Click here for full display of Rockingham by-election results.
Final result (August 5)
Quicker than I would have expected, the WAEC has conducted its full preference count. My projection was correct in suggesting independent Hayley Edwards, who polled 15.9% of the primary vote, would overtake Liberal candidate Peter Hudson, on 17.7%, to make the final count. This she did with 4823 votes (22.1%) at the last exclusion to Hudson’s 4590 (21.0%). The final result was 13,412 (61.4%) to Labor’s Magenta Marshall and 8443 (38.6%) to Edwards, compared with the 10.8% winning margin projected by my preference estimates. The media feed still records an abandoned TCP count between Marshall and Hudson, which is reflected in what’s on my results page.
Live commentary
11pm. I have corrected a preference model that was significantly flattering Labor in its projected lead over Hayley Edwards, which I now have at 10.7% rather than 16.1%. Edwards rather than the Liberal candidate is projected to make the final count because the preference model has her ahead 23.1% to 21.9% at the final exclusion. Happily, the WAEC’s notional two-candidate count means the Labor-versus-Liberal two-party preferred is recorded for posterity, at least up to the present point in the count. For all the contingent factors that undoubtedly contributed to the 22.5% swing, it is notable that it exceeds Labor’s statewide winning margin in 2021.
8.31pm. The WAEC feed doesn’t seem to have updated for about 15 minutes, which might suggest that’s it for the evening.
8.24pm. My results page is updating again after a bit of a blip, and the two-party swing against Labor is now north of 20% — higher than they would like, but not a disaster of the scale intimated by the opinion poll. That’s if the Liberal candidate indeed finishes second ahead of Hayley Edwards, which remains unclear. In any case, Labor’s Magenta Marshall has at all times looked to be in the ballpark of 50% on the primary vote and in zero danger of actually losing.
7.56pm. The WAEC is now letting its TCP results into the wild, which show what I reckon to be a 17% swing against Labor. This would be a bad result for them under normal circumstances, but it may be noted that the Utting Research poll recorded a 24% swing, without the loss of a Mark McGowan personal vote being a factor as it presumably is here.
7.52pm. Types rows have been restored to my results table.
7.30pm. A huge 5037 votes that I thought were going to be counted under “Rockingham EVC” have been added as “Early Votes (In Person)”. That they don’t have their own line on my page at this stage due to the bug discussed below is disappointing, but the results are entirely in line with the norm in having Magenta Marshall with almost exactly half of the primary vote.
7.28pm. My probability gauge isn’t quite doing what it should be — “probability unavailable as preference count unclear” should only be appearing if the distinction could conceivably make a difference to result, and it is clear Labor will win regardless of whether Hayley Edwards or the Liberals make the final count.
7.22pm. Eight booths now, but the overall trends are so clear that distinctions between them individually are pretty fine.
7.15pm. The special hospitals, remotes and mobiles has reported, so I was wrong to say earlier that the “types” rows wouldn’t be needed this evening. This amounts to 129 formal votes, the results of which you can work out by comparing the difference between the “polling booths sub-total” and “total” rows.
7.12pm. The early voting centre has reported with only 688, which I’m puzzled by because nearly 11,000 early votes were cast and most of them would have been there. Perhaps the result there will be updated progressively. Since my projections assumed as much, they are now out of whack.
7.07pm. Five booths in on the primary vote, and either there are no TCPs reporting yet or the WAEC is sitting on the results because it’s not sure the Liberal candidate will come second.
7.05pm. The WAEC’s TCP count, for which we have no results yet, is Labor versus Liberal.
7.02pm. I’ve done an ugly fix on the big issue by removing the “types” rows, which we won’t be needing tonight anyway. Four booths now on the primary vote, and while Labor has dipped below 50% on the raw primary vote, I’m projecting them to make it just over when all the votes are in. Nothing to separate Hayley Edwards and the Liberal candidate for second, but this will ultimately be academic.
6.58pm. Making progress on the bug fixes, but not all the way there yet. Now we’ve got three booths in, and the projection continues to be for a Labor primary vote in the low fifties.
6.53pm. We have a booth in, and while my system has a bug of some description, the result you can see is Bungaree Primary School, where the swings are consistent with Labor scoring slightly over 50% of the primary vote.
6pm. Polls have closed, and with that the start of a nervous wait for both candidates and those of us who are hoping our live results pages will function smoothly. For the time being my set-up assumes the WAEC’s notional two-candidate count will be between Labor and Liberal, but it can’t be ruled out that they will have gone with independent Hayley Edwards rather than the Liberal. The WAEC has unusual practices on this count – if memory serves, it only publishes the results once it is satisfied the candidates it has picked are the correct ones. My own projection will make up its own mind as to who the leading candidates are, and will work off preference estimates unless and until the WAEC has numbers for the same pair of candidates as those determined by my system.
Saturday morning
Today is the day of the by-election for the Western Australian state seat of Rockingham, arising from the unanticipated retirement of Mark McGowan. Here I offer my customary overview of the by-election; here, my coverage of a state opinion poll this week that greatly increased the level of interest surrounding the by-election; and here, if you’re a Crikey subscriber, my account of the potential federal implications of a collapse in Labor support in Western Australia, if that’s indeed what we’re seeing. This post will supplemented with live coverage of the account from 6pm this evening, and the site will as usual feature its famed live results, offering projections, probability estimates and a booth results map.
The West Australian is reporting that Rockingham deputy mayor and independent candidate Hayley Edwards is emerging as an outside chance, after a straw poll it conducted at pre-polling found 22 out of 73 respondents saying they had voted for her, compared with 34 for Labor candidate Magenta Marshall (leaving only 17 for others, among them Liberal candidate Peter Hudson). Another report in the paper today breathlessly relates a late change in Labor’s how-to-vote card has moved Edwards from third to eighth, ahead only of obscure independent Peter D. Dunne. This is sold as a “sign the ALP is worried the election for the long-held and uber-safe seat will be decided on preferences”, though the logic behind this is unclear.
It is one thing for a result to be “decided on preferences”, which will happen in the seemingly plausible event that Labor’s primary vote falls from the 82.8% Mark McGowan received in 2021 to below 50%. But the preferences of those who vote Labor will only enter the equation if its candidate fails to reach the final count, which would require what Kevin Bonham described yesterday as “the worst result in the history of everything”. It should be noted that the change in Labor’s how-to-vote card is a response in kind to Edwards, a former party member whose own how-to-vote card equally has all comers but Peter D. Dunne ahead of Magenta Marshall.