Antony had a good post over on his blog today about how the current big polling leads Labor enjoys may not end up flowing through to a large election victory. My own view on the subject is that I’m completely and utterly open minded about the prospect – it’s worth explaining why.
If we look back over the last 10 years or so of federal and State politics, there’s been a number of examples where political parties have enjoyed large and consistent polling leads. In fact, there’s been five really strong examples where a party has completely dominated the polling for most or all of their term, and two moderately strong examples where a party dominated the polling late in their term. Of the 7 times a party has dominated the polling, on 6 of those occasions the party has gone on to get election results consistent with those large polling leads.
The only exception was Kevin Rudd and Labor in the 2007 federal election.
The other interesting thing to note is that when you just look at incumbent governments and their polling, of the six incumbent governments that experienced big polling leads, all six of them went on to win large election victories consistent with those big polling leads.
First up, we’ll take a look at the five strong examples of polling domination in Qld in 2001 and 2004, Victoria in 2002, South Australia in 2006 and the Federal election in 2007. We’ll track the polling on both primary vote terms and estimated two party preferred terms through to the actual election result using Newspoll data. I had to construct a two party preferred estimate for some of this polling (Qld, NSW and Vic) as Newspoll either didn’t measure or didn’t publish that estimate (click the charts to expand).
Next, the two weaker examples of NSW in 1999 and 2003, where large polling leads eventuated, but relatively late in the terms.
If, for these seven cases we have where large polling leads existed, we take the average of the Newspolls for the 12 month period leading up to the election and compare that poll average to the actual election results, we get:
The “Difference” measure is the election result minus the polling average. What we see is that the only case where the election result was substantially lower than the polling average was in the case of the last Federal election – which also happens to be the only example of an Opposition having a large, consistent polling lead anywhere in the country in the last decade or so.
So, did the Labor Party in 2007 have their large polling lead reduced to a small election victory because of the magic of federal political history, a magic that doesn’t exist at the state level – or did it occur because they were an Opposition without the campaign benefits of incumbency to keep the last minute swingers?
On the one hand, recent election history suggests that every government that enjoys consistently large polling leads ends up with a large election victory. Yet, on the other hand, longer term federal election history – as Antony pointed out – suggests that these large election results never or rarely happen. Which one is or will be ultimately correct?
My honest answer is “I don’t know” – which is why I’m completely open minded about the potential size of any ALP victory this year. Especially since every one of these governments that ended up with big victories off the back of big polling leads were Labor governments.
On something completely different, Crikey has a readers survey going to help further develop the Nature of Crikey <insert ominous music>. So if you read this place, it would be really appreciated if we could pinch about 5 minutes of your time and know your thoughts on what you like or don’t about Crikey and what you’d like to see more or less of.You can even win stuff.
35 Comments
Antony’s opinion could make sense considering the Essential Report results you blogged about recently. How can you interpret these contradictory results (i.e. very strong support for Labor generally cf lower support on specific metrics)?
Perhaps a number of respondants still retain the negative notion of ‘Howard = the Liberal Party’ on a superficial level but respond more positive when presented with specific current Liberal policies/positions? If – and this is may be a very big ‘if’ – the party can retain a sense of stability and remain a safe distance from Howard’s most polarising policies, this lingering negative sentiment of the Libs as a ‘brand’ will wane and their 2PP support increase?
Or are we all merely mindless weathervanes, swinging in the capricious political breeze…?
I won’t be disappointed if the rodents get a walloping.
Federally there haven’t been any landslides on the scale of those state landslides mentioned above for a very long time. So we don’t have any pre-federal landslide polling to compair with.
Well I guess whatever happens pollsters will look back at the 2010 result as a significant precedent.
There are factors pushing and pulling in each direction so it’s pretty hard to know for sure. Still I expect that there will be a small swing to the government (that in itself is remarkable for a 1st term federal gov) but not to Holt 1966 levels.
The contradictions in conservative ideology have rescently been laid bair by environmental issues such as climate change and they won’t have time to evolve before the election. CC could do to the Liberals what communism did to the ALP. Plus the demographics are slowely working against the Libs.
I’m in the don’t know category as well.
One caution about state/federal comparisons is there is a long history of big Labor wins at state level, but not at Federal level. Traditionally at Federal elections Labor was viewed as the Canberra party, which in part explains why Labor has a batter record in Queensland and WA state politics than it has ever had in those states at Federal elections. That dynamic may have changed in recent years as the Coalition has slowly forgotten about its states rights past.
There is also an argument that in the past Labor’s poor federal results were caused by Labor’s appalling record in Victoria until 1980, and relatively poor record in Queensland. It may be that now Labor is in government, and relatively popular, and improved in Victoria and Queensland, that those traditional upper limits on its national support may have fallen away.
People I imagine tend to fall back to what they are comfortable with the closer they get to election day and I reckon this is what saved the Liberal Party from a much larger loss.
Rudd got people to feel comfortable about him but there were still plenty who wanted to stay with that nice old man (as one colleague described him).
When we get the election day this year the big stand out issue will still be the economy. People will feel that Australia was hardly affected by the GFC and that the government had some large part in that. And with the GFC still blowing in the wind somewhat people will want to stick with the government and that t nice fair haired nerd.
I’m note sure that precedence can be relied upon too much, the world has been changing rapidly, the USA on the decline, China ascendant, things look different.
Rudd and Co will only need to seem as nerdy and boring as they did last time to keep their polling margin.
The ABC and right wing media will do all they possibly can to promote Abbott and the Liberal Party. But there is too much of a creepy sinister feel about Abbott for the public to have any warmth for him.
I would predict a narrowing toward the government come election day….meaning a widening of the poll margin.
Poss, I took out a crikey subscription deal for 2007 because I wanted to be in the know for the election.
It was fine, I guess.
Then I found this site and PB, and the whole thing shifted.
I realised that I could have my finger on the pulse far better by logging in here and at PB every day than by subscribing to Crikey.
I tried to log out of Crikey, but they wouldn’t let me. They reckoned I had changed my email, which I had not. There was no way to find out what the problem was. I gave up.
So I guess I could send Crikey to the eternal blackness of “report spam” on my gmail account, or just hit delete for all the crap about how I am missing out on important information. Not.
The weekly squatters edition I scan when it comes, and pretty much all of the time I am glad I never stumped up a subscription.
I just don’t get first dog on the moon, it seems from a universe far away. Or the moon.
I get far more from this site and PB than I ever did, or ever could, from Crikey. They just seem irrelevant.
TP@5:
That’s a brave prediction. I hope you are right.
Polling at the 2007 federal election showed a very late swing toward the coalition (Nielsen’s last rogue poll notwithstanding) and I reckon it was because of the scare campaign the coalition ran regarding interest rates and union domination of government (train stikes, petrol strikes, power strikes, teachers strikes, airport strikes, building strikes, dock strikes…did I miss any?). Anyway, it seems likely those scares won’t have much traction now and unless there is a good scandal or the mother of all industrial actions between now and August the election will be 55-45. Whether that counts as an electoral drubbing someone else can decide.
55? big call.
I’m also in the not-sure-yet category on all this.
Firstly Government and Opposition are not comparable; there is a tendency for election results to bounce back towards the government of the day compared to the polling if that government isn’t miles in front already. So any argument that the 2007 election result was a low-point compared to the polling on either side, therefore all the current polling is suss, doesn’t automatically hold up. It only raises a question mark, and a questionable one at that.
Secondly, any attempt to extrapolate now from the Government’s poll leads to the next election result on the basis of previous findings about election results of governments that dominated in their last 12 months in term has a bigger hazard than state/federal comparison issues. That hazard is that we don’t yet know what very much of the polling landscape for this government of that last 12 months looks like. If we, for instance, now see Labor’s polling slide into the 52s (2PP) and then Labor wins 53-47 or so, that will be consistent with the pattern, but it won’t answer the question that is being asked right now. That question is about the reliability of the high 50s and even low 60s polls we have been seeing up to now and that largely fall outside the 12 month window. Those polls may have been accurately measuring voter intention at the time, but voter intention changes, and polls taken more than 12 months out have relatively little predictive value when it comes to actual results.
Current Coalition cf. Labor under Calwell? I wasn’t around for Calwell but I think that one can be argued either way because they are completely different kinds of trainwrecks. The current Opposition has had three leaders, none of whom were elected decisively or convincingly, and is probably more ideologically confused than any other I can remember at federal or state level (or even perhaps overseas in major western nations) with the possible exception of the Tassie Liberals under Cheek (who got annihilated). Its best hope is that Labor will spend the rest of the term giving it free kicks.
Labor’s history of not dominating? This is only the third new Labor government we’ve had in a very long time and of the previous two Whitlam’s was never going to win any landslides, while Hawke’s 1984 election was held very early in the term and featured a massive blowout in the informal vote, without which factors Labor may well have won that election much more convincingly.
So I too am open to the possibility that what the polls are saying around now may well be right and that the Opposition could yet be smashed 55-45 or even worse. Or as it is still several months out from the election, there is time for it to all turn out to be nonsense and the Opposition to be vaguely competitive. That said I don’t trust the extra point or two Labor frequently gets in Morgan F2F or ER and am paying more attention to the rolling phone poll trend here as an indicator of where voter sentiment is at.
For the sake of testing my suspicion that the high polls will mean something if they continue, I will make and check a test prediction (though I will not be putting money on it). I predict that, subject to the following subjective fudge caveat, Labor’s 2PP vote at the next Federal election will be not more than 2 points lower than its phone poll average 2PP for the 6 months leading up to the start of the campaign. The subjective fudge caveat is that the test may be called off before the election if Labor’s campaign is marred by major scandals, extreme incompetence or very major unfavourable world or national events.
PS To throw in a Tassie datum (bearing in mind the major electoral system differences and the lack of 2PP) the first Bacon government held average leads of around 20 points over the opposition (undecided removed) in EMRS polling in the 12 months before the 2002 election, and outpolled that opposition at the election by 25 points. That is probably comparable to the state-level mashings by Labor governments that Possum cites via Newspoll.
Possum, Antony,
Someone suggested that the opinion polls do not accurately represent the Coalition vote because:
a) People who have signed up to the Do Not Call register are not contactable for polling
b) People on the register tend to be from higher income households, ie more likely to be Coalition voters
… thus excluding the effect of this element from the results.
Any validity to this theory? And if so, do the pollsters have a means by which to compensate for this effect?
Kevin, the 1984 increase in informal vote was caused by a surge in ‘1′ only votes induced by the new Senate voting system. AEC research counted all these votes out by party. If added to the party totals, it would have lifted Labor’s national vote by 0.35%. Checking seats by seat, it did not change the result in any seat.
Cuppa,
The DO Not Call register doesnt apply to pollsters.
Thanks (I think) Antony@9. Luckily I have no psephological credibility to lose by making such predicitions. I shall also take this opportunity to reiterate my predictions that Aston and Dunkley will go ALP at the next election.
I do think though that a lack of landlines could influence polling data.
Evan, I’m getting some stuff together for an article about that.
Thanks Possum. Another myth put to bed.
DrMick
If Labor dont get 100 seats, I’ll demand a Royal Commission into whythey didnt!!!
I’m on the Do Not Call register, and I still get calls from pollsters and charities. They’re the ones that are exempt. So there’s some first-hand evidence for you.
Re the poll figures: we can look at trends etc, but I would have thought the biggest factor would be the ways in the which the parties actually campaign, no? The numbers don’t exist independently of the issues. Obvious point, I know; but it does mean that any prediction we make right now could be swept away very quickly depending on what happens.
Possum allowed me to make a very confident prediction that Labor would win in 2007, so I do trust pseph forecasts to an extent. But IIRC, it was on the back of gaffe after gaffe by the Coalition. They regained traction late in the campaign almost despite themselves. I think Howard’s incumbency – 11 years of it – was alone the trigger for the narrowing. There were a lot of swinging voters who just felt ’safe’ under him, and moved back late.
The trouble this year is that that we don’t know whether Abbott will run a steady ship or go postal. My money’s on steady ship. But I’m not confident either way.
My first post – go easy on me!
Rubber gloves on for Aguirre!
I’m not on the Do Not Call register (must get around to it one day), but I don’t get polled either
… Actually, that’s not strictly correct: I did get polled by Morgan about eight years ago. I wasn’t ‘into’ politics then, so I don’t recall whether they asked questions of that nature.
PS Aguirre: Now you’ve made the first post, the ones to come will be a cinch!
Cuppa, I just whipped up this to explain why you (or most people for that matter) never get polled.
http://twitpic.com/10835c
I think comment #8 has it right why the 2007 federal election is the anomaly. The last weeks of the campaign saw a torrent of “Don’t vote labor – they’ll stuff our economy” advertising. It was a unsophisticated fear campaign and to some extent it worked. With that fear based, risk-of-the-unknown, better-the-devil-you-know stuff gone from the equation I think that the ALP will break records in 2010, just like the Victorian ALP smashed all sorts of records in 2002.
This time the ALP will be dishing out the fear, and with Tony Abbott in charge of the libs there is plenty of material to work with!
Wow, Poss, statistically a person would have to live for something like 300 years to be polled even once. It’s a bit like winning the lottery. I guess I was lucky then.
Cuppa: Even if the “Do Not Call” register applied to pollsters, the pollsters all use demographic weighting when they report their results. ie, if they find that a particular demographic (say, males from high income households) are under-represented in their sample for a poll, they’ll weight the responses from that demographic a little higher to compensate.
I get polled lots – as in, at least half a dozen times in the last ten years. So does that make me 2400 years old??
Oh, and I’ve been rejected by pollsters because I didn’t fit the demographic they were looking for.
Which, BTW, is how they make sure that they get the age groups represented reasonably – if they look at the people polled so far, and there is a demographic under represented (proportional to its numbers in the population as a whole) they will deliberately go looking for them.
So the phone call will start (after the usual blurb) with “Is there anyone aged between X and Y in your household?” and if there isn’t, they don’t ask you anything else.
OK, thanks. Looks like I got sucked in by Hawke’s excuse for not winning more comfortably!
As for polling, I get polled heaps! Mostly EMRS; I’d rarely go a year without getting polled by them at least once.
Zoomster and KB, do you happen to live in marginal electorates? That could make you more pollster-targeted. When I lived in a marginal electorate I got polled a few times, but not since I moved to a safe seat.
One reason for the continuing high TPP for Labor is the “apparently” high Greens Primary vote which is then distributed to Labor according to the pollsters preferred process.
A Greens Primary vote in the teens is something I will only believe when I see it replicated in an election.
Greeny, even when you look at the respondent allocated preference version of the two party preferred that Nielsen and Morgan both do, there’s no great difference. In the polls that only have the Greens around 8-12, the ALP primary just jumps as a result.
At federal level not even remotely (Denison is now so boring that it is about to be handed on a platter to some complete nonentity I went to school with) but then again most of the polling I get is state-level.
On the demise of landlines affecting the results of phone polls.. as Poss and others have pointed out, this would tend to actually be flattering the Liberals as the mobile only under 40s tend to favour the ALP/Greens, so artificially distorting things and making the Libs look more competitive than they really are.
While not being enamoured of the Holy Roller Rudd/Conroy Puritan Party Government, The only issue for the next election is the size of the ALP majority and whether they crack 100 seats and control of the Senate. A Lib win is just fantasy time.
I think there has been a secular shift in favour of Federal Labor by large swathes of the public – essentially in cohorts under the age of 55/60 – and this has been fairly stable for at least 4 or 5 years. This reflects, in the first instance, the inadequacies of the Liberal Party itself. Since Rudd’s ascendancy, it has also reflected the ability of Labor to control its messaging. If anything, they are getting even better at this as time goes on.
At the last election, the incumbency factor – especially the recurrent “fear of change” – worked in favour of Howard, who achieved a last-minute lift in support. This did not save him, but prevented an even bigger Labor win. This time, the fear factors will work for Labor and they can look forward to a very pronounced victory.
Labor can expect to win seats in nearly all States from a miserably out-of-touch and confused Liberal Party.