tip off

Did Greens HTV Cards win Labor any seats?

The AEC has released the preference distribution data from the election, so it’s worth having a bit of a squiz at the way preferences flowed from the Greens to the ALP, as there’s some interesting little bits in there.

To start with, it’s worth looking at the broader picture on the vote side – so the primary vote and the primary vote swing for the Greens and the ALP, by state and territory, came in like this:


That gives us a bit of a feel for the results – so in places like NSW we had the swing away from the ALP being much larger than the swing towards the Greens, while in places like Victoria we saw the opposite occur, where the swing to the Greens was much larger than the swing away from the ALP.

So we have a pretty non-uniform result across the States and Territories in terms of the magnitude of the Greens picking up votes at the net expense of Labor.

On the preference side of things, the flow of Greens preferences to the Labor Party at the national level reduced by nearly a point, from 79.96% at the 2007 election down to 78.84% in 2010. If we break those Greens-to-Labor preference flows down to the state and territory level, we find that only the NT had a particularly large change – with flows dropping by 4.3% since the last election:


NSW is an interesting case, where not only did the ALP take a pretty big hit on their primary vote, but the preference flow from the Greens to the ALP also dropped – suggesting that the ALP really was quite on the nose  and not just suffering a protest vote from the left end of the spectrum.

In terms of the two party preferred, we can look at how many points of ALP two party preferred that Greens voters ultimately delivered to Labor at the election.


What we find is that Labor experienced an increase in the number of points of two party preferred that the Greens gave them in every State and Territory – though with quite a bit of variance. In NSW, the Greens gave Labor an increase of 1.7 points over the 2007 election result, while in the ACT it increased by 5.3 points.

The headline numbers are also quite interesting, with Greens voters in Victoria, WA, Tassie and the ACT now all delivering the ALP double digits worth of two party preferred. This is important mostly for “stating the obvious” type reasons – the Labor party is becoming more reliant on the Greens delivering them the ultimate post-preferences vote they need to win seats.

This raises a couple of questions. Firstly, how many seats did the Labor party win because of Greens preferences?

The answer here is easy – out of the 72 seats Labor won, 48 of those were reliant on Greens preferences for victory.

The second question is a bit trickier: What effect on Greens preference flows did Greens HTV cards have?

To answer this, we can do some regression work. Firstly, we need data on the electorates where the Greens preferenced Labor over the Coalition compared to where they didn’t run a recommended preference flow on their local HTV cards. We can get all that from here.

Next, we need preference flows by electorate, which the AEC kindly provides here. We can’t use all 150 electorates here as there were some issues in some electorates which distorted things slightly – so if we use only those electorates where the final two candidates were Labor and Coalition (removing the issues), that gives us 142 seats worth of data to look at.

If we regress the Greens-to-Labor preference flows by electorate against a dummy variable which represents those electorates where the Greens preferenced Labor on their HTV cards (96 of those 142 electorates), we find that the existence of a Greens How To Vote card preferencing Labor increases the Greens to Labor preference flows by an average of 3.3%.

But, that’s not the full story. One of the issues here is that the preference flow rate is correlated with the size of the Greens primary vote in such a way that for every 1% the Greens primary vote increases, the preference flow to Labor increases by around 0.6% – presumably as that increase comes from people voting for the Greens that would ordinarily be Labor voters. So once we control for the size of the Green primary vote, what we end up with is this (a cut and paste from the stats software):


For every 1% increase in the Greens primary vote, preference flows from Greens to Labor increased by around 0.6%. Holding that constant, what we also find is that the existence of a Greens HTV card preferencing Labor increases that preference flow rate by 2.71% These results are highly statistically significant.

Also worth noting is that I checked for a donkey vote effect in a number of ways (to make sure that any donkey vote wasn’t distorting the relationships were looking at here) and there wasn’t one that was even remotely close to being statistically significant.

What do these results mean in practice?

Well, let’s take the tightest ALP win at the election in the seat of Corangamite as an example. Here there was a Greens HTV card preferencing Labor ahead of the Liberals and the result was a Labor two party preferred result of 50.41% to the Libs 49.59%, or 47,235 votes to 46,464 – a Labor lead of 771 votes.

Greens preferences in Corangamite flowed to Labor at a rate of 80.31% (while the broad “Others” flowed to Labor at 41.92%). If the Greens didn’t have a How To Vote card that preferenced Labor, we would expect that the preference flow rate would have dropped by around 2.71% – bringing the Greens preference flows to Labor down to 77.6%.

With that 77.6% Greens preference flow, Darren Cheeseman would still have won Corangamite for Labor by 50.1% – or around 190 votes.

As that was the closest Labor seat, there was no seat in the country where the size of the average Greens HTV card effect was larger than the Labor Party’s winning margin –which really is quite interesting.

That’s not to say that Greens HTV cards didn’t deliver Labor some seat, somewhere -just that if such a thing did occur, it was more a result of fortuitous local variation than any broader, nation wide  HTV card effect.

Greens voters, on average,  didn’t appear to pay a great deal of attention to their party’s How To Vote cards.

Then again, they rarely do.


Please login below to comment, OR simply register here :

  • 1
    Posted October 4, 2010 at 1:34 pm | Permalink

    Obligatory “begs the question” pendant comment here.

  • 2
    Posted October 4, 2010 at 1:38 pm | Permalink

    Well, we wouldn’t want that! I’ve replaced that naughty word :-P

  • 3
    Posted October 4, 2010 at 2:05 pm | Permalink

    I haven’t had chance to run the same regression or use my graphic software, but it’s interesting to compare with a post I did on the 22007 results.
    The constant term is the same, this strange exogenous 70% starting point for Green preferences to Labor. The Coefficient on the % vote is much flatter, but that would make sense because the Green vote was higher, and the rate of Green preferences must become asymptotic to 100% as the Green vote rises.
    The donkey vote disappearing might also relate to the generally higher Green vote, which means as the Green vote rises, the proportion of it’s vote that is donkey vote shrinks and could lose its significance.

  • 4
    Kevin Bonham
    Posted October 4, 2010 at 2:24 pm | Permalink

    I suggest that the reason there was no seat at all in the country where the expected HTV effect was larger than Labor’s winning margin was the notable absence of ultra-close seats, and in particular very close Labor wins, at this election. The closest seat was 50.43 2PP. At the 2007 election there were eight seats that were twice that close or closer, and three of those were won by Labor. Most of the time Greens HTV preferences would be worth a seat or three but this time none of the seats were close enough.

    A question I find it useful to keep an eye on is “How many seats did Labor win from behind on Greens preferences?” The reason this matters is that Labor winning seats where they only need 25% of Greens preferences to get across the line means nothing because it can be taken for granted that the Coalition will not get the majority of Greens prefs any time soon. Whether Labor’s majority of Greens prefs will stay as high as it is in the long term, is another question.

    I believe there were eight such seats this election (nine in 2007).

    Labor also lost one seat from ahead on Green preferences – in a very technical sense. Andrew Wilkie’s 2CP gain on Green preferences of 1652 votes exceeded his victory margin of 1575.

  • 5
    Posted October 4, 2010 at 2:32 pm | Permalink

    Possum, just playing around, if you construct a donkey vote variable where if the Greens appeared at the top of the ballot paper, if Labor is higher on the ballot paper the value is 1, and where Coalition is higher the value is -1, you get a regresiion varibale about the same size and probability as the presence of a direction of preferences, about 3%.

  • 6
    Posted October 4, 2010 at 3:25 pm | Permalink

    The donkey vote is a weird one this election Antony. I can find it all over the place using different types of variable to spec it out, and nearly everytime I find it having a larger effect on the Greens primary than any other party.

    Yet, so saying, I can’t find any significant relationship between that donkey vote and preference flows from Greens to ALP.

    Want to see something really interesting – try regressing Green primary vote swing against ballot position.

    Weird dynamics.

  • 7
    Posted October 4, 2010 at 3:37 pm | Permalink

    I know plenty of Greens voters, but to the best of my knowledge, none of them actually take HTV cards on election day.

    I still find it hard to fathom how anyone votes Green and preferences the Coalition. The only two factors I can think of are the personal appeal of particular candidates, or an “anyone but Labor” mentality. I don’t think anyone whose vote was effected by those factors would take much notice of an HTV card.

    I know this probably hard information to get, Possum, but are there any seats where the Greens directed preferences in only one of the 2007 & 2010 elections? Looking at how the preference flow changed for the same seat with and without a HTV, relative to the national trend, could be interesting.

  • 8
    Posted October 4, 2010 at 3:40 pm | Permalink


    Comment 7.

    Oh come on, tells us what happens!

  • 9
    Posted October 4, 2010 at 3:44 pm | Permalink

    Oh, a factor that intrigued me was the fact that the 80% thingo actually held.

    Take a seat where the green Vote went up by 5%, 4% went to back the ALP and 1% went back to the LIbs,. It reinforces the theory that the progressive small l libs are capable of, and do do, exactly what the ‘progressive’ lefty is doing,. Ie making a strategic and highly thought out vote.
    Lastly, when will the Greens stop being a protest Vote?

  • 10
    Posted October 4, 2010 at 3:45 pm | Permalink

    Barking – It’ll be a whole post tomorrow or Wednesday on the weirder bits of the donkey vote

  • 11
    Posted October 4, 2010 at 3:49 pm | Permalink

    Depends what you mean by donkey vote. In relation to preferences, the relevant point is always the flow down the ballot paper rather than any boost for being in first position. When I constructed the variable I described, the donkey vote was as significant as the Green decision on directing preferences to Labor.

    I have the suspician the high informal vote in NSW might be mucking up some dinkey vote models.

  • 12
    Posted October 4, 2010 at 3:51 pm | Permalink

    Why analyse ballot position against Green swing? You’d have to analyse it against change in ballot position, and that’s unreliable because of redistributions.

  • 13
    Posted October 4, 2010 at 4:07 pm | Permalink

    Antony – bloody good question!

    I had a squiz at ballot position when I was using it to build up a few different types of donkey vote variables. I accidentally regressed the green primary vote swing against it and found a p-value of 0.0001 and and R-Sqr of nearly 10%!

    You get spurious regressions every now and then, but not like that.

  • 14
    Posted October 4, 2010 at 4:08 pm | Permalink

    So a vote for the Greens is a vote for Labor except in Melbourne where a vote for the Liberals is a vote for the Greens.

    Interesting if depressing analysis.

    What is staggering isn’t just the number of seats where Labor came from behind on prefs from Greens but the scale of it.

    Why should a primary vote have the same value as a preference?

    37,000 Lab primary vs 42,00 Lib primary in a 96,000 electorate of Corangamite for example with a 3.3% informal vote.

    Cheeseman is very lucky. And so is Gillard.

    Mind you, she’s certainly paying the piper. And in spades.

  • 15
    Posted October 4, 2010 at 4:14 pm | Permalink

    Ah, that’s social science statistics for you. Rutherford’s gold leaf experiement, now that’s a statistical result that produced a new paradigm.

  • 16
    Posted October 4, 2010 at 4:18 pm | Permalink

    James K – the solution to that conundrum is optional preferential voting. You can’t assign weights to preferences without the count becoming almost impossible to conduct. Just imagine every time you excluded a candidate you had to assign a new transfer value to each ballot paper. Let the voters assign their own weights to preference by deciding how many preferences to give.

  • 17
    Posted October 4, 2010 at 5:41 pm | Permalink

    Possum – one variable that you may have missed is the ‘Bob Brown hates preference deals’ (http://news.ninemsn.com.au/national/election2010/7948797/greens-allege-dirty-tricks-in-qld) flyers that the Libs handed in Queensland and WA (and i assume other states), which encouraged greens voters to preference the coalition over Labor. Although who knows if it had any meaningful impact on the preference flows…

  • 18
    Kevin Bonham
    Posted October 4, 2010 at 9:31 pm | Permalink

    I still find it hard to fathom how anyone votes Green and preferences the Coalition.

    Loads of possible reasons. This list is far from exhaustive:

    * Some Greens voters are socially conservative as well as environmentally inclined. Because of the environmental preference they vote Green first and foremost, but then they give their preferences to the Libs because they see the Libs as more socially conservative.

    * Some Greens voters living in safe Labor seats like to give their preferences to the Libs to try to make their seat more marginal.

    * Some Greens voters actually think the Coalition is more environmentally conscious than Labor.

    * Some Green voters preference the Coalition to “punish” Labor for not being more environmentally conscious than it is.

    * Some Greens voters are upper-middle class and have environmental concerns but see Labor as being the lower-class party and just don’t relate to its traditional interests, or think that it relates to theirs.

    * Some Greens voters are ideologically converted to the Coalition cause but move in circles where voting for the Coalition is a no-no, so they can get around that by voting 1 Green 2 Liberal.

    Each of these would be only a small fraction of those Greens voters who preference the Libs. Many, I think, would be doing so because of candidate issues.

    What is staggering isn’t just the number of seats where Labor came from behind on prefs from Greens but the scale of it.

    The number of such seats itself was hardly staggering, as I mentioned above (though I haven’t checked the absolutely final figures to see if very late counting added any to the list.) But the scale of Labor comebacks on preferences – Brisbane was a great example, down by 15.5 points on primaries, final margin less than 2.5 (but still lost).

  • 19
    Rod Hagen
    Posted October 5, 2010 at 8:47 am | Permalink

    I know plenty of Greens voters, but to the best of my knowledge, none of them actually take HTV cards on election day.

    Mmmm. Green HTVs were certainly a popular choice amongst the punters at the booth I was at in McEwen, especially amongst younger voters. Long term Green voters might not bother with them, but there were quite a lot of “first timers” in the pack this time around!

  • 20
    Rod Hagen
    Posted October 5, 2010 at 8:54 am | Permalink

    I still find it hard to fathom how anyone votes Green and preferences the Coalition.

    I’m sure there are plenty of Liberal voters out there who like to think of their party as the one led by people like a Menzies, or a Peacock or a Turnbull, or even a Fraser. They don’t take kindly to Abbott and others of his ilk who run the Liberal shop at present. They vote Green accordingly, but can’t quite come at giving their pref to labor.

  • 21
    Rod Hagen
    Posted October 5, 2010 at 9:00 am | Permalink

    It would be interesting to know if there is any relationship between Green preference behaviour and the comparative age makeup of particular electorates.

  • 22
    Posted October 5, 2010 at 9:37 am | Permalink

    If you’re looking at the effect of Green HTVs, don’t you have to consider the issue of which booths actually had Green HTVs and which didn’t?? As far as I’m aware, the Greens simply don’t have the resources to hand out HTVs at most booths. Even my booth in Moreton, which registered one of the highest Green votes in the country, had absolutely no Green presence whatsoever (not even a poster). Isn’t that a relevant factor in an analysis of this kind? Or am I missing something?

  • 23
    John Kotsopoulos
    Posted October 5, 2010 at 12:51 pm | Permalink

    It is ridiculous that a 2nd, 3rd, 4th or even lower preference can end up having as much value as a 1st preference as the count progresses and the candidates drop out of contention. Optional preferential voting has some attraction but a better gauge of people’s wishes might be to require preferences to be expressed for their top 3 candidates with each preference having a value based on whether it is a 1st, 2nd or 3rd preference. All 3 preferences would be counted even for the last two candidates with 3 points for 1st, 2 points for 2nd and 1 point for 3rd.

    Under these circumstances it is feasible that a soft Lib or Labor voter could help elect a member from the other side. But so what. This is still their choice and a far better outcome than having someone elected who gets 20% of first prererences and lucks in because of the way the preferences fall.

    (My maternal grandmother who was also of European origin never got to grips with preferential voting and on one occasion famously declared that she could not see why she should give one candidate 3 votes and the others only 2 or 1. Her solution was to award all 3 candidates 1 ‘vote’ each.)

  • 24
    Andrew Norton
    Posted October 5, 2010 at 1:47 pm | Permalink

    “I still find it hard to fathom how anyone votes Green and preferences the Coalition. ”

    The Australian Election Study (not yet available for 2010) suggests that the Greens pick up part of the stray ‘I don’t like any of them’ vote. As other commenters have suggested, it is a mistake to think that voting Green necessarily signals green ideology.

  • 25
    Posted October 5, 2010 at 2:08 pm | Permalink

    Second, third and so on preferences count as much as a first preference, because in the end the only votes that have a value above zero are those that end up with one of the two final candidates.

    “Preferences” aren’t multiple votes. You have one vote, and in our system that vote is expressed as a complete ranking / ordering of the candidates.

  • 26
    Kevin Bonham
    Posted October 5, 2010 at 6:28 pm | Permalink

    This is still their choice and a far better outcome than having someone elected who gets 20% of first prererences and lucks in because of the way the preferences fall.

    Except that generally such a candidate needs luck, but it isn’t just luck that they win. For instance Andrew Wilkie, who is probably the paradigm case of the above rare feat, can be said to be a worthy winner because he is the Condorcet winner – based on these results, he is the candidate who is preferred in a straight two-candidate matchup with any other candidate in the electorate. It’s true that one other candidate (Jackson – Labor) got far more primaries than Wilkie did, but so what? That other candidate still got way less than half, and among those who didn’t vote for him, had very little support, so why should he win?

    As for the system proposed in #24 (3 points for 1st pref, 2 for 2nd, 1 for 3rd) it doesn’t actually acheive its desired goal, if I’ve understood it correctly. Suppose there are three candidates, Liberal Labor and Independent. Each major party candidate gets 45% of the vote, but the major party voters each put the other major party last. The Independent’s 10% of primary voters split their #2 votes. Each major party gets 45%x3+5%x2+50%x1=195% of the number of voters. The Independent gets 10%x3+90%x2=210% of the number of voters and wins.

  • 27
    Posted October 5, 2010 at 6:36 pm | Permalink

    Do we have any totals of voters who do NOT have labour, national or coalition in either of the first two spots?

  • 28
    Posted October 5, 2010 at 7:49 pm | Permalink

    Negotiations have been finalised between Labor and The Greens for preferences in the Queensland election. The Greens are directing preferences to Labor in fourteen seats, while receiving preferences in one seat – Indooroopilly.

  • 29
    Kevin Bonham
    Posted October 5, 2010 at 8:07 pm | Permalink

    Do we have any totals of voters who do NOT have labour, national or coalition in either of the first two spots?

    There wouldn’t be a way to work that out apart from direct sampling. Many voters would vote 1 Greens and then 2 for a more obscure minor party that had already been eliminated in that count by the time the Green was eliminated.

  • 30
    John Kotsopoulos
    Posted October 5, 2010 at 11:17 pm | Permalink

    Kevin if each major party puts the other party last and their supporters follow the card so be it. Caf my point is that a 2nd, 3rd or 4th or 5th preference should not carry the same value as a 1st. It is simply unfair.

  • 31
    David Richards
    Posted October 6, 2010 at 9:17 am | Permalink

    John – such a system is unwieldy and also feeds into the “wasted vote” meme used against minor parties, even if to a lesser extent than does OPV. Either you go MMP, or you stay with preferential voting, either full or OPV, with all votes counted with equal weight.

    I suspect you are coming at this from a pro LNP – anti Green/ALP stance.

  • 32
    Posted October 6, 2010 at 11:14 am | Permalink

    John #31

    It doesn’t – a 2nd – nth preference becomes increasingly unlikely to make it into the count that determines the final two horse race as the preference increases down the list.

  • 33
    Posted October 6, 2010 at 11:51 am | Permalink

    Thanks Antony GREEN.

    I’m a big fan of your work on our national broadcaster.

    Also glad to see you support Possum’s blog.

    I certainly prefer our system to PR and FPTP.

    Optional preferences does sound like a reform worthy of consideration.

  • 34
    John Kotsopoulos
    Posted October 6, 2010 at 1:33 pm | Permalink

    David Richards “I suspect you are coming at this from a pro LNP – anti Green/ALP stance.” Boy you sure have pulled the wrong rein on that one.

    Counting everyone’s 1st 2nd and 3rd preferences and distributing the points is hardly unwieldly. It’s a fairer system that applies in most other situations where a vote is called for.

  • 35
    Posted October 6, 2010 at 3:18 pm | Permalink

    If it’s good enough for the Brownlow it’s good enough for federal elections.

  • 36
    Chris Berkeley
    Posted October 6, 2010 at 3:36 pm | Permalink

    Another point to consider about Greens is that they are getting better organised and are, prima facie, much better organised than the major parties in some areas. For example ib Chifley/Parrammatt/Greenway there were a number of booths that were jointly serving two of these and some electors were voting in their third division in as many polls. There was mass confusion at such booths. The only party that had turned up with HTVs for both divisions was The Greens.

    For example at Blacktown South the Greens polled 5.68% (P’matta) and 11.09% (Chifley) (compared to 2.22 in 2007 when the booth was Chifley only). The Liberal helpers were people from outside the area being paid for the day and the ALP people were either geriatric or had a relatively poor command of English and no real local knowledge.

  • 37
    Rod Hagen
    Posted October 6, 2010 at 8:54 pm | Permalink

    Interesting comment, Chris. At the booth I was at in McEwen the Greens had three HTV people for most of the day, as did Labor. THe Libs had only one at a time (and the first of these found that his replacement was nearly three hours late). Similar situations prevailed at the booth at which I actually voted. Here there were 2 Green, 2 Labor, and 1 Lib.

    Not sure what the situation was elsewhere, but in McEwen I get the impression that both The Greens and Labor were far, far better organised on election day itself than the Libs, though , judging by my own letterbox, the Libs spent far, far more on glossy pre-election advertising than the two other parties combined.

  • 38
    Posted October 6, 2010 at 10:08 pm | Permalink

    Possum (or other smarties)

    Is it possible to infer what proportion of voters changed allegiances from the Libs to the Greens in Vic and SA? The Greens vote in those states went up more than the ALP went down.

  • 39
    Posted October 7, 2010 at 7:19 am | Permalink


    Unfortunately not – we can’t tell what sort of dynamics went on in the changes, just the total net changes themselves.

  • 40
    Posted October 7, 2010 at 11:30 am | Permalink

    Message to Labor = chasing the Coalition over to the far right ain’t workin’…and don’t let Rupert Murdoch run the opposition’s weekly campaign.

  • 41
    Posted October 7, 2010 at 2:53 pm | Permalink

    Needless to say I disagree with quantize whose silly leftism is impossible to caricature.

    I suspect Gillard’s harder-line conservative rhetoric on asylum-seekering illegal boat entrants and no carbon tax promise before the election saved her ass.

    Now she’s swung silly left again to keep her looney left base and the Greens (same diff) happy.

    She’s proved to be no less dishonest than her predecessor.

    Leftists win by moving away from their base in rhetoric (lying) and conservatives win by sticking close to their principles.

    This adherence to conservatism needs to be despite media pressure to have small ‘l’ Liberal leaders (as Opposition leaders only) and the media defamation of conservative leaders.

    JG introduces a carbon tax this term she’ll be carbonised toast at the next election assuming she lasts that long as leader.

    Thus endeth the lesson.

  • 42
    Posted October 7, 2010 at 3:03 pm | Permalink

    Yes, because a mere 70% of people want a price on carbon. Electoral suicide.

  • 43
    Posted October 7, 2010 at 3:10 pm | Permalink

    Show me the reference for that assertion autocrat.

    It must have some basis. It’s oft repeated.

    Rudd lost because of his initial rhetoric and the manner he backed away from carbon tax/cprs. Not because he backed away per se. Obviously Abbott was licking his lips to fight the election on that issue and with Rudd.

    I suspect you’ve been reading too much Hartcher/Keane drivel.

  • 44
    Posted October 7, 2010 at 3:18 pm | Permalink

    Correct JamesL your mindless regurgitation of extreme right rhetoric is always

    ‘Needless to say’

  • 45
    Posted October 7, 2010 at 3:40 pm | Permalink

    There have been lots of surveys that show this – do your own homework. Alternatively, just continue to be wrong all of the time.

  • 46
    Posted October 7, 2010 at 4:06 pm | Permalink

    There’s a pair.

    Classic numbskull leftyism.


  • 47
    Posted October 7, 2010 at 10:25 pm | Permalink

    I think we have our Lord Monkton audience member..

  • 48
    David Richards
    Posted October 8, 2010 at 9:35 am | Permalink

    JamesK is the resident village idiot. I simply ignore him.

  • 49
    Posted October 8, 2010 at 3:31 pm | Permalink

    That’s very good by your usual standards quatize.

    See what happens when you sleep on it before pressing the ‘Post Comment’ button?

One Trackback

  1. …] This post was mentioned on Twitter by David Olsen, Possum Comitatus, Tobby, CassieST, Dean Lombard and others. Dean Lombard said: RT @Pollytics: On Pollytics: Did Greens HTV Cards win Labor any seats? http://bit.ly/9s2koV Some regression work tells an interesting story […

Please login below to comment, OR simply register here :