Today, Newspoll via The Oz (and a spiffy new flash app) released their quarterly aggregation of polling that provides us with state and demographic breakdowns for the first 3 months worth of Newspoll data in 2009. Since we’ve got 32 charts (!!) to get through, it might be wise to break this up into a few posts. So today we’ll have a look at voting intention, tomorrow we’ll do the satisfaction dynamics and Sunday we’ll go nerdy and run some election simulations.
So first up, we’ll take a squiz at the ALP primary vote:
As we can see, Western Australia is still the odd man out for Labor with an ALP primary that’s been tracking an average of 7 points lower than the national average over the same period since the 2007 election – even though Labor are currently shown to be just over 2 points higher in WA than their 07 election result. Also worth noting is how the female primary vote has swung 4 points to Labor since the election, while the male vote and the 18-34’s have actually swung 3 points away from Labor since the election. But all of that swing away plus more has moved across, on net, to the minors rather than the Coalition as we can see by taking a squiz at the Coalition charts:
The big movement here is among females and the 35-49’s (justifying our incessant banging on about how the ALP is deliberately targeting the Family Tax Benefit set) where the Coalition has had a 9 point swing away from them since the election with the 35-49’s and a whopping 10 point swing away from them on the female vote! Even though the female vote is historically far more volatile than the male vote (please discuss 
), this demographic collapse must be, by far and away, the biggest Coalition concern.
The over 50’s are the only real demographic strength left for the Coalition at the moment, even though they’ve swung toward Labor since the last election like everyone else. If it wasn’t for the over 50’s vote, the Coalition would be staring at wipeout levels of voter support – knock 2 points off their TPP every fortnight – which really goes to show the complete absence of political nouse in Tony Abbot with his February claims questioning the proposed rise in the age pension.
The other chief concern for the Coalition would have to be their performance in South Australia, which is looking pretty abysmal at the moment – even though Labor isn’t benefiting from those lost Coalition voters in primary vote terms, it’s actually the minors that seem to be gouging on the SA Coalition vote which we can see by having a squiz at the Others charts:
The other demographic movement with the minor’s vote would seem to be the Greens picking up larger numbers of the 18-34’s, although the minor vote is up in every state and demographic since the election.
Looking now at the two party preferred, where the results focus on geography rather than demographics, it’s the same old sad story for the Coalition:
That movement in SA from the Coalition to the minors in net terms on the primary vote really starts to burn on the TPP once preferences get distributed, with Labor picking up a 5.5 point TPP swing in SA off the back of only a 1 point primary vote gain.
On the MoE’s here, were looking at a total sample size of well over 5500, so for the gender breakdown we get an MoE around the 1.9% mark, while for the states it varies between about 4% for the small states like SA and WA down to a touch under 2.5% for NSW and Victoria.
For something completely off topic but hilarious, Crikey News on Twitter today had this to say about Conroy’s Q&A performance last night:
Someone should make Russian Mafia tshirts. “im at ur dentist, ruining yr internetz”….
11 Comments
Poss,
I’m a relative newcomer to these things, so forgive me if this has been flogged to death before.
Even to my untrained statistical eye, it is clear that in every state, every demographic, there is a sharp peak to the coalition at the election, accompanied by a sharp, temporary crash in Labor’s primary and 2PP vote.
Is this the factor of incumbency that the Coalition enjoyed at the election? For months people say they will vote Labor, then chicken out at the last minute and vote for the Libs because they are scared of change; only to jump back on the bandwagon after the election and profess their love for Lord Kevin once again?
It could well be Dave – incumbency often provides a few points of safety net for a government. There’s the theory of the Shy Tory – where polls for a whole number of different reasons end up underestimating the conservative vote – where the uncommitted voter tends to be conservative, and those that refuse to participate in polling are big conservative leaners. There’s probably something to that in Australia even though it’s hard to prove. The other hard part is trying to pull apart any Shy Tory effect from an incumbency effect – but Labor governments don’t appear to get as large a rush toward them when they are incumbents, as Coalition governments do.
I wonder if anyone has done any research on why people participate in market research? It may be that some people do it thinking that they are contributing to some kind of collective good, which would apply less in the individualist subculture within the conservative demographic?
There’s quite a lot of theories around on why, and plenty of market research on market research to back them all up
The last thing I read on that was a piece of research by Survey Sampling international
Blurb:
http://www.surveysampling.com/?q=en/about/news/mar3108
Paper:
http://www.surveysampling.com/files/imce/_paper_9_Understanding_respondent_motivation.pdf
If you ignore all the dressed up psychobabble, there’s a few interesting points under it all.
With political polling, there’s an entire generation of people out there that were brought up to think that the two topics you don’t talk about over the dinner table are religion and politics. Considering the current age of that generation, I could see how a Shy Tory effect could be lurking in the demographic that declines to be surveyed.
Poss
Totally off topic but I hope this isn’t a relative of yours?
http://www.news.com.au/couriermail/story/0,20797,25250069-3102,00.html?from=public_rss
Persistent little bugger tho!!!
I have a hypothesis on female voter volatility in comparison to male voter volitility.
I have a notion (which I think came from something I read) that men have a greater standard deviation of political belief, assuming a binary left-right scale. Even if the median man and median woman have similar political beliefs, there are still more men at each extreme and proportionally more women in the centre.
I guess this could be verified with something like the Australian Election Study.
But anyway, in a two party, compulsory voting, environment we’d expect more volatility from the female voters.
Since both parties should be chasing after the median voter, the frontlines between the parties should be in that ideological zone on the spectrum. People is this zone can have minor shifts in their own beliefs, or the frontline can shift through the usual inane news cycle dramas, or the weather or whatever, and they can still end up on either side of the ideological divide quite easily.
Meanwhile, those further out the left or right, even if they are just as fickle and inane and influenced by news cycles/weather/hormones, will just be oscillating deep in labor/liberal territory.
And if men do have a greater standard deviation on the binary scale, proportionally more of them will be having their random walk in areas that don’t affect voting intention. And conversely, proportionally more women will be doing their’s where it does effect voting intention.
So there it is. Women’s voting intention is more volatile because of the sausage fests at your local Communist Paty/Citizens Electoral Council meeting.
The Shy Tories? I think I just found the name for my next band
I guess this has been asked before, but here goes: Is it known or are there any theories why Morgan (including its phone polls) and Essential Research consistently show higher ALP and lower coalition voting intentions than Newspoll and ACNielsen? It is not unusual for Morgan and ER to have the ALP’s vote in the 60s, but to my knowledge this has never happened in a Newspoll or ACNielsen poll.
Morgan phone polls pretty much track the other phone pollster results and their last poll before any given election is statistically the same as the other phone pollsters. Their face to face polls definitely have a lean toward the ALP (you’ll also notice that the final Morgan poll is always a phone poll rather than a face to face poll), and that probably comes from their sampling frame – people who are at home in the middle of the day and agree to be polled face to face have some sort of structural political lean toward Labor.
Just why they do is a mystery.
Essential Report doesnt include “leaners” in their figures. For instance, Newspoll asks people who they would vote for if an election were held today, and if they were uncommitted, who are they leaning toward. Essential on the other hand just removes those leaners, making the Essential polls effectively a poll of decided voters, and decided voters only.
As a result, if Essential has a larger ALP vote than the other pollsters it suggests to us that either the ALP vote is harder than the Coalition vote (as in the Coalition vote as estimated by other pollsters has a large component of uncommitteds that are leaning toward the Coalition) AND/OR that Essential has a slightly skewed sample that isnt exactly representative of the broader public. We can’t disentangle those effects until an election occurs – and as Essential are relatively new pollsters on the block, their D-Day will come at the next election where we’ll be able to see how their polls stack up against the election outcome.
Thanks, Possum.
I see that the most recent Morgan phone polls are much the same as the others, though there were six in a row in early 2008 ranging from 60.0 to 63.5 2PP.
If F2F polls have a known bias I wonder why Morgan sticks to them. I remember a debate on the 7.30 Report many years ago in which he was very firm and a little abrasive in claiming that F2F is the right way to do a poll and phone the wrong way, while Sol Lebovic was just as firm that F2F is wrong and phone is right. Maybe he’s defended F2F too strongly to drop them.
> the female vote is historically far more volatile than the male vote
)
> (please discuss
Where discuss==speculate wildly ?
There are so many factors at play, it’s hard to pick just one.
Richard Green has already done a nice job on degree of staistical deviation in political belief above. What about I do ‘identification with brand’?
One factor in voter lock on to a given party is how strongly the voter identifies with that party. This is oversimplifying, but how much they think the party is like themselves.
All the major Australian parties are male dominated, the processes that create their shared values are male dominated, and their representatives are mostly male. They have a male image.
As such, it’s easier for men to align themselves more precisely with one party, and thus to see themselves as clearly unlike the other party. For women, it is a bit harder to align themselves with either party. The psychic distance to either party is greater, and the perceived distance between the parties is thus reduced.
Result: men are more likely than women to feel a strong allegiance to a party.