Canning by-election live

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

Two-party: full preference estimate

# % Proj. Swing
ANDREW HASTIE (Liberal) 36933 54.4% 54.8% -7.0%
MATT KEOGH (Labor) 30989 45.6% 45.2% +7.0%

Primary vote

# % Proj. Swing
Liberal 34291 46.4% 46.7% -4.4%
Labor 26544 35.9% 36.2% +9.5%
Greens 4479 6.1% 6.0% -1.4%
Palmer United 2264 3.1% 2.3% -4.6%
Family First 531 0.7% 0.8% -0.5%
Australian Christians 2280 3.1% 2.8% -0.3%
Others 3519 4.8% 5.1% +1.5%
Formal 73908
% of enrolled voters 65.5%
% of projected turnout 89.2%
Booths counted (of 46) 46

Two-party: raw count

# % Prefs Pref. Swing
LIBERAL 40600 54.9% 48.3% -0.6%
LABOR 33307 45.1% 51.7% +0.6%
Formal 73907
% of enrolled voters 65.5%
% of projected turnout 89.2%
Booths counted (of 46) 46

City of Armadale booths

# % Proj. Swing
LIBERAL 15490 49.2% 49.2% -9.3%
LABOR 16025 50.8% 50.8% +9.3%
Booths counted (of 18) 18

City of Mandurah booths

# % Proj. Swing
LIBERAL 10868 58.3% 58.3% -3.3%
LABOR 7765 41.7% 41.7% +3.3%
Booths counted (of 10) 10

Other booths

# % Proj. Swing
LIBERAL 10575 59.5% 59.5% -6.7%
LABOR 7199 40.5% 40.5% +6.7%
Booths counted (of 18) 18

Concluding summary

In political terms, the result is anti-climactic, in that the swing is neither a triumph nor a disaster for either side. However, it’s interesting to note that the swing was concentrated in Armadale, given that a lot more suburban territory cut from the same socio-economic cloth sits beyond the electorate’s northern boundaries, in Hasluck and the newly drafted seat of Burt, both of which are marginal Liberal. Another standout factor from the results is that the non-major party vote was down on the last election, contrary to the usual by-election form. The drop in the Greens vote can only partly be explained by competition from Animal Justice and the Pirate Party, and the loss of two third of the Palmer United vote wasn’t entirely matched by an increase for other right-of-centre minor parties. Contrary to the indications of some polling, Labor’s share of preferences was essentially unchanged.

Live count commentary

10.15pm. Results above are final for the night, there evidently being no counting of postals this evening. The results include two pre-poll voting centres – Armadale, which is included in the booth-matching calculations as the centre was in use at the 2013 election, and Mandurah, which isn’t and wasn’t.

9.01pm. The Armadale pre-poll voting centre — which along of the PPVCs I’m including in my booth-matching results projections, since it’s the only one that was in use at the 2013 election — has pushed the Labor swing over 7%. It will be interesting to see if there’s a general pattern of Labor performing better on pre-polls, since part of the vote will have been cast before the leadership change. Conversely, this may just be the oft-cited Armadale swing in action.

8.44pm. Two more results added, typical of the whole in being a strong result for Labor from Armadale and a weak one from Mandurah.

8.41pm. Still to report: two large booths at Halls Head in Mandurah; two Armadale booths, plus two others are yet to report two-party preferred; and in the vicinity of Armadale, full results from Harrisdale, and two-party from Byford.

8.32pm. More Armadale-heavy data has added a further increment to the Labor swing, although outstanding booths from Halls Head in Mandurah may well rein it back.

8.27pm. The addition of some further results from Armadale, where the swing is approaching 10%, has added half a point to the projected Labor swing.

8.22pm. I’ve been progressively adding newly reporting booths without comment, there being no surprises.

8.02pm. A very soft result for the Greens, who have gone backwards, and the minor party vote in general. Modest though it may be, Labor has at least achieved its two-party swing off its own steam.

8.00pm. Contrary to my expectations, the results have come in at a bewilderingly fast clip. But the swing has stayed around 6%, and the impression of a swing strongest in Armadale and weakest in Mandurah has been consistent.

7.49pm. I strongly suspect the AEC has the Labor and Liberal vote in Boddington entered the wrong way round. Otherwise there’s been a 40% swing there.

7.42pm. Five booths in from Armadale and now and three for Mandurah, and so for the swing in the latter looks to be the smaller. Despite that, the swing projection has remained been pretty stable at around 6%.

7.32pm. Fair bit of surgery required there on my results display, but I think it’s in order now. The swing Labor looks to be headed for is about 6%. It doesn’t seem that their preference share has picked up at all.

7.18pm. A fairly solid 11 booths in now, and I’ve had the swing at around 8% for a little while now. Antony Green only says 6%, but he’s less aggressive than me in extracting projections from incomplete results.

7.10pm. Three booths have been added on the 2PP vote, and these results are now being used to project preference flows on to booths that have reported the primary vote only.

7.05pm. A bunch of booths in now, including a couple from the larger centres, and the swing is looking stable so far — but I’m still going off 2013 election preferences. I’ll sort that out in a minute.

6.53pm. Carcoola and Dandalup North booths added on primary vote — so still only booths from the smaller semi-rural centres.

6.35pm. Still a lot I’m trying to work out here, but we’ve got primary votes in from the Preston Beach booth, and from that very small data point I’m projecting a result in line with expectations.

6pm. Polls have closed, so welcome to the Poll Bludger’s live coverage of the Canning by-election count. The table below will be updated as the numbers are reported to track both the raw vote and projections based on booth-matching and preference trends. With 12 candidates in the count, progress might be a bit slow. There are some small semi-rural booths in the electorate, and we should be hearing from these first in about an hour.

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.

313 comments on “Canning by-election live”

  1. Postals mostly added now. They’re actually swinging slightly more weakly than the booth votes, though the Liberals did make a big effort on them.

  2. Damo @296

    Animal Justice and Pirate Party debuted at 2.2% combined. I didn’t see their preference tickets, but I can imagine how they’d go.

  3. Thanks Kevin. So the pre-poll votes look almost identical to the poll day votes in the 2 major centres. I guess a number of factors here but certainly suggests not much Turnbull effect.

  4. Actually it could be that the Malcolm effect was negative.

    In 2013 the Libs did better on postal votes compared to ordinary votes on polling day by 5.65%.
    This time the Libs did better on postal votes by 6.49% compared to polling day.

    They did better when Abbott was PM -presuming a significant number of postal votes were cast when Tony was still boss.

    The reverse applies to the ALP.
    In 2013 their postal votes and ordinary votes were about the same – roughly 26%.
    This time their postal votes were much lower – by 3.88% – than the votes cast when Mal was #1.

    Significant?

  5. I doubt it. The numbers look like the 2014 WA Senate re-run and the NSW state election. The Liberal Party actively campaign for postal votes where Labor has tended to put much less effort into postal votes. Half of the postal vote applications in Canning were forwarded to the AEC by the Liberal party.

    This encourages more Liberal voters to use postal as opposed to pre-poll voting, which shows up in the postal vote party shares. Because of compulsory voting, this sort of campaigning shifts the time and place of voting rather than change the number of people voting.

    The postal vote application numbers on the AEC website list when applications were received and dispatched, not when votes were completed and received back. Ballot papers are removed from declaration envelopes before counting, so it is not possible to tally party vote shares by when votes were cast, unless you were a scrutineer and able to know something about the batching procedures of how votes are bundled and counted.

  6. It looks like both major parties made a bigger postal effort in Canning for the by-election than in 2013. In 2013 Coalition party postals were 38% of all postals and Labor party postals were 0.1%. In 2015 as of Thursday, Coalition party postals were 48% and Labor party postals 11%.

  7. This seems to have gone a bit quiet but Tony Abbott is now going around saying that the by-election would have been won with him as leader and that’s why Malcolm Turnbull moved against him during the by-election.

    It seemed to me that it would be easy enough to guesstimate what difference Turnbull made by comparing results on early votes with results on ordinary votes between 2013 and this by election (primary votes, not two party preferred). By my calculations the swing against the Libs on early votes was 7.55 and 4.31 on ordinary votes. So Ok they still would have won on a 7.55 swing, but it would have been a much bigger swing if that had been sustained to ordinary votes, so it looks like Turnbull did cause an improvement.

    I haven’t seen any psephologists refuting what Tony Abbott is saying, sorry if there was and I missed it, but I think it really needs to be emphasised, because unfortunately it looks like Abbott is up to his old tricks of ‘misrepresenting”.

  8. Valerie Kevin Bonham basically disputed Abbott’s 57/43 on the main thread earlier today.

    Based on all the external polling the Liberals were running at around 51/49 prior to the leadership change so any internal polling prior to the change coming up with a 57/43 result must have polled a very select group.

  9. Apart from the Newspoll of August 15/16 before the campaign proper started, the polls pre-Turnbull only showed a close result by using questionable preference allocation. I made comments about it, as did Kevin Bonham and William several times

    William wrote:
    those who have been following this thread will be aware that ReachTEL’s two-party totals have been reached through surprising Labor-friendly numbers on respondent-allocated preferences, such that you would want to see primary vote numbers before analysing them too carefully – and so far, no such numbers are available.

    Now that we have actual preference allocation, it would be interesting to see how misleading the published client commissioned ReachTEL polls were in their published final 2PP. Instead of a close result the 2PP would have shown an easy Liberal win under Abbott or Turnbull. As Kevin Bonham points out on his blog the vote share of pre-polls and postals did not vary much from the final booth results and therefore do not show the talked about shift in voter intention before and after Abbott.

    I am part of Canning and my impression was that with the blow-ins trying very hard to make this election a Abbott referendum, the Canning electors were pushing back and were determined to make this a election about candidates. In Canning it has been about the candidate for some time. How else did Randall manage to win booths like Westfield which vote 2 to 1 Labor in the state election? Armadale is one of the safest Labor state seats, yet Randall won all booths in 2013 except the relatively small Brookdale booth.

    The ReachTEL client commissioned polls (I answered all except one of them) are a very clear lesson again that client commissioned poll results should be treated with utmost caution. Many questions are asked in different ways. The results which are published, although undoubtedly accurate, are selective results. One early poll, which I suspect was from the unions was never published, although selective results were, having reportedly “fallen of the back of a truck”. ReachTEL itself should be concerned that by providing its clients with only the raw figures, allowing the clients to calculate their own 2PP by friendly allocation of undecided and preference flows could reflect on ReachTEL’s brand. What would have been shown to be dubious published headline results was masked this time by the change in Liberal leadership.

    I am certain the final result would have been 55/45 regardless. This is what the polls DID show if they had been analysed by competent phepsologists like William and Kevin. The client calculated undecided and preference flows were rubbish.

Leave a Reply

Your email address will not be published. Required fields are marked *