As my first post today I thought I would summarise the state of play in the House of Representatives.
As of last night I had called 141 seats as part of my analysis at the Guardian. This includes 72 Coalition seats and 63 Labor seats, as well as six crossbench seats – the five won at the 2016 election plus Warringah.
I have three other seats which I was close to calling but hesitated because there is a lot of pre-poll votes yet to report: Eden-Monaro, Hunter and Swan. Examining these seats this morning it looks likely Hunter will go to Labor while the other two are close (one going each way at the moment). If you assume these three go to the party in the lead that’s a total of 73 Coalition and 65 Labor, leaving you with seven seats in play:
- Bass – LIB by 321
- Boothby – LIB by 1311
- Chisholm – ALP by 325
- Cowan – ALP by 1575
- Lilley – ALP by 1323
- Macquarie – ALP by 620
- Wentworth – LIB by 801
If these seats all break to the party currently in the lead this will give the Morrison government 76 seats – the slimmest of majorities. They would need to rely on the casting vote of the speaker unless they made other arrangements with the crossbench.
Best case scenario for Labor would see the Coalition reduced to 73, with Labor on 71, and with a seventh crossbencher with Kerryn Phelps winning Wentworth.
Best case scenario for the Coalition would see them win 80 seats, which would be a comfortable majority.
The middle scenario would be almost exactly the same as in 2016. Labor and the Coalition would win the same number of seats, with one more crossbencher. Which is remarkable considering the last three years.
I’m sure I will touch on this in more detail later, but it’s worth noting that there has been another slight increase in the vote for minor parties and independents. These candidates polled 23.47% of the primary vote in the lower house in 2016, the highest ever figure for candidates outside the major parties. On current figures minor parties and independents are polling 24.73%. Despite almost one quarter of the electorate voting outside the major parties, they will collectively hold over 95% of seats in the lower house.
In addition to the seats listed above, I count eight seats that have changed hands. This includes Corangamite and Dunkley which were redrawn as notional Labor seats and have been won by Labor. Apart from these two redistribution gains, the only Labor gain so far is Gilmore. The Coalition has gained Braddon, Lindsay, Longman and Herbert, while independent Zali Steggall has gained Warringah.
I’ve produced the following map which shows the swing on two-candidate-preferred vote (except for the eight seats where we don’t have such a figure). You can toggle to see the 2CP vote for the winning candidate, except for the five seats where we have no 2CP count at the moment. I have grouped together independents, Greens, One Nation, Katter and Centre Alliance as ‘others’ – only in Kooyong was there both a strong independent and Greens candidate, in that seat the ‘other’ refers to the Greens.
While the Coalition did generally gain swings, there was lots of places where the reverse happened. Of the 133 seats where we have a Labor vs Coalition two-candidate-preferred count, the Coalition gained a swing in 85 seats, and Labor gained a swing in 47 (there is currently no swing in Tangney).
There is a clear difference between the trends in each state. Almost every seat in South Australia and Queensland swung to the Coalition (barring 3 Queensland seats and one SA seat). About two-thirds of seats in NSW swung to the Coalition while two-thirds of seats in Victoria swung to Labor.
| State | Swing to ALP | Swing to LNP | Unknown |
| NSW | 13 | 27 | 7 |
| VIC | 21 | 12 | 5 |
| QLD | 3 | 25 | 2 |
| WA | 6 | 9 | 1 |
| SA | 1 | 6 | 3 |
| TAS | 2 | 2 | 1 |
| ACT | 1 | 2 | – |
| NT | 0 | 2 | – |
On a final note, I should touch on the polls.
I don’t spend a great deal of time analysing polls, but I think it would be a mistake to treat this as a complete shock from a polling perspective.
If you look at the Bludgertrack polling average there had been a clear trend of polls drifting back towards 50-50 since the collapse in support for the government last year.
The final polling results were about 2% off the final outcome, which is within the range of the margin of error. We don’t have anyone who does the kind of probability-based modelling that we see from outfits like FiveThirtyEight, but a sensible model would have given the government a sizeable chance of winning, if not a majority. In that way the result reminds me of the Trump victory in 2016 – the polls were not significantly off, but expectations were such that the result was a shock.
One question I’ll be interested in answering is how well the seat polls performed. I haven’t looked at them in aggregate, but I definitely noticed a trend of seat poll results which seemed incompatible with the national polls. I think most of us assumed it must have been the seat polls that were wrong, but now I am wondering. Still, that’s a topic for another day and for others who have more expertise in polls.
That’s about it for now. I’ll have more analysis to come on the close races in the Senate and the House, as well as a podcast, over today and tomorrow.
The polling was not fundamentally wrong. Labor were the genuine front runners for over three years. But their momentum was lost at the most inopportune moment, right before the finish line. Most polls did actually demonstrate this in some way or the other. Ipsos on Thursdays had ALP on a 33 per cent primary vote. This appears to be precisely where Labor has landed. News Poll tightened substantially from 53-47 TPP to 51-49, (barely) within the margin of error.
I think Ben has hit the nail on the head when he said that “the polls were not significantly off, but expectations were such that the result was a shock.” After three years of Labor winning poll after poll, people will of course expect Labor to win. But expectation is not the same as outcome.
Labor will spend the next several months analysing what went wrong. I think there are a combination of factors. Their campaign was frankly, very underwhelming, especially compared to the 2016 “Mediscare” campaign. And Labor’s tax reform agenda was perhaps too difficult of a sell for an opposition seeking government. But they were also facing some very strong headwinds. These included an $80 million Clive Palmer war chest, and very unfavorable press coverage by the Murdoch media.
Ben if the polls weren’t wrong then what data were the bookies basing their odds on? Sportsbet and Tab were offering odds of 6 to 1 for the Coalition on Election Day and on Betfair you could get odds of 9 to 1, and those odds had been increasing as we got closer to the election. Every poll had Labor ahead and so you can’t simply look at individual polls and say that because the results of each poll were within each poll’s margin of error, there wasn’t any systematic polling errors.
Just touching on the polls. When something consistently misses the mark, as polls have increasingly done, one has to look at issues with underlying assumptions as a possible cause. As hard as they try to get an even demographic, pollsters are still reliant on people not hanging up as soon as they realise it’s a poll. The most likely demographic who would do this would be the less educated Conservative voter (conservative as in how Abbott, Hanson et al would define it, not the Menzies ”conservative’), hence polls would underestimate those votes, which has been reflected not just in the recent election, but also the recent NSW state election, Brexit and even Trump. On the other hand, in electorates such as Waringah where voters are more politically aware and, well, educated, the response rate is likely much higher and the polls are much closer to the reality. Given recent polls vs election results, a possible ballpark figure for this ‘correction’ could be between 2-5%, which would be very significant with anything other than a landslide result.
Maybe margin of error is not the right term for what I mean when I say the polls were “not significantly off”. I more mean that a systemic error of the type that has happened is not unheard-of and we should be factoring something like that into our judgement when the race is so close (which is what 538 does).
Much will be said in the washup about pollster ‘herding’ over the last weeks of the campaign when I think a more sober analysis down the track will focus on how ‘minor’ players are presented and whether the assumptions made about their preferences have been accurate or otherwise. An interesting stat for me, which I would appreciate a pseph looking into at some point, is the inordinate high H of R informal vote in some ‘safe’ labor seats. Examples, Fowler, McMahon, Blaxland, Chifley where the labor incumbent suffered a swing on primaries of about 6% and yet the informal is 10-13% which is way higher than the 3-4% I’m seeing in most seats. Tying into this is Greenway where Michelle Rowland had only a relatively small swing against her and the informal vote in that seat is also relatively smaller (still 6.6% though).
Perhaps in many ways the pollster methodology was correct, but the support for labor was much softer than could be detected by simple quantitative analysis and when the punter came to cast his or her ballot Morrison’s unrelenting attacks and labor’s overall poor campaign came home to roost.
Is there any chance of any big changes with 30% of the vote still to be counted in most seats? It it normal to have this much vote not declared the day after? And since these are mainly pre polls could they swing very differently?
Shy Tory factor? Possibility? This looks allot like 1970 UK general election, Where the polls Favoured Labor, But the conservatives ended up winning a majority. 1992 too. It seems people are afraid to tell the pollsters that they are backing a controversial government and say they are voting Labor, When they really are not.
The opinion poll companies my need to regroup or stick to deal with surveys on consumer goods. As one TV commentator said last night opinion poll companies rely on landline phones as the know were the number is roughly located but it does not work with mobiles which could be anywhere. Also why try and predict the result?
Voters have not forgotten Shortens hand in undermining Rudd and Gillard and other voters the same with Abbott with his sniping behind the scenes too.
After many results are know we still have Dutton and Hunt alive political. Watch you back PM Morrison.
Shy Tory factor doesn’t explain the Victorian state election miss last year, which seems to be corrected. More like a problem adapting to new methodologies
PS,
Pre-polls and especially postal votes tend to favour the conservative parties, so it’s more likely that a small Labor lead will be overcome by the Liberals (e.g. Chisholm, Macquarie) than the other way around.
Absolutely agree. Labor’s small lead in some of those undecided seats will be eroded. I wouldn’t be surprised if the Coalition end up around 78-79.
If all current seats stay as they are on the ABC, (Wentworth,Boothby,Bass for the Liberals) and Labor (Chisholm,Cowan,Macquarie and Lilley) It will result in 76-69, The exact same result as last time
I wonder about the impact on polling responses of broader public debates on social issues and indeed our whole mode of public discussion now, frankly. This might seem fanciful but I’m tentatively moving towards thinking that the modern social media environment is creating a kind of ‘communications dissonance’ between what people will say out loud, even anonymously to a stranger on a phone, and what they might really think. Shy, indeed. If a pollster asks you ‘how important is immigration to you compared to the economy as a voting issue’, I’d be cautious about their answer. Lots of people who put ON and UAP first in Qld might tell you they did so because of ‘Adani’ or ‘franking credits’…and that’s certainly how the analysis will go. But is it so, just because a voter says it is? Is it really? I just reckon – hunch is all – the problem with polling now is a lot deeper than an outdated methodology. People are profoundly guarded about moral issues now, I think.
I don’t even really mean ‘social issues’ or ‘moral issues’ in any specific sense (only one or two electorates in Qld voted against SSM for example). I think it’s more a general ‘moralising vibe’ nerve kind of deal (cf. the ‘hip pocket’ nerve of pollster lore). A deep-seated qualitative aversion to being hectored from a moralistic platform that jars too much with your own. And that’s hard to pick up in any poll, precisely because even asking the kind of question designed to assess it tends to set it off to defensive protective mode. I’d suggest conservative-inclined but basically decent people are unable to/cautious about even articulating it to themselves. But especially to a stranger they might regard as part of the ‘elite political bubble’…which pollsters are.
Just a fumbling thought…
I agree Papillons. I remember a comment by Hillary Clinton to this effect during the 2016 US presidential election. Clinton said during the campaign that she has “both a public and a private position” on Wall Street. As it turns out, voters also may have public and private views on politics and other inviduous issues. The only time voters can ever express those views with complete anonymity is at the ballot box. I think part of the problem might be due to the lack of mature debate and genuine discussion out there. Rather than encouraging an open dialogue, our culture has a tendency to simply shame or ostracise people whom some disagree with. I’m definitely not saying we should start empathising with every neo Nazi imbecile out there (cue Fraser Anning), but people in a civilised society should be free to express an opinion without being judged
Ben Once again Tally Room has played a very useful part in the election process. It remains one of the very few sites where political discussion takes place in a civilised manner. It is amazing what the positive effect an editor has.
I was surprised by result. I thought ALP were going to win and have very mixed feelings.
I want Tony Abbott Social policies with Menzies economics . Bill Shorten’s industrial policy would drag us in right direction providing the Senate can apply the brakes.
Wine Diamond thanks for positive feedback Daniel and Adrian you have made Tallyroom interesting. What happened to Benee?
All parties should now be commencing a review of what went wrong.
ALP review after last election was a relatively public document which was a positive for democracy.
Tony Abbott accurately described this as a nasty campaign. Corkscrewing was a new low in Australian politics.
My suspicion that someone with a Corkscrew in pocket at midnight was probably putting up Corflutes . I have used an old corkscrew for holing Corflutes. If this is a political activist the incident amounts to terrorism.
Whilst legally not the case I see a difference between daubing one’s opponents Corflutes with a political message and daubing same Corflute with an obscene abuse.
Palmer’s $50-$80 Million cash flash must not be allowed to occur again.
Where the hell did Anning get the money he spent? I will be looking and taking boycott action if I can. I will never use Wotif due to their funding of Greens. I disagree with BDS but not their right to Boycott. Sorry Adrian some of us support Israel.
What went wrong with polling? The statistical analysis discussed elsewhere may be of value to pollsters but it is of little use to those of us who are Poll users.If it is Landline phones that are the cause of the problems,
Pollsters may have to return to home visitation with clipboard questioning. If it is herding then pollsters do not believe their own polls. I personally have trusted polls especially old Bulletin magazine polls and In recent years Newspoll.
Agreed Bugsy.
I think Max Koslowski has been on the right track for some time on this. Food for pollster thought.
https://www.theage.com.au/federal-election-2019/the-expert-who-predicted-trump-brexit-and-scott-morrison-20190519-p51owf.html
Re: my last – of course I meant Professor Stantic (see article). Apologies, Ben…a long night!
Some TV commentators were gloating on Sunday that UAP won no seats but it was not necessarily about seats. It was about the UAP and other smaller parties like ON preferences going to the Coalition and that’s what happened. That’s what so called preference deals are about not UAP etc preferences going to the Coalition.
Opps – the last sentence above from me should read “That’s what so called preference deals are about not Coalition preferences going to UAP etc”.
Rebekah Sharkie for Speaker?
She gets a significant payrise and Libs have a workable majority.
I think it’s almost certain that the Coalition will be returned with a majority of 76-78.
Boothby & Wentworth have been called by the ABC leaving them just 1 short, which I suspect will come from a combination of Bass, Chisholm and/or Macquarie.
Quite the result, to not only go against virtually every single poll released since 2016, but to actually increase their majority and get a swing towards them, the Liberals owe Morrison the whole term.
Is Hunter still doubtful with a slim chance of the nationals winning or One nation?
Comments are closed.