984, when Labor picked up a slight swing upon the retirement of Doug Anthony. No doubt this reflected an unlocking of the loyalty accumulating to brand Anthony, which between father Larry and son Doug had occupied the seat for an unbroken 46 years. Even so, the other Hawke government by-elections weren’t far behind, with the exception of Bruce where voters seemed to take a shine to Liberal candidate Kenneth Aldred for some reason. Coincidentally or otherwise, the two worst swings, in Gippsland (a 6.1% swing against Rudd Labor in 2008) and Parramatta (a 7.0% swing against Whitlam Labor in 1973) were suffered by the two shortest-lived governments of the modern era.

However, Griffith looks quite a bit less exceptional if the eight state results I can identify going back to the early 1990s are thrown into the mix. Four swings in particular dwarf those in Griffith, the two biggest being at by-elections held in country seats in New South Wales on May 25, 1996. Results in Clarence and Orange provided a fillip to Bob Carr’s year-old Labor government and a severe blow to the Nationals, perhaps reflecting the party’s recent acquiescence to the Howard government’s post-Port Arthur gun laws. On the very same day, Labor had an historically mediocre result against the Liberals in the Sydney seat of Strathfield, and finished third behind the Democrats in the Liberal stronghold of Pittwater.

The third and fourth placed results are from early in the life of the Bracks government in Victoria, when Labor pulled off rare victories in Jeff Kennett’s seat of Burwood in 1999 and Nationals leader Pat McNamara’s seat of Benalla in 2000. Also higher up the order than Griffith is the Elizabeth by-election of 1994, held four months after Dean Brown’s Liberal government came to power in South Australia. This may have indicated the popularity of outgoing member Martyn Evans, soon to be Labor’s federal member for Bonython, who had been designated as “independent Labor” for most of his ten years as a state member. Rob Borbidge’s Queensland government of 1996 to 1998 did less well, with the looming Liberal collapse in that state foreshadowed by swings to Labor in the Brisbane seats of Lytton and Kurwongbah.

All of this is laid out in the chart above, which ranks swings to the government (positive at the top, negative at the bottom) from the eight federal and eight state by-elections just discussed. Red and blue respectively indicate Labor and Coalition governments, the lighter shades representing state and the darker representing federal. Stats enthusiasts may care to know that the model y=10+44.3x explains 38% of the variability, where y is the government’s eventual longevity in office measured in years and x is the swing to the government across 15 observed by-elections. For what very little it may be worth, the positive 0.5% swing in Griffith associates with 10.3 years in government.

Saturday

Midnight. Finally got around to adding the Coorparoo pre-poll voting centre result.

9.22pm. Or perhaps not – Coorparoo pre-poll voting centre still to report, which is likely to amount for a lot – 5859 votes cast there at the federal election.

8.51pm. Morningside 2PP now in, and I’d say that’s us done for the night.

8.36pm. Camp Hill reports 2PP, leaving just Morningside. Glasson and LNP reportedly not conceding, but 2.4% leads (which accounts for the fact that the LNP is likely to do better on postals – Labor’s raw lead is 3.3%) don’t get overturned on late counting.

8.29pm. That’s all the fixed booths in on the primary vote; the outstanding ones referred to in the table are special hospital booths that may not actually exist (but did in 2013). Camp Hill and Morningside still to come in on two-party, and then I think we’re done for the night. It’s been a pretty quick count.

8.03pm. Two more booths a slight move to the LNP.

8.00pm. Four more booths in and a slight tick in Labor’s favour on the swing projection.

7.53pm. One more primary result and a number more on two-party preferred, it remains unclear who will end up with bragging rights to the negligible swing. What is clear though is that Terri Butler is over the line.

7.45pm. Thirty-two of 43 booths reporting, and the picture of a status quo result is unchanged.

7.36pm. A big rush of results that taxed my data entry chops to the limit has produced very little change to the projection, which essentially looks like no swing at all.

7.26pm. Coorparoo Central was a tricky one from a booth-matching perspective, as it’s a “merger” of two booths from the 2013 election.

7.25pm. Coorparoo Central, Greenslopes, Morningside South and West End in on the primary; Buranda West and Norman Park South on two-party. Upshot: a bit more breathing space for Terri Butler, who will be difficult to pull in from here.

7.19pm. The informal vote seems to be down pretty solidly.

7.18pm. I note that none of the booths from the electorate’s north-western latte belt have reported yet.

7.15pm. Bulimba, Carina Heights, Greenmeadows and Norman Park find Labor still with its nose in front, despite a slight swing against. Annerley has also reported a two-party result, so I’ve switched on preference projections based on the booths that have reported so far. This finds Labor’s share of preferences up 6% on 2013.

7.12pm. Annerley, Bulimba Heights and Norman Park South booth results provide better news for Labor, with Butler now pulling into a projected lead. However, I’m still going off 2013 preferences here, as only two very small booths have reported two-party results.

7.03pm. Holland Park and Buranda West are in, and also Murarrie on two-party, and the swing to the LNP is sticking, as is the extremely close projected result.

6.53pm. Another small booth on the fringe of the electorate – Mount Gravatt East in the south-east – and it’s another bad result for Labor, down almost double digits on the primary vote. I’ve switched off the preference swing calculation for now, so the two-party is going off 2013 preference flows.

6.50pm. The preference result is in from Holland Park West, and Labor has 7.2% more preferences than it got in 2013 – but we’re only going off 14 votes here. Nonetheless, my model is extrapolating off it to project the result for the other booth, causing Labor to go up about 1%. Bottom line: hold off reading anything into anything yet.l

6.42pm. Both booths are on the very fringes of the electorate: Holland Park West in the south and Murarrie in the east. The dynamic nearer the city may well be very different. I’ll stop getting a “#VALUE!” result on the Stable Population Party when I get a result from a booth where their vote in 2013 wasn’t zero.

6.38pm. Two very small booths on the primary vote provide a measure of encouragement for Bill Glasson, suggesting a very close result if 2013 preferences are any guide.

6.30pm. There have apparently been 2090 ballots cast at the Whites Hill booth, which compared with the 2083 cast at the election suggests a pretty healthy turnout.

6pm. Welcome to the Poll Bludger’s live coverage of the Griffith by-election. With polls closing round about now, first results should be in in maybe about an hour, although this can be a bit variable. Results in the table will show raw figures for the primary vote and booth-matched swings for both primary and two-party vote, together with a projected two-party total based on the booth-matching. Where available, the latter will be based on booth two-party results; at booths where only primary vote totals have been reported, two-party projections will be derived from 2013 preference flows taken together with the “swing” in preferences recorded across booths where two-party results have been reported. I’ll be copping my results off the ABC Elections page, as the AEC annoyingly does not publish booth results as they are reported (or at least, never has in the past). So those without a minute to lose should note that my table updates will lag about that far behind the ABC.