Something worth taking a look at is how the new electoral redistributions have changed the pendulum in the context of how the current polling is playing out. If we apply the latest combined Nielsen/Newspoll quarterly aggregates at the State level to the new pendulum,it gives us an idea of which Coalition members are fighting for their life. The Coalition held seats below are those where the ALP has theoretically overtaken Liberals on vote estimates. All the figures are in the form of ALP two party preferred vote – where the 2007 election results on the old boundaries for each seat is given, the projected ALP two party preferred vote based on the current polls as they apply to the new electoral boundaries is given, and the practical swing (the difference between these outcomes) for each seat is given. (Note – please don’t call this a generic “swing” or Antony will put you over his knee and spank the crap out of you 
It’s simply the difference between the 2007 election result and the current polling )
Keep in mind this is based on a pendulum, so even though the results are given to 2 decimal places (just to show how close some of these seat results would be) we can’t be that accurate in the real world – but it’s certainly indicative of the likely results were an election held in the past three months and where the results reflected the polling over that period.
In the real world, some of these seats wouldn’t fall, some would fall with much much larger swings, and possibly a few seats much further up the pendulum would fall – but just using the pendulum and applying the current polling, these are the seats now notionally sitting in the ALP column which are currently represented by members of the Coalition.
We can also take a look at what Coalition seats the pendulum suggests would become very marginal, seats that the Coalition would hold by less than 2%.
This last one is interesting because of the Liberal Party Heartland nature of seats like Menzies, Higgins and Goldstein turning up in the list – all in Victoria. Because these three seats – and we can also probably throw Casey and Aston into the mix here as well – are unlikely to be swinging as much as the nominal polling is currently suggesting, it means that other places in Victoria are swinging more than is suggested simply to make up the weight. The problem this causes for the Liberals is that either other seats up the pendulum are moving instead by significant margins (possible for one or two seats, but unlikely as a general across the board case), or alternatively, Labor held seats and marginal Coalition seats are experiencing a much larger swing, building a formidable safety margin into Labor’s currently held seats and possibly even seats that are currently Coalition marginals in Victoria.
At the last election, of the 37 House of Representative seats in Victoria, the ALP won 21 of them, where 11 of those 21 ALP seats had a two party preferred vote of over 60%. On the current polling applied to the pendulum in Victoria, the ALP would win 26 seats in Victoria – but get this – they would hold 19 of those 26 seats with a two party preferred vote greater than 60%
But it gets worse for the Coalition in Victoria. Because seats like Goldstein, Higgins, Menzies, Aston and Casey wouldnt be expected to swing as much because of their demographics, other seats would have to pick up the weight of the swing. Similarly, if we look at the ALP side of the pendulum, it has Batman sitting on a two party preferred of – and I shit you not – 81%!
You just dont get seats sitting on margins like that. If we look at the ALP seats that would be sitting on a TPP of greater than 70% under the new polling, we have nine of them:
So we would expect those seats to swing smaller than is suggested as well, leaving other seats to take up the weight of the swing. If that “swing weight deficit” from the heartland Liberal Party seats and the ultra-safe ALP seats was distributed across the rest of the pendulum, the most likely outcome is that of the 26 seats that Labor would likely win were an election held recently, around 22 of them would have an ALP two party preferred of more than 60%!
That would make 60% of all Victorian seats ultra-safe Labor.
The only place looking worse than Victoria for the Coalition is South Australia. If we apply the current polling to the pendulum in SA, the only Coalition representative in the entire State would be Pat Secker, holding on to what would be the marginal seat of Barker by 1.6%.
Yet here, again, Mayo, Barker and Grey shouldn’t swing as much as the state average (although Grey caused a scare for the Coalition in the lead up to the last election) because of the demographics of those seats and Port Adelaide is already ultra safe for Labor, making it unlikely it would reach 77%. That suggests that the rest of the State would be swinging more to make up the weight, turning all of Labor’s currently held seat into ultra-safe ones, and washing Boothby and Sturt into the ALP columns with significant Labor margins.
49 Comments
Do these numbers explain why Turnbull is starting to talk about spending more time with his family? It is a dismal road ahead of him at least two terms out of Government. I still don’t know why the Greens just stall on these numbers. They never seem to get past where they are now. And on election day the numbers, if I remember rightly, drop a little.
AS Pauline would say Please explain Poss.
Putting on my rational hat here (rarely worn) it not good for government is it if so many of what could be seen as reasonable Libs are wiped off the map? Eventually you end up in the Political Nirvana called NSW and no-one wants to go there.
That first table is a bit loopy because you’re “swing” column is partly swing and partly change in electoral boundaries. You get a 10.75% swing for Greenway which is not a swing at all, as most of it is caused by the change in electoral boundaries. Swing should always compare like with like or it gets confusing.
Antony, it’s using the perspective of the actual member – hence swing+redistribution.
Using Greenway as an example, Louise Markus had a 4.5% margin after the 2007 election, but on current polling now faces the prospect of the ALP getting around 56% on the TPP.
We’re smart people here – readers wont be confused.
Oh yes they do. Call something swing and people think you mean swing.
It’s not called swing anywhere above – its the “Swing to the ALP + Redistribution”.
If people want to walk to the beat of a different dictionary, there’s little I can do about it.
I’ve added a note to the post Antony, hopefully ensuring that no confusion arises.
C’mon Green, be a little generous. This isn’t discussion for the lowest common denominator. We also apprecite that nominal redistribution figures are always to be taken with a grain of salt.
And Batman… jesus 81 tpp would be some kind of record. Out of curiosity poss, what IS the record tpp result for fed seat??
Why don’t you call it “Change in Margin” which is what it is, the change having two components, swing and redistribution.
Sorry Dan, but the data is mixed. The heading says new pendulum, all the seats and members listed are old pendulum, but the last two columns are an estimated post election pendulum. I know the point Possum is trying to make, but the redistribution creates a discontinuity between the 2007 election results and the post-2010 result.
Dan, over the last 5 elections there’s about half a dozen seats that get around the 75% mark for one side or the other. The one with the largest margin was Batman for the ALP in the 1998 election, getting 76.43% TPP.
Antony,
It’s only a discontinuity if one looks at it as being something it’s not.
For instance:
That sort of framing and description is something we dont usually see mentioned – even though that is *exactly* the way Don Randall would be looking at it.
Yes, sure – the swing in Canning is actually an estimated 6.6% and the electoral boundaries are responsible for a further 1.2% shift to Labor. But sometimes I think it’s worth looking at this through the eyes of the Members involved and how their electorate has and is practically shifting underneath them – not as an exercise in psephology mind you, but as an exercise in understanding politics.
Ooooooooh I love it when psephologists go at it .
Durutticolumn, the Greens never get past those numbers because their socio-economic policies only appeal to a small percentage of the population. There is about %10 of the population which will vote for the far Left (and opposite to that is the %10 or so who will vote for fringe far Right wing parties) but beyond that ammount, their policies on the economy and society are too extreme for moderates to vote for them, even if they agree with their environmental stance. Jack the Insider put it perfectly in his column the other day. The Greens are not just a “Green” party, but also a party for refugees from the far left.
If the Greens wants to make more headway as Green party, then they must become centrist on socio-economic policy. It is unlikely they could convince more Australians to vote for them in their present state. I understand that then they would no longer be everything that they stand for now, but that is precicely the problem. Being Green and being Left wing are not inherently the same thing.
Duritti went:
On the one hand, it’s certainly not good for the quality of governance – bad opposition can easily lead to lazy government. The other problem is that the biggest deadwood for the Libs are in the safest seats, meaning an Opposition where mediocrity is concentrated in any electoral wipeout.
But on the other hand, longer term, any election wipeout also leaves more seats for new blood to come through at subsequent elections assuming of course that the Oppo can win them down the track – so there’s upsides and downsides longer term.
Dewy – the far right (by that I assume you exclude ALP, LP, Nats from such a description) are nowhere near 10%… they’d be hard pressed to get past 4%.
Either that means the far right is not as popular as you assumed, or that the far right inclined voters have found homes in one of the tweedles.
D Richards, agreed, support for the far right is probably much lower than %10, I was really just generalizing there. My point is the same though. Fringe ideologies are not going to convince the mainstream to vote for them, except probably in very exceptional circumstances.
Dewgong.. I guess I was thinking out loud. I agree with your assessment. They just haven’t cracked out of that stereotype. But while I don’t agree that they are far left I don’t think their politicians are able to communicate well enough to convince people they not entirely cracked. I heard Christine Milne’s speech to Press Club on climate change and carbon reduction and renewable energy and it was one of the best I have heard on the subject from anyone. But we don’t see much of her. I know many love Bob Brown but he annoys the beejeesus out of me for some reason.
Poss I agree on new blood but we still waiting in NSW I suppose if it is the landslide predicted then enough talented Libs will wash up on the shore to be able to provide the new government with some cred Mind you i think my dog and cat could do a better job. Than Nathan and his band of no-hopers, pimps and urgers
Durritt, I’m not sure it is entirely a steryotype. They might not be “far” left but they are certainly much further to the left than any other of Australia’s major political parties and far enough to convince most people not to vote for them.
Compared to the ALP and Lib/Nats, even a genuine centrist party would look like a lefty party
The Dems (pre Natasha and pre Lees) would now be left of the ALP and Libs/Nats
The true far left would be as small supportwise as the far right
Greens are left of centre… but not all that far left
Both majors have done a good job of demonising any third party, and enough people swallow the bulldust to stymie any attempt at breaking the nexus
Hi Possum, I have a burning fundamental question about this kind of
analysis (I’d love it if Antony could chime in as well…).
Ignoring the re-distribution issue, and any demographic difference
between seats, I have never understood why ‘uniform swing’ is used in
the way it is, from a fundamental mathematical perspective. What seems
to happen is the following: Taken overall (be it by state of federally
or whatever) we see that there was (or will be) a swing (change in %
vote for a party) of X%. Therefore, if we assume a uniform swing, then
the predicted vote for a given seat is Y + X.
However, ‘uniform swing’, being used to add percentages, seems to be a
silly way of doing things. In the ‘all things being equal’
demographics wise way of thinking, what you really want to say is that
a given LNP voter (say) has the same chance of changing to Labour
across different electorates. That is what people think ‘uniform
swing’ means, but that is not how it is applied. A seat sitting at around
50% will get the same answer either way, but as an extreme example,
say a seat sits at 95% Labour, and there is a 10% swing. The
conventional approach tells us that, demographics aside, 105% of
voters will be voting Labour!
Now, I’ll spare the details but it is very easy to show that using
this simple description of ‘uniform vote change probability’ you find
a simple formula for it. Lets take a swing to Labour as an example,
and the formula is given by
Probability of random LNP voter switching to Labour (on net) = Labour
vote at last election * 2 * swing to Labour at this election / (1 -
Labour vote at last election)
where swing is defined in the conventional sense of (vote after – vote
before). The factor of 2 comes from the way swing is expressed, if you
expressed it in terms of vote margin between the two parties then you
don’t need that factor. Note that I’m expressing everything as
decimals (i.e. 55% = 0.55).
Now, what does this tell us? Let’s take the seat of Batman as used in
the above analysis. In your analysis you take the 07 Election vote to
be 76% to Labour (once rounded) and use the Victorian uniform swing of
5.1% from recent polling to suggest Batman is notionally at
81.1%. You suggest that this is absurdly high, and expect the swing to
be less because of demographics or the fact that ‘You just don’t get
seats sitting on margins like that’.
Now, lets look at this in the way I suggest above. So, take Victoria
as a whole. The vote at the 07 election was 54.3% and the uniform
swing from polling sits at 5.1% currently. This means that a Victorian
LNP voter, on net, has:
P = 0.543*2*0.051 / (1 – 0.543) = 0.121
or in other words about a 12% chance of changing their vote. We can
then use this to predict the Batman swing result by turning the
probability formula around. We get
swing = P(1-last Labour vote)/(2*last labour vote)
Now, we use a uniform probability of a LNP voter changing vote across
Victoria, but now calculate what swing that would give us given that
only Batman had a Labour vote of 76% at the last election. We get
swing = 0.121 * ( 1 – 0.76) / (2*0.76) = 0.019
or in other words 1.9%. This predicts that Batman will sit on 77.9% if
we assume that any Victorian LNP voter will, on average, have the same
chance of changing to Labour regardless of where they live.
The difference between appraoches is stark, and is a much more obvious reason for
suspecting that seats won’t end up going like the naive ‘uniform swing’
analysis suggests.
Really, this has bugged me for years. Lets push this to really show
how absurd the conventional ‘uniform swing’ approach is. With your
predictions of Batman moving from 76% to 81.1% what you are saying is
that LNP voters (on TPP) in that seat have a probability of
P = 0.76*2*0.051 / (1-0.76) = 0.323
or in other words about 32% chance. That requires Batman LNP voters to
be about three times more likely to change to Labour than the state
average. THAT is why Batman will not sit at 81.1% after the next
election, because the basic mathematics doesn’t suggest that it will, not
because of demographics or any inherent notion that seats just don’t
go beyond certain margins.
The approximation of simply adding the uniform swing percentage to the
margin of a seat only works as long as the seat is quite close to 50%,
for anything else it is a very poor, and even absurd assumption.
Of course this actually suggests that safe LNP seats are less safe
than would otherwise be expected, but demographics is an important
factor. But even when you consider demographics, you get a far more
sensible result by applying a factor that alters the probability of
each LNP voter changing to labour than simply adding percentages in a
mathematically invalid way.
No no no, too mechanistic. The uniform swing is the National swing, and the swing in seats is essentially distributed around the national swing, because each seat is one component (150th) of the national swing.
The manner that causes different seats to end up in different parts of the distribution depends on what is causing the swing. I can name elections (eg 1987) where the swing was enormous in safe Labor seats. There are elections, very unusual ones, where the swing was huge in country seats (1974, Victoria 1999). There are elections where the different behaviour of the mortgage belt caused a particualr distribution of swings (1998) or elections where the state swings varied enormously (eg 1990).
It is expressing the swing as a distribution that is behind all those funny graphs Possum does using monte carlo simulation. You could adopt the model you’re talking about, but it would just produce smaller swings in the safest Labor seats, bigger swings in the safest Coalition seats, and have no impact on the overall result because that is entirely determined by the swing in the marginal seats. It doesn’t matter if the uniform swing produces the wrong margin in Batman because it just doesn’t matter.
The uniform swing model works okay as long as you recognise it works because on average it works, not because something makes it work. Your model would make little change to the swing in marginal seats.
to distribute around the mean
If the Greens are a green party (which many dispute) then they would be a conservative party, ie they would be in favour of conserving things. It’s biggest trick of the last half-century that right-wing parties (who conserve nothing) have managed to walk away with the ‘conservative tag’, so that parties which advocate rational policies are able to be painted as extreme.
I don’t see it’s particularly conservative to run a planet-wide experiment in the degradation of natural systems, with no control. Sounds pretty radical to me.
“Your model would make little change to the swing in marginal seats”. Exactly, becuase in that regime, the ‘uniform swing’ approximation is correct (or at least sufficiently accurate) such that both approaches match.
What is not true though, is “each seat is one component (150th) of the national swing”. That is simply incorrect mathematically, at least if you express swing as a percentage (which is what is always done). If you take the mean of the swings in all electorates you do NOT in general get the national swing. Not if you are playing with the percentages, you have to do it in terms of acutal voters (i.e. 500 voters here, 4000 there) if you want to do that.
What I’m talking about is that before you consider the different cause of vote change etc etc, which are obviously very important to consider, you have to simply have the correct way to tally things up, and adding percentages together is simply wrong mathematically, you are not averaging what you think you are averaging. If you don’t pick this part of the error in uniform swing out first, before you then consider all those other crucial factors, then you are simply chasing your tail.
We know that the real situation is more complicated than applying any uniform measure to individual seats, but there is no reason to intentially start from an even more wrong start point, purely because of flawed mathematics. We may as well start from a sensible position after applying a uniform measure, and then think about what else might be going on.
It just seems lazy to do something wrong, just because there are other sources of error in the system. You may as well control the thngs you can control as best you can.
Antony, has anyone researched these anomalies (post 22)?
Seems relevant, at the present, with the non-capitals supporting the Coalition against the trend in the cities. Almost certainly (as there is nothing else) the ETS causing this concern in the bush.
If this is true, and prevailed till an election, it could well be another instance.
True, the national swing is not technically the average of the seat swings. However, because all seats have roughly the same number of votes, then the national swing will be almost exactly the same as the average of the seat swings unless the swings by seat are not normally distributed, and they are nearly always normally distributed.
What you’re talking about is using a proprtional swing model rather than a uniform swing model. There’s a whole literature on the subject if you look around for it. Here’s a quick sample http://johnrentoul.independentminds.livejournal.com/129567.html
Nipper, there is huge research on those subjects. There are books and papers published after every election on the subject.
I should have added before, thanks Antony for giving your view, it is much appreciated. Hope I didn’t come across too strongly in my reply!
The fact is that people use this uniform swing stuff all the time. Your own website has various ‘election pendulums’ were you can plug in a uniform swing and see the nominal outcome, so you must see the use of some amount of ‘uniform measure’ analysis.
Taking the current Federal situation, we had a TPP for Labour of 52.7% at the last election and current polling (according to Poss on the right hand side) has Labour on 55.9% for a ’swing’ of 3.2%. Now, we could plug that into your pendulum and get the notional Labour and LNP seats out. We would know that this is unlikely to be correct, but starting from those results we can think about individual seats circumstances to modify the situation in a seat of interest.
Now, lets look at how wrong we are using the ‘uniform swing’ approximation. Notionally a seat with a LNP TPP of 53.2% at the last election now sits at 50%, so anything less than that is notionally Labour if we apply a uniform swing.
Now, if instead we apply a uniform probability of vote change, we get that around 7% of LNP must have changed there vote, on average across the nation, to go from Labour TPP of 52.7% to 55.9%. So, lets take our seat with LNP TPP of 53.2%. Using uniform probability we would say that the swing will be 4.1%, suggesting that rather than sitting at 50%, this seat is notionally Labour (assuming current polling applies uniformly) by a margin of 0.9%. This is a not insubstantial difference, and comes about from simply applying a national average poll vote to an individual seat in the most obvious way.
I can’t see why this is not a better way to do those simple ‘election pendulums’ than a model which actually assumes that voters will have a different probabiliy of changing votes based on where they live, which is precisely against what the idea of assuming a uniform measure is trying to do.
Sorry, got ninjad by Antony. Anyway I still point out that even if all seats had the same number of voters, the national swing is still not the average of the individual swings. Yes if the distribution is normal it will work, but that’s only because the errors for seats either side cancel out. If you want to get the best prediction you can for a seat, you still do a better job than a scheme where you know you have the wrong answer, but know that this will be cancelled out by an equally wrong answer in some other seat.
You point out the literature on ‘proportional swing’ which indeed appears to be what I’m talking about, although not that I’m saying it has any better predictive power. It is simply the way to go from a national phone poll type result to a notional result for a given seat with the fewest assumptions.
Nippy
a couple of factors which might explain the rural/city divide:
1. Most rural areas have their own sources of news – local newspapers, TV and radio – which do not necessarily give issues the same slant they are given nationally. (There is also a tendency for these to be conservatively biased, partly due to the perceived audience).
2. Few rural seats will have an ALP candidate yet (other than those already held). So there’s noone to put an alternative point of view.
I would expect that once – and, in some cases, if – ALP candidates hit the hustings, these figures will swing back (no predictions as to how far).
I’ve been doing books of election results for 20 years, and the sort of difference between swing in marginal, safe and ultra-safe seats you are talking about does not occur. Here are the FED 2007 figures in Safe (>10%) Failr safe (6-10%) and Marginal (<6%) seats by party. Note the NATs had only a small number of seats.
ALP 5.0 / 4.9 / 5.0
LIB 6.3 / 5.9 / 5.2
NAT 4.7 / 6.9 / 7.8
Here's 1998 figures from an election with a 5% swing election, rare at Federal level
ALP 4.2 / 5.1 / 5.3
LIB 5.5 / 4.2 / 3.0
NAT 6.5 / 5.3 / 3.5
We use the uniform swing model because it works relatively well, and I see no reason to abandon it unless there's a reason in the results to abandon it.
No, if all seats have the same number of votes and the swings are distributed normally, the the national swing will be the same as the average swing. Each swing represents a percentage of voters in the electorate. A 4% swing says that 4% of voters in a seat changed. You can average these swings and get the national swing. No one calculates it that way, but its effectively the same.
I lost a post there somwehere. I was trying to say that from doing books of election results for 20 years, I see nothing that shows any pattern of swing suggesting a proportional swing is the more appropriate different. Here’s two elections where I list the swing by marginality and party. The numbers are Safe seaTS (>10%), fAIRLY sAFE (6-10%) AND mARGINAL (<6%).
2007 FEDERAL
ALP 5.0 / 4.9 / 5.0
LIB 6.3 / 5.9 / 5.2
NAT 4.7 / 6.9 / 7.8
Here's 1998
ALP 4.2 / 5.1 / 5.3
LIB 5.5 / 4.2 / 3.0
NAT 6.5 / 5.3 / 3.5
Those are 2 elections with a big national swing. Note there were only a small number of National Party seats.
This is just a test, I think my posts aren’t getting through.
mmmmmm, the post where I explained swing by marginality of seat isn’t getting through. I was trying to say that in my 20 years of doing election results books for Parliament, I always do tables of swing by pre-election margin. And there’s almost never anything like the relationship you’re talking about.
You’re coming through loud and clear Antony, you posted while I was typing my post #28, which is why the chronology of the discussion is a little confused.
Well, I’m not convinced. I’ve just checked back through result book after result book, and the only example I can find where the uniform swing model failed in safe seats was 1988 in NSW, when the overall swing was 8.3% to the Coalition, but only 4% in very safe Coalition seats. If your method was correct, then all attempts to use the unform swing model would under-predict how many seats fall, and I’ve never seen anyone produce a figure saying the uniform swing models are sytematically under predicting. Mathematically it would be more likely to occur at elections where there is a big swing.
Josiah Thomas (Labor) polled 85.8% of the vote in a two-candidate race against a Liberal in the seat of Barrier, NSW (based on Broken Hill) at the 1910 election. I think that’s the record in a contest between the two major parties.
Zoomster, take your point. But isn’t the swing (non-capitals) against Labor a recent phenomenon? The structural (media) impediments for Labor also existed before the rise in Coalition stocks.
Thanks Antony, the data speaks for itself. It seems somewhat odd though. On the face of it just says that the more voters there are of a particular persuasion in a given local area, the more likely a supporter of a different party is to switch, or conversely your more likely to stay with a party if those around you are. This makes sense, what is surprising is that the magnitude of this effect would be so finely tuned such that it made the simply uniform swing model work.
I can’t argue with the data, but I’d be interested to have a looksee at the results to try and make sense of this. Is there a good source with lots of historical data in an easily machine readable format somewhere?
Scratch that, just found a wealth of data by clicking on Psephos’ name link.
Zoomster, I have my own theory about the city/country divide.
That is that country people generally exist within a smaller, more tightly connected social network. Not only does that encourage group-think, but it also makes a fertile territory for less-rational conservative ideas. In other words, if you think of ideas as memes then those memes engage in a lot more inter-breeding in the country
Where people are more connected socially such as in cities and have a wider, deeper social network, they’re more likely to have had their ideas challenged and to avoid the simplistic messages that you get from conservatives. (Yes, I’m well aware that its a lot more complex than this.)
Its an ugly truth that country people are also more conservative socially, more religious, more homophobic and so on, and for much the same reasons. I know that’s not politically correct of me, but it does deserve more mention in psephology all the same..
Cud Chewer, I grew up in National party heartland (in Wagga Wagga, regional NSW) and then moved to Sydney more than a decade ago. I think most of what you’re saying is reliant on some pretty out of date and innacurate sterotypes! Let’s go though them one by one:
* In regional communities, people tend to have larger and deeper social communities than in cities. There are a lot more lonley people in Cities than in the country.
* You are suggesting that the more people talk about things the less rational they become. We’d better close down this site then!
* There is much more racism in Sydney than regional NSW. In fact apart from my Grandfather (who lived in Sydney) I never actually experienced racism in action (as in observed racist remarks etc) until moving to Sydney (after living 18 years in the country). Its not as though this was because the population was completely white either.
Regional Australia does tend to vote for the National Party, who are the arch conservative party. However, the recent trend has been for Independants and Liberal memebers to eat into the Nats seats. Primarily this is because I think the Nationals don’t know why people vote for them. The main issues are more about rural services etc, rather than social conservatism that is what gets the Nats much of the vote, but it is also why more socially liberal but rural focused Indepedants are taking over. If you look at the seat of Riverina (which Wagga sits in), Kay Hull is the popular local National with independant leanings who has crossed the floor a number of times over issues like the sale of Telstra and voluntary student unionism. However she is no rampant social conservative and has been involved in the arts in Wagga for many years, long before she became a polly. She doesn’t get a huge margin by appealing to racism and homophobia and would surely lose votes if she tried.
Yes there are plenty of racist wife beating walking sterotypes in the country, but there are plenty in Sydney as well.
If anything, the much of the rural/city divide on some issues is probably to do with the age of the population. Generally rural and regional areas have fewer 20-30 year olds, since a lot of people move to bigger cities after school and either never return or return once they start a family later on. That would be my theory anyway.
Antony, a couple of your posts were in the spam bin and I’ve pulled them out.
Sorry about the delay, I was in bed
Antony and/or Possum is there a good measure of electorate variability? I’m thinking of something like standard deviation or variance. What I’m after is some way to infer the number of swing voters in a given electorate. Or does this come out roughly even for most electorates. I imagine you have some feel for which electorates are partial to a swing and which are fairly immune and which have a long term demographic trend.
EP, generally elections have a standard deviation of between 2.5 and three and a bit. If we take the average of the last 3 elections, by state, we get:
NSW 2.6
VIC 2.2
QLD 2.7
SA 2.1
WA 1.8
National 2.7
As for something more precise, like the number of swing voters in a given electorate – that will often depend on what the issues of the day are. Different issues ring different peoples bells in different electorates or groups of electorates.
Qld, for instance, is inclined to have larger swings than other states – but mostly in regional and outer suburban electorates rather than inner suburban ones. Things west of the range in NSW don’t swing much at all – but can (like with One Nation and when well known Indies stand).
There’s probably not any given measure of swing voters at the electorate level – but you could probably design a rough approximation of one specifically for what ever it is you’re trying to do. (I say that anyway, Antony will probably disagree on this last bit)
Historically, the largest electoral volatility occurs in mortgage belt seats. As I always say, the point where couples are having children and buying a house is the moment in life when most people find the balance between what they earn and what they spend is under greatest pressure. At that monent, anyone’s underlying political allegiance is most likely to be questioned by the financial pressure of the present. It’s why the Howard government did so well with family tax benefits and maintaining low interest rates. And the mortgage belt is the mortgage belt because it has the highest concentration of the sorts of voters responding to these pressures, so with more of these sorts of swinging voters, it sees bigger swings.
But it all depends what is driving voter change at an election. At the 1990 Federal election, you saw bigger swings in Victorian electorates because the Victorian government was an issue. In 1987, whether the Hawke government was a real labor government caused bigger swings in safe Labor seats. In 1998, whether the Coalition was doing enough to lower migration was an issue in areas where One Nation did well. The backlash in country seats against a string of measures by the Whitlam government generated unusually large swings against Labor in country seats.
As for Possums last comment, you probably could come up with a measure of that, but you first have to predict what is the factor that will cause certain blocks of swing voters to swing. Voter behaviour is massively mutli-variate, and at any particular election only some of the variables are in play. You have to figure out which ones.
But, as I said in the first paragraph, young home-buying families carry the biggest bang for your buck for political parties because they are up to swing and they are contentrated in certain seats. Older voters and farmers are also concentrated, but they are not prone to swing and so attract less attention.
Bogdanovist @ 43..
Rural communities do not have the depth of social connectivity. They may claim to stronger social ties, but the point I was making was more technical. Its about the depth of social networks and the extent to which they have more deeper links. This is why I say that conservative ideas are more likely to be challenged in cities because there are more links deeper into the network.
I did not mention racism. That’s complex. I think racist ideas are less likely to be challenged in small communities. But racist ideas are more likely to form in cities – especially those experiencing culture clashes and growth.
What I did say though is that there is more religiosity in the country and for the same reasons – your social network is less likely to confront you with alternate ideas. Same goes for homophobia. Like it or not there is a big city/country divide there.
You do have a point regarding the migration of younger people. However, I’ll believe you when Wilson Tuckey loses his seat to a moderate