Melbourne by-election live

Monday

Rechecking and a little over 400 more postal votes have nudged Labor’s lead up from 754 to 772. Here’s a piece I had in Crikey yesterday:

Notwithstanding the Greens’ unduly stubborn refusal to concede defeat, it is beyond doubt that Labor is over the line in the Melbourne byelection. Its candidate, Jennifer Kanis, holds a 754-vote lead over Cathy Oke of the Greens, with only a few thousand votes outstanding and the tide of late counting running in Labor’s favour.

The result has surprised election watchers, national newspapers and, most memorably, Sportsbet, which went a step too far with its regular publicity stunt of paying out on sure-thing election results before the actual event.

As is often the case in byelections, there are enough intricacies in the result to allow interested parties to craft narratives to suit, be they Christopher Pyne comparing Labor-versus-Greens apples with Labor-versus-Coalition oranges, or Adam Bandt claiming a slight rise in primary vote share meant the electorate had “gone green”.

My own take on the result is that the Greens fell victim to an unexpectedly strong determination of Liberal supporters to deprive them of their votes.

One recourse was absenteeism, which saw turnout slump from 86.9% at the 2010 general election to no more than 67%. Another was informal voting, the rate of which shot up from 3.8% to 8.7%. Given the intensity of media interest, and the electorate’s high levels of educational attainment and civic engagement, these are remarkable figures.

Clearly some Liberal supporters managed to struggle their way through the ballot paper, but few seem to have given their support directly to the Greens, who have actually polled about 750 votes fewer than at the state election. That they were able to increase their overall share probably has more to do with relatively high turnout among their supporters than votes shifting in their favour.

Liberal votes instead scattered among the crowded field of minor candidates, of whom the best performers were Fiona Patten of the Australian S-x Party (6.6%), Stephen Mayne (4.7%), conservative independent David Nolte (4.7%) and the three Christian parties (6% combined), all of whom showed at least some tendency to poll most strongly where the Liberal vote had been highest in the past. Reflecting the pattern of Liberal preferences when they were directed against the Greens in 2010, these votes (which would have included a share of left-leaning supporters of Patten and Mayne) flowed about 60-40 to Labor.

Past state byelections had given the Greens cause to expect better. When the Liberals sat out the Marrickville byelection in inner-city Sydney in 2005, the Greens vote shot up 10.5%. In the Western Australian seat of Fremantle in 2009, Adele Carles claimed the seat for the Greens in the absence of a Liberal candidate by adding 16.5% to the party’s primary vote — and turnout actually increased.

That things were so different in Melbourne may well suggest that conservative voters are feeling more hostile to the Greens than they were a few years ago.

The result also fits a pattern of the Greens underperforming at state level in Victoria relative to federally. When Bandt won the federal seat of Melbourne in 2010, he polled 37.6% in the booths covered by the state electorate. This was almost exactly what Oke polled on Saturday, when the Liberals’ 28% share of the vote was up for grabs, and well above the 31.9% they polled at the 2010 state election. While this may partly reflect the fact that the hot-button issues for the Greens are most salient at federal level, it could equally be a reflection on a state parliamentary party that lacks a strong media performer.

As for Labor, while it can’t take too much joy at having dropped 3000 votes from the general election, it has room certainly for relief and perhaps even a flicker of satisfaction. Its primary vote has fallen 2.4%, which is about what pseph blogger Poliquant calculates as par for the course at byelections where the Liberals don’t field a candidate.

It is also clear that the 4.2% vote for independent Berhan Ahmed came largely at Labor’s expense, having been concentrated in a small number of booths where the Labor vote was correspondingly down (Stephen Mayne relates that Labor received about 80% of his preferences).

Certainly there are bad signs for Labor in the result as well, but they are nothing it didn’t already know about: that half its primary vote in Melbourne has vanished over the past decade, and that it is  becoming increasingly reliant on preferences in stitching victories together. However, it has equally been reminded that such victories can indeed be achieved, and that however calamitous things might be for it in Queensland and New South Wales, in Victoria the ship remains more or less afloat.

Sunday

Apologies for the Crikey-wide outage that appeared to kick in at about 11.30 last night. The VEC has announced on Twitter there are only 1000 postal votes to come, although it would surprise me if the current count of 3728 pre-poll votes were the final story, given there were 6268 of them in 2010. However, even if there are a few thousand votes still outstanding, they will offer the Greens no prospect of overcoming a 754-vote Labor lead that will widen further with the addition of the remaining postals.

I have reset the above table so it just shows raw results, in doing so removing what was projected as a 0.5% lead to the Greens. This reflected a 6.7% swing to the Greens on booth votes, compared with an overall margin of 6.2% from 2010. The projection went on to be buried by the addition of 3000 postal votes, which the VEC unusually decided to get stuck into on election night (together with 3728 pre-polls, which behaved more in line with the polling booth votes and thus made little difference to the overall picture). The postals split 59.6-40.4 Labor’s way, and while this actually represented a swing to the Greens of 1.6% compared with postals in 2010, the effect was to drag the overall swing below 5%. Another factor was that the Greens did extremely well on absent votes in 2010, which by-elections don’t have.

Labor’s win has come as a surprise to me, and I know I’m not alone in pseph-dom in this count. I had expected to see a pattern similar to that in the 2009 by-election for Fremantle, which had supported Labor, Liberal and the Greens in similar proportions to Melbourne in the past, and where homeless Liberals appeared to fall in behind Labor’s rival by way of taking a kick at the main enemy. Besides the result, the most radical difference between the two elections was turnout. Very unusually for a by-election, turnout in Fremantle (which I am measuring in terms of formal votes cast) actually increased, from 79.6% to 83.5%. Even on a favourable projection, turnout in Melbourne appears to have slumped from 83.7% to around 63%, a result interestingly similar to the South Brisbane by-election held a few months ago to replace Anna Bligh.

This makes it instructive to consider the election in terms of raw numbers of votes rather than percentages. There are roughly 45,000 voters on the Melbourne electoral roll, of whom about 7500 can be expected not to vote at a general election. Normally this could be expected to increase at a by-election to around 11,000, but this time it shot up to 15,000. No doubt Liberal voters were over-represented here, and its tempting to contemplate how different things might have been if the Greens had chosen a candidate as attractive to Liberal supporters as Adele Carles proved to be in Fremantle. However, it should not be assumed that the collapse in turnout can be entirely understood in terms of Liberals sitting it out, as there were also 3000 fewer votes for Labor as well as 750 fewer for the Greens.

Liberal voters made their impact felt in a a 7500-vote increase for “others”, most of which was garnered by (religious) conservatives and liberals. The latter were particularly prevalent around the CBD, where the Liberals have a considerable constituency. The standout example was David Nolte, who polled around 10% in Docklands and East Melbourne and also at the university end of Carlton, but very weakly elsewhere. Another independent with strong localised support was African community leader Berhan Ahmed, who polled 15.9% in Hotham Hill, 10.5% in Carlton and 10.1% in Flemington, but only 4.2% overall. There was a corresponding drop in the Labor primary vote in these booths. The other minor candidates to recover their deposits will be Fiona Patten of the Australian Sex Party, who is on 6.6% overall and reached double figures in and around the CBD, and Stephen Mayne, who failed to crack 5% but has a notable base of support in East Melbourne (11.3%).

Saturday

11.22pm. While I’ve had my eye off the ball, the VEC has caught me off guard by adding huge numbers of postal (3066) and pre-poll (3975) votes, the former of which have, as far as I’m concerned, decided the result for Labor. Labor has received 1702 postals to just 1156 for the Greens, a split of 59.5-40.5: 1702 (59.5%) to 1156 (40.5%). Pre-polls have slightly favoured the Greens, 1914 (51.3%) to 1814 (48.7%), but the overall result is an unassailable lead 754-vote (1.4%) to Labor.

8.49pm. Examination of the results from 2010 shows up a very telling point: the Greens did exceptionally well on absent votes, scoring 54.4% on 2PP. However, absent votes are those cast in polling booths outside the electorate – which is to say that they don’t exist at by-elections, because there are no polling booths outside the electorate. That would seem to suggest that my projection is flattering to the Greens.

To those who are confused by all this – and in particular by the disparity between my figures and the VEC’s – what I have done here is calculated the swing on the booth results, which are all we have at the moment, and that swing is 6.6%. Labor scored 57.4% on booth votes in 2010, and 50.7% today. After other votes were added in 2010, Labor’s vote came down to 55.8% – so on that basis, a 6.6% swing would suggest they are headed for a narrow defeat. But as just noted, the reason they came down was that the Greens did so well on absent votes. The non-existence of such votes at this by-election puts a rather different complexion on things.

8.45pm. Flemington 2PP added, so the projection is final for the night.

9.35pm. With all but one booth now in on 2PP, my projection now leans a little further to the Greens. BUT … at this point, that matters less than what the dynamic of pre-polls and postals is going to be. There could be any number of reasons why they might be a little more favourable to Labor (in relative terms) than they were at the state election, and that’s all it would take. I’ll have a think about that and get back to you, but with the negligible exception of the one outstanding 2PP result, my projection has achieved all it’s going to achieve this evening, which is to say that it’s too close to call.

9.29pm. Still awating Docklands, Flemington, Melbourne and South Kensington on 2PP, remembering that all this is likely to do is nudge the preference share slightly in one direction or the other.

9.27pm. Final primary vote result in (Flemington), and it tips the Greens into the lead on my projection.

9.15pm. The addition of eight 2PP results in one hit didn’t change the complexion of things any: Labor’s share of minor preferences changed from 60% to 61%.

9.14pm. I’m back. We’ve now got 10 of 14 booths on 2PP and 13 of 14 on primary (Flemington the holdout), and it’s as close as close can be.

9.07pm. South Kensington and Melbourne have reported, but my spreadsheet’s crashed. With you in a minute or two …

9.00pm. Half-hourly results dump any moment now …

8.50pm. I’d say the VEC site is providing half-hourly updates, and we’ll get another blurt of results in about 10 minutes.

8.45pm. At North Melbourne booth, Stephen Mayne reports Labor got 32.5% of his own preferences, 92% of Nolte’s and 57% of the Sex Party’s.

8.38pm. Still to come: Flemington, Melbourne and South Kensington, and 11 of the 14 booths’ two-party counts.

8.37pm. Carlton Central and East Melbourne primaries added, and my projection is staying lineball.

8.31pm. The VEC has published 2PP results from three booths, which suggest my preference splits were exactly right after I made the adjustment just noted to Sex Party preferences.

8.28pm. After half an hour of silence, the VEC has just unloaded seven booths in one hit. Poor effort. My figures now align what ALP sources just told James Campbell. On intelligence from Stephen Mayne, I’ve adjusted Sex Party preferences from 70-30 to Labor to 50-50.

8.23pm. So the ALP has results from seven booths, but the rest of us only have two.

8.20pm. Sunday Herald Sun reporter James Campbell tweets: “ALP sources say vote it will come down to preferences but with almost half the booths reporting 1st preferences they are behind.”

8.17pm. Stephen Mayne reports East Melbourne booth primaries are ALP 466, Greens 436, Mayne 175, Sex 151, Nolte 144 – which suggests to me little or no swing, which would be an excellent result for Labor.

8.10pm. That RMIT booth has apparently gone 55-45 to Greens, which suggests a swing of about 7-8% – further encouraging the idea that it’s going to be close.

8.04pm. So in a nutshell, the Greens’ raw primary vote lead gets closed on my 2PP projection because a) the better performing minor candidates are preferencing Labor, and b) these two booths collectively were relatively strong for the Greens in 2010.

7.58pm. Twitter reports “catering situation at ALP HQ has improved”.

7.56pm. Keep in mind also I’m assuming 70% of those voting for minor candidates favour the party favoured on the how to vote card. The better performing candidates are tending to be those favouring Labor. If they show more (or less) independence than I’m presuming, the projection could be off.

7.47pm. Very similar swings in booth booths. Labor basically steady on primary vote, Greens up 6% and 4% respectively. Both booths broadly representative of the electorate as a whole as well, North Melbourne East a little above average for the Greens (remembering that the swing calculations take that into account).

7.45pm. North Melbourne East and Parkville booths added, and my word it looks tight …

7.40pm. Slowest count ever.

7.24pm. Conversely, more Twitter talk is of lineball results in Carlton, which is the Greens’ best area. Some actual results would be helpful …

7.21pm. Twitter talk is of 3% swing away from Labor and 7% to Greens – assuming this is off the primary vote, it points to a Greens win in the 55-45 vicinity.

7.13pm. Word on Twitter is that the Greens won the RMIT booth with 489 votes to Labor’s 300, which would be more than encouraging for them if so.

6.47pm. The fact that there are 16 candidates on the ballot paper might cause the count to be a little slower than usual.

6.25pm. Some further technical detail while you wait. Until booths begin reporting two-party preferred results, preferences will be distributed on the basis of 70-30 splits according to their how-to-vote cards, or 50-50 where no recommendation was made. When two-party booth results become available, the preference splits from booths which have reported two-party results will be projected on to the ones that haven’t.

6pm. Welcome to the Poll Bludger’s live coverage of the eagerly awaited Melbourne by-election count. Polls have closed, and the first results should be in in around three quarters of an hour. The table above will be display both raw and projected figures as the 14 booths progressively report. The first two columns will provide raw primary votes and percentages. The third “swing” column will show the primary vote swing for those parties which contested both this election and the 2010 election (Labor, Greens and Australian Sex Party), calculated by comparing the booths which have reported with the same booths at the election (which required some tinkering in one or two cases where booths have moved or are not being used). The two-party preferred swing will do the same. The latter will be compared against the total result from the 2010 election to project the outcome shown in the “2PP (projected)” column.

Author: William Bowe

William Bowe is a Perth-based election analyst and occasional teacher of political science. His blog, The Poll Bludger, has existed in one form or another since 2004, and is one of the most heavily trafficked websites on Australian politics. View all posts by William Bowe