Wentworth by-election final result: Phelps 51.2, Sharma 48.8

With the last remaining postal votes counted, Kerryn Phelps emerges with a winning margin of 1851 votes.

There’s a fair bit going on on the blog right now, so for those of you focused on the top of the page, note the various threads below this one: the main discussion thread, Adrian Beaumont’s post on the US mid-terms, my comprehensive Victorian election campaign summary and the one where I plead for donations.

And now to the matter at hand: the last batch of postals have finally been added for the Wentworth by-election, for which my full accounting of the results can be found here. The late postals have broken 232-164 in favour of Kerryn Phelps, who emerges with a winning margin of 38,989 (51.2%) to 37,138 (48.8%), or 1851 votes. All that remains to be done is the formal distribution of preferences, which might yet turn up a few small anomalies.

Wentworth by-election live

Live coverage of the count for the Wentworth by-election.

Live publication of results, updated by the minute with full booth results and swings, can be found here. Commentary of the progress of the count follows below.

Thursday evening

The remarkable swing to Phelps on postals continues, with today’s batch favouring her 306-243. This means the progress of her two-party vote across the five batches has gone 36%, 40%, 45%, 52%, 55%. She also gained with the addition of pre-poll declaration votes, which favoured her 189-112. Her lead is now 38,757 to 36,974, or 1783 votes, which is probably more than the number of late arriving postals that constitute all that’s left to come. Phelps’ margin has ticked over from 1.1% to 1.2%, and looks sure to stay above 1%.

Wednesday evening

The dwindling daily addition of postals put a further nail in the coffin today, breaking 242-224 in favour of Kerryn Phelps. Her shares of the two-party vote across the four batches of postals that have been added to the count have been, consecutively, 36%, 40%, 45% and 52%. The provisional votes were also added to the count, and they were higher than usual in number and heavily favourable to Phelps, who received 246 to Sharma’s 175. Phelps’ leads is out from 1554 to 1643, with roughly 2000 postals and 250 declaration pre-poll votes to come.

Tuesday evening

The count continues to drift away from Dave Sharma, with today’s postals favouring him only 433-361, giving him 54.5% where he needs nearly three-quarters. Phelps’s current lead is 1554 votes, with maybe 3500 still to come.

Monday evening

Today’s counting consisted of finishing off the rechecking of ordinary votes. It appears 54 votes in the Paddington PPVC were reassigned from Sharma to Phelps; other than that, the effect was to cut 209 votes from Sharma’s total and 234 from Phelps’s. So evidently a lot of the action on rechecking has consisted of ruling informal votes that were initially admitted to the count. The upshot is that little has changed since yesterday, except that the window seems to have closed on a major anomaly being identified in rechecking, which was Sharma’s best hope. I had a fairly extensive look at the progress of the count in a paywalled article in Crikey today.

Sunday evening

Today’s events as they unfolded:

• Anxieties in the pro-Phelps/anti-Liberal camp that set in as the largest pre-poll voting centres recorded their votes late last night cranked up a notch in the morning as the first and biggest batch of postals were added to the count. These broke 3356-1858 in favour of Sharma (later revised to 3346-1851), his 64.4% share being fractionally more than he would likely need to rein in what remained of Phelps’s lead. Not long after, Antony Green wrote on Twitter: “No (Phelps) is not home. The difference between on the day voting and voting in advance is wider than I’ve ever seen at an election. In conversation with very senior party people today, they have the same opinion.”

• Around 10:30am or so, Sharma got a further small boost when the two Special Hospital Team booths were added to the total, collectively breaking his way by 266-54.

• After that though, the pendulum swung back. The AEC set to work on the routine recheck of the ordinary votes, starting with those booths where the preferences flows recorded yesterday appeared to be anomalous, as was keenly observed by Kevin Bonham. This confirmed that Phelps had indeed been short-changed in the Bondi Beach and Bellevue Hill booths – because, according to Antony Green, the preferences from primary votes for the Liberal candidate had been entered the wrong way around. As a result, Phelps’ 2132-1714 lead in Bondi Beach blew out to 2427-1330, and Sharma’s purported 1305-985 lead in Bellevue Hill turned out to be only 1152-1119. About half the booths have had their votes rechecked to this point, the net effect of the others being neutral.

• Then a second batch of postals went 698-467 to Sharma, or 59.9% to 40.1% – less than he would have needed even before the rechecking raised the bar.

The main votes yet to be counted are late-arriving postals – I see no reason to doubt my earlier judgement that the final number of postals will be very close to the 9392 that were cast in 2016, since the number of applications received was almost identical. That leaves maybe 3000 postals outstanding, along with provisionals and pre-poll declaration votes, of which a high-end estimate would be about 500. This leaves Sharma with about 3500 votes outstanding with which to close a gap of 1616, meaning he will need about 73%.

That’s not going to happen, so it will take the emergence of another error in what remains of the rechecking to make a Sharma victory plausible. Precedents do exist, such as the decisive 1000 votes that showed up for Cathy McGowan as she grappled with Sophie Mirabella in Indi in 2013. But if the remainder of the count proceeds normally, Sharma only seems likely to reel his existing deficit in by around 700 votes, giving Phelps a winning margin of around 900 votes, or 0.6%.

Sunday morning

To cut the following long story short: this isn’t over.

Those who were still paying attention at the close of last night’s action were thrown into a spin when Dave Sharma did remarkably well out of the pre-poll voting centres, which these days account for many thousands of votes and do not report their results into well into the evening. In particular, the 6431 votes of the Rose Bay PPVC broke 4473-1958, which slashed Phelps’ lead from 4.2% to 1.9% – creating just the slightest opportunity for Sharma to pull a rabbit out of the hat on postals.

In the last of my updates in the section below, I calculated that Sharma would need 70% of postals to close the gap, but it seems this was an overestimate. The number of postal votes issued at this by-election has been almost identical to that in 2016 (12860 compared with 12796), so it’s a very safe bet the number of formal postal votes will be around the same, namely 9329. We can also expect 500 or so provisional and pre-poll declaration votes, but it’s the postal votes that are most interesting because they skew conservative. Malcolm Turnbull did around 9.5% better with postals in 2016 as compared with ordinary votes, on both the primary and two-party preferred vote.

If that bears out this time, Sharma can expect to reduce his present deficit of 2590 votes by around 1400. However, it’s not impossible that he will do significantly better than that. Given the trend of polling and the general course of political events over the past week or two, it could be surmised he would do relatively well on votes that were cast earlier in the process. Kevin Bonham points out that the Mayo by-election is particularly auspicious for the Liberals in that Rebekha Sharkie gained a 3.5% two-party swing on ordinary votes against the Liberals, but there was actually a 5.1% swing in the Liberals’ favour on postals. No doubt this was unusual, but it does demonstrate that it would not be without precedent for postals to weigh towards Sharma heavily enough to sneak him over the line.

However, some objections have been noted to the results as currently published:

• The Rose Bay PPVC is an extreme outlier in having a primary vote swing of only 1.9% against the Liberals, where in every other booth it was in double digits (not counting the 16 votes cast through the blind or low vision telephone voting service), and this is not reflected in any unusual movement in the Labor primary vote. However, this very likely reflects the fact that the Rose Bay PPVC wasn’t in use at the 2016 election, and the “historic” vote totals provided by the AEC to facilitate booth-matched swing calculations (including those featured in my own results facility) were well off the mark. Specifically, the “historic” totals only account for 1459 formal votes, of which 983 were credited to the Liberals, compared with the 6431 that actually appear to have been cast. As such, I see no reason not to think Sharma was indeed being undersold in early assessments of the count, as it was not appreciated how much of the harbourside vote was locked up in the Rose Bay PPVC, waiting to be unleashed at the very end of the night.

• Probably more substantively, Kevin Bonham has identified curiously weak preference flows for Phelps at Bondi Beach and Bellevue Hill. These would be consistent with 450 votes that properly belong to Phelps having been wrongly placed in Sharma’s pile. If the imminent rechecking of votes indeed proves this to be so, the hill would look just that little bit too high for Sharma to claim. But as Bonham also notes, there could just as easily be other inconsistencies awaiting discovery that could tip the balance the other way.

Election night

Midnight. The last pre-poll voting centres tightened things up quite a lot – not quite enough for Sharma, but there won’t be much in it in the end. Phelps ends the night with a 1.9% lead, which would leave Sharma needing a more-than-plausible 70% or so of postals.

Continue reading “Wentworth by-election live”

Wentworth by-election minus zero days

One last report of internal polling for the Liberals suggests the situation for them is bad, though maybe not quite as bad as some reports this week have suggested.

Dawn breaks, and we have arrived at the end of the campaign without a single media-commissioned opinion poll. All those results from ReachTEL (which you can find listed on the sidebar of my by-election guide) were commissioned by left-of-centre political concerns; the other source of intelligence has been the purported internal polling provided by the Liberals in the name of scaring wavering voters back into line. The last such is from today’s Daily Telegraph – make what you will of the following:

Internal party polling, obtained by The Saturday Telegraph, shows the Liberals are faring better in the contest in Sydney’s wealthiest electorate than they have admitted, with a primary vote of 39 per cent, although this is still not an outright winnable position … Earlier this week, polling conducted by the Liberal Party collapsed to as low as 41 to 59 two party-preferred. However, tracking polling has improved since then.

If all goes well, I’m hoping to take a leap forward with my election night results reporting this evening, I having finally – I hope – wrapped my head around enough Python and Javascript to offer fully automated live results, updated by the minute from the Australian Electoral Commission media feed. For a preview of how this will look (with a few imperfections), I have retroactively applied the set-up to the final results from the Super Saturday by-elections in July: go here and click on one of the five electorate tabs.

Have a look down any given page, and you will see booth results in a nice and neat tabular form, with tabs to allow you to choose whether they appear as raw votes, percentages or swings from the last election. I flatter myself that this will be the handiest way of observing booth results as they come in: the AEC doesn’t publish them at all on its site on election night (unless there’s been a recent change of policy), and navigating Antony Green’s display at the ABC requires a bit more effort.

You will also note “win probability” gauges, copying the concept the New York Times made famous on the unforgettable night/day of the 2016 presidential election. This will only get a workout if the AEC publishes a Liberal-versus-Labor notional two-party count, as this will be the only way to allow for a swing calculation. If that’s indeed what happens, it will come with the fairly substantial qualification that the result will apply only if Liberal and Labor are indeed the last two candidates in the count, and it’s not the general view that that’s what will happen. One way or another, the facility should be regarded as experimental for the time being.

Wentworth by-election minus one day

Some account of how preferences might flow between Dave Sharma and Kerryn Phelps, if that’s indeed how it pans out.

Wentworth has clearly evolved into a three-horse race, with independent Licia Heath apparently having failed to gain much traction, despite some high profile backing. Liberal candidate Dave Sharma is all but certainly to lead the primary vote, but a lot depends on who finishes second out of independent Kerryn Phelps and Labor’s Tim Murray. So far as the excitement of election night is concerned, a lot equally depends on which two candidates the Australian Electoral Commission picks for the notional two-candidate preferred count. I would assume they will play it safe, follow tradition and conduct the count between Liberal and Labor, but we won’t actually know until polls close tomorrow. If they get it wrong, we will be relying on word-of-mouth accounts from scrutineers, which are sure to be widely propagated on social media, for even a vague idea of how things are playing out. That’s assuming the result is not so close as to make it unclear who out of Phelps and Murray ends up ahead – and this too will depend in preferences, on which we will likewise have no official guide on the night.

Continue reading “Wentworth by-election minus one day”

Wentworth by-election minus three days

As the Wentworth race enters the home strait, Liberal internal polling reportedly records a blowout lead for Kerryn Phelps.

Panic stations in the Liberal camp ahead of Saturday’s Wentworth by-election, according to latest reports:

The Australian reports that “leaked” (by which they mean “dropped”) Liberal internal polling has found Kerryn Phelps breaking out to a 55-45 lead, after being level pegging a week ago. Caution should always be taken with reports of internal polling, which is invariably selectively released even if not actually fabricated, and it is clear in this case that the objective is to scare potential protest voters back into line. Nonetheless, an apparent turn against the government is all too explicable given its performance over the past week or so, and is also consistent with the evident desperation of the Prime Minister’s thought bubble about moving the Australian embassy in Israel to Jersualem – an all too obvious pitch at the 12.5% of the Wentworth population that identifies as Jewish, the highest proportion of any electorate in the land.

• The reported movement in internal polling has been reflected on the betting markets, with Ladbrokes now has an unspecified independent a very slight favourite to win, from $2.75 at the start of the week to $1.70, while Dave Sharma is out from $1.33 to $1.80. Labor’s Tim Murray is out very slightly from $7.50 to $8. If you’re up for a bet, particularly a losing one, it would help me out if you signed up through the Ladbrokes advertisement on the sidebar.

• Evidently recounting complaints from the Kerryn Phelps camp, a report by Joe Hildebrand of the Daily Telegraph made overheated claims that a poll conducted for independent candidate Licia Heath a fortnight ago was an example of “push polling”. In particular, a question on Heath’s campaign for a new high school in the electorate prompted an unidentified Phelps campaign operative to assert, “if that’s not push-polling I don’t know what is”. It would seem the spokesperson does not, indeed, know what push polling is – properly understood, the term refers to efforts to disseminate false information under the guise of conducting an opinion poll. But the information in this case is not false, and the poll was very clearly a poll, even if it may not have been a particularly good one. Kevin Bonham notes that such a question may have contaminated voting intention responses if it had preceded it in the question order, but my attempt to clear this up did not elicit a response.

UPDATE: Now Greenpeace has produced results of ReachTEL poll that tends to confirm the picture painted by the reported Liberal internal polling – after allocation of a forced response question for the 2% initially undecided, the primary votes are 33.5% for Dave Sharma (Liberal), 26.4% for Kerryn Phelps (independent), 21.7% for Tim Murray (Labor), 9.2% for Dominic Wy Kanak (Greens) and 5.6% for Licia Heath (independent). Applying the remarkable results for respondent-allocated preferences, which finds over 90% flowing to Phelps, Phelps emerges with a crushing lead of 62.4-37.6. The poll was conducted Monday from a sample of 661. Respondents were also asked about their vote in 2016, and the results aligned fairly well with the actual result. Among the other findings were that a remarkable 66.0% agreed, including 54.0% who strongly agreed, with Alex Turnbull’s assertion that the Liberal Party had “been taken over by extremists on the hard right“. For perspective, the course of four ReachTEL polls to have emerged through the campaign has run like so:

Wentworth by-election minus one week

Reported internal polling from Wentworth suggest the Liberals have a nervous week ahead of them.

With one week to go, two polling snippets to launch a dedicated Wentworth post and discussion thread:

Andrew Clennell of The Australian reports Liberal internal polling shows Kerryn Phelps is “likely” to finish ahead of Labor’s Tim Murray, and that Phelps emerges “a fraction of a percentage point ahead” of Liberal candidate Dave Sharma after preferences.

• Both Kerryn Phelps and Tim Murray are credited with two-party leads of around 55-45 over Dave Sharma in an online poll conducted for Voter Choice, a research project by James Cook University doctoral candidate Kathryn Crosby. While the poll has a respectable sample of 736 and appears to use judicious weighting, the self-selecting nature of the sample warrants a degree of caution. UPDATE: Kevin Bonham has looked at this more carefully, and advises still more caution.