Image for Election 2010: The campaign about nothing

What does it take for a modern Australian first-term federal government to get itself voted out of office, and has Labor actually done enough wrong to lose?  This two-part question is the key to trying to forecast the outcome of the 2010 federal election.  The answer to the first part is unknown, since it hasn’t been done since 1931.  The answer to the second part is probably not, but you really have to give them points for trying.  Campaigning against perhaps the most objectively thrashable Opposition of the last thirty years, Labor has spent most of the last four months in a mess. At the moment it appears they have pulled back from the abyss in time to avoid losing.

If there was any real rationale for dumping Kevin Rudd as Prime Minister over his inability to sell the Government’s message, we would have expected to see something markedly different from Julia Gillard.  Instead, we switched from a leader struggling to sell his message because of distractions created by poor policy and backflipping to a leader struggling to sell the same message for pretty much the same reasons, with the added bonus of a tasty string of leaks from an unknown source thrown in.  Rudd was the fall guy for policy failings that were not just his own but endemic to his leadership group, and may even have contributed to his replacement by Julia Gillard by too often agreeing with her.  With Gillard’s attempts to resolve the policy problems left by Rudd ending in a mixed record of one qualified success (the mining tax), one token attempt (border protection) and one abysmal shambles (the citizens’ assembly on climate change) it has not been possible for Labor to sell the leadership change as having really accomplished anything.  As they could have run on their economic record anyway, the Labor campaign has been little more than a banal exercise in tokenistic identity politics, reinforced by the ultimate in political bottom lines: vote for us, because at least we’re the other mob. 

At this stage, the Coalition deserve credit for trying to make the best of what they’ve got; the problem being that what they’ve got is not particularly good.  The party is still in denial about the reasons why it lost the last election, thinking that the 2007 vote was mainly about John Howard’s excessive incumbency and that very minimal repairs are now required to make the party electable under normal circumstances.  There seems little realisation that the party is still fundamentally distrusted on industrial relations and on basic service delivery.  The Coalition’s other problem is that to attack the Government as a risky proposition because of its spending and leadership instability, they have had to make their own leader appear stable and risk-free, which does not come naturally to Tony Abbott at all.  Hence we have the new Abbott, the supposed conviction politician who in fact changes his positions frequently and has spent most of the election campaign effacing himself on most of the issues he really feels most strongly about. 

I did expect that this election would present a striking contrast in personal styles, but with both leaders attempting to hide their personalities and moderate their views, and with relatively little real policy differentiation, the 2010 election has turned into the election about nothing.  It has been characterised by an unusually high level of meta-campaigning (eg the constant debates about whether to have debates at all) and by an obsession with irrelevancies of political celebrity.  The latter trend started when it was clear that Bob Hawke would play a major role in Labor’s re-election campaign, but that rather than contribute anything of much substance he would do so in the role of an ageing political rockstar type, famous more for being famous than for what he did as PM.  Following this, we have seen five of the six living ex-PMs and most of the former Opposition Leaders as well active in some way.  This not only reflects the lack of much real life out on the trail but also that the current party leaders are not up to the calibre of some of those who came before.  The ultimate devolution of the campaign’s reporting into silly not-news came when Mark Latham confronting the PM, ostensibly as a journalist but asking a question nobody but him cared about, was suddenly seen as a substantial test of prime ministerial style.

One of the most common complaints about this campaign, heard from everyone from Andrew Bolt to Michael Gawenda and doubtless further still, has been about just how boring and inane the campaigning has been, to the point that its very boringness becomes a thing to examine in its own right.  Warping the rock-stardom analogy a little further, and considering how political addicts view classic elections like Keating vs Hewson in much the same way as rock fans view classic albums, the 2010 election is the political equivalent of Bob Dylan’s “Self Portrait” album.  Whoever wins, whatever happens next, it can only come as a relief after such a lightweight and silly campaign. 

In attempting to determine who might actually win all this nonsense (which is much more interesting and difficult to follow than the campaign itself) there are two important objections to the idea of a Coalition victory. The first of these is that first-term majority governments don’t generally lose; the last seven in a row have been returned with reduced majorities, and even at state level there have been few cases of first-term defeat in the last 60 years.  The second is that the single most reliable predictor of a government defeat is that that government is trailing badly in the 2PP vote more or less throughout the campaign.  A rolling average worse than 48-52 suggests an opposition will win; anything higher than that and the government’s survival chances even if behind, off the back of the late-breaking undecided votes and the benefits of incumbency in winning marginal seats, are good.  And this government is not even behind at the moment. The government’s rolling average of all phone polls (see Pollytics Here and on the sidebar) dipped just below 50-50 briefly in May before Rudd was removed, and again a couple of weeks into the campaign, but is now tracking up towards 52-48.  Maybe this flatters them a little, but at this stage polls are pointing to a nationwide swing of about 1%, enough to cost the government some seats but not enough to be likely to stop it winning.

A seat-by-seat appraisal as examined through bookmakers’ odds (which in the case of final seat odds has proven to be remarkably accurate in the past) also shows the Coalition to be struggling to get its nose in front in quite enough marginals.  There are two ways to look at Labor’s starting seat tally, and one is to consider the 88 seats it notionally holds going into the election, while the other is to start with the 83 it holds right now.  Because much of the difference consists of Coalition seats that just scrape into “notionally Labor” territory via redistribution, I prefer the latter method this time around.  Labor starts with 83 seats and loses Lowe which has been abolished.  It is currently strongly favoured to pick up Greenway (NSW), which was redistributed in its favour, and McEwen (Vic) where the sitting member holding a trivial margin is retiring.  Labor is strongly expected to lose six seats at the moment: Robertson and Macquarie (NSW), Flynn, Dawson and Leichhardt (Qld) and Melbourne to the Greens (Vic).  The Liberals have moved into slight favouritism in Bennelong (NSW) and the leading bookies currently disagree about Hasluck (WA) and Lindsay (NSW) and have Labor just in front in Solomon (NT).  The Coalition would need three out of those four lineball seats for a hung parliament of any kind, but even winning all four would only be good for 74-72-1-3, which would almost certainly result in a Labor minority government.  So to win office even in minority, the Coalition needs to not only make all the above posts winners, but also break into seats where Labor is currently favoured.  The next line of targets includes seats like Forde (Qld) and Bass (Tas) where Labor has been around $1.70 (though Bass is finally shortening to aroung $1.50 as I write) and Corangamite (Vic) where it is slightly shorter.  But considering that Labor has been out at $1.70 in Bass for so long for no real reason other than that it was close last time, we have to wonder what basis there is for any hope for the Coalition in others at $2 odds or longer.  In any case, individual seat markets may re-align substantially in the last few days.

Although Labor’s position appears strong on a 2PP basis and OK on a seat by seat basis, and although Gillard is almost as heavily backed to be PM as Kevin Rudd was, media coverage is again dominated by claims of an unexpected result. These can be boiled down into three interlinked claims: that it will definitely be very very close, that a hung parliament is likely, and that the picture in the marginals points to a likely win for the underdog. 

All these claims are heard at virtually every federal election and generally turn out to be false.  This time I think the first two have rather more legs than they usually do, but are still being hyped too much, while the case being presently made for the third is badly flawed.

To start with closeness, it is in the interests of both parties to hype that the result will be close no matter what the reality, since governments don’t like to seem cocky while oppositions don’t like to seem uncompetitive.  Journalists also like to hype a close result because that makes the election much more interesting than if one side is winning easily, and some analysts like to hype close results as a way of reducing their margin of being wrong.  However, close results are reasonably rare whatever may be said about them in advance.  Since the famous 1961 election, in which the Menzies government averted a tied parliament and a likely fresh election by 130 votes in the final seat to be declared, there have been single-figure seat margins in 1969 (seven seats), 1972 (nine), 1974 (five) and 1990 (nine with one independent).  1993 and 2004 were especially prominent flops for the narrative of inevitable closeness but there have been others as well. 

This time, there currently seems a high chance that the election will come down to a small number of seats, but it’s possible that this will change.  Indeed, in 2007, one week from the election, seat betting made Labor favourites in only 77 seats.  They eventually won 83 and were very unlucky to not win a few more.  If Labor remains strongly favoured up til polling day, then there should be a realignment with some of the seats now considered to be lost for Labor moving back towards their camp.  However, assuming Labor lose the six seats they are currently widely expected to lose, that still gives them just 78, which is fairly close.  To do better than that they need to either save some of those seats or start chipping into Coalition marginals, as well as saving all their other seats that they are at risk of losing.  Apart from McEwen and LaTrobe (Vic), Sturt (SA) and Boothby (SA) the Coalition seats on margins of less than 4% are mostly in states where Labor is on the nose and swings towards it are unlikely.  Some seats on higher margins in Victoria might not be out of reach. 

If close results are a relative rarity, hung parliaments are even scarcer still. As Peter Brent points out  the hung parliament scenario is another old chestnut trotted out at election times to spice things up, or by commentators who want to sit on the fence, but it doesn’t happen all that often. No matter how close the election appears to be, the result has to land in a rather narrow range for a hung parliament to occur.  How narrow that range is depends on the size of the crossbench.  This election it will probably occur if Labor wins 72-75 seats, or 71 seats provided the Greens win Melbourne.  That’s a wider range than in most federal elections.  While there has not been a hung parliament following a federal election since 1940, in most of those elections the crossbench hasn’t even existed, and therefore for a hung parliament to occur the seat tally would have needed to be tied.  So the record of no hung parliaments for 70 years is a bit misleading in terms of just how unlikely one is now.  Even so, supposing that 73-73-1-3 is someone’s expected outcome of the election, the outcome has to only fail to match that by three seats in either direction for a majority result to occur.  A hung federal parliament is still unlikely at this election, or any other where the crossbench is just a few percent of the parliament, but for once there is a vaguely realistic chance that it could happen. 

Most of the worst analysis seen at this election has involved attempts to use marginal seat arguments on a state-by-state basis, often based on unrepresentative sampling.  The typical marginal seat argument runs along the following lines: although Labor leads in the 2PP, polling of selected marginal seats and/or the whole state in Queensland/NSW/perhaps WA has shown certain swings in each state, and applying those swings uniformly to each state, X seats in those states will be lost, while Y seats will be gained in Victoria and maybe South Australia, and X-Y exceeds the number required, therefore the Coalition wins a bunch of seats and government.

A good example of this argument was the 4000-vote Galaxy poll of selected marginals in the five largest states.  It is a shame to see such a large polling sample wasted on such a bogus extrapolation method as occurred in the analysis of this poll.  For instance, in NSW the poll polled Macarthur, Gilmore, Macquarie and Eden-Monaro, and on this basis predicted a 2.4% swing against Labor, based on which the Coalition supposedly takes Robertson, Macquarie, Bennelong, Eden-Monaro and Page as well as the notionally Labor seats of Macarthur and Gilmore.  But these seats are unsuitable as measures of a swing against Labor in Labor marginals, since they include two notionally Labor seats with Liberal incumbents, seats that Labor probably does not hold out much hope of winning anyway, compared to seats like E-M, Page and Bennelong where they are bound to be making more of an effort.

In Queensland the poll included Bowman, Dawson, Dickson and Flynn, but again two of these are Coalition-held.  On the basis of swings said to exist in these seats we are supposed to throw the Coalition all of the Labor-held seats of Longman, Forde, Leichhardt, Petrie, Brisbane and Bonner as well.  Never mind that all these seats except Leichhardt are clustered around Brisbane, while Flynn and Dawson are regional seats north of Brisbane, where attitudes to Labor both state and federal appear to be quite different and where the swing to Labor was enormous last time and hence is likely to trend back more heavily this time around. 

I could also mention the view that Galaxy underestimates the Labor vote by a point or so anyway (which makes a big difference to this sort of exercise), but hopefully my point about these marginal-seat polls is clear by now.  If reading a result off national 2PP figures and an expectation of uniform swing is unsound because of regional differences, then reading a result off state figures with an expectation of uniform swing within each state is unsound for exactly the same reason.  There are far more national polls, they don’t sample questionable selections of seats, and their combined sample size is far greater than that for these marginal-seat polling exercises.  Individual-seat polling of the marginals would be more illuminating, if only there was enough good sampling being done close to election day.  There rarely is.

A more interesting marginal-seat polling exercise has been the automated call survey by JWS Research, with a sample size of about 400 voters in each of 54 seats, a massive survey effort.  We have little experience of this polling method in Australia so it remains to be seen whether it stands up to scrutiny.  Of course, because of the margin of error of the samples, it is bound to tip some seats as wins for the wrong side, but overall this should cancel out.  The poll suggests a net gain of four seats for the Coalition, reducing Labor to 79, although it does not survey Melbourne, which Labor could lose to the Greens.

Another variant of marginal-seat analysis involves looking at individual seat bookmaker odds, adding them up, getting a number less than 76, and concluding that Labor is not favoured by the markets to win a majority.  A good example of this is Simon Jackman’s article in the Australian yesterday.  There are two problems with this kind of calculation.  The first is a thing called “longshot bias” - seat markets tend to overestimate the chances of the underdogs.  In an informed market, candidates at final odds longer than $3 hardly ever win, and candidates between $2 and $3 win a lot less often than probability theory says they should.  So while Jackman points out that a lot of Labor seats are losable, they are actually not as losable as the betting markets say. The second is that electorate results are not actually independent of each other.  If Labor has an average 80% chance of winning each of five electorates in the same area, for instance, this does not mean that the Coalition will probably win at least one of them.  Rather it means that Labor will probably win all five, but there is an outside chance that there will be an unexpectedly strong swing to the Coalition and that it will claim several of them together.  I prefer to look at how many seats a party is actually favourite in, noting the seats that are indicated as close, but assuming that the incumbent actually has better prospects in them than the markets indicate … especially after a campaign as unfathomable to the electorate as this one.

With Labor currently looking at a 2PP of at least 51, the question is why, if we assume such a 2PP, we are even considering scenarios under which the Coalition might win.  After all, history shows that if a party narrowly loses the national 2PP but wins the election, it will be the Government and not the Opposition (this has happened four times in the last fifty years).  Much the same pattern is shown at state level, although the Liberals did manage to lose their majority and subsequently office in South Australia in 2002 despite a 50.9% two-party preferred.  In general, though, governments are well placed to use the resources of office to manage swings so that they fall where they do less damage. A counter-argument made by Brent is that because of Labor’s replacement of Kevin Rudd they are not running on their record in this election, and that the campaign is more like a battle of two Oppositions, in which an advantage for the government in converting 2PP results to seats can’t be assumed.  Another possibility, however, is that the government does still enjoy incumbency advantages, but that they will show up as the undecided voters make up their minds in the final few days.  And if that occurs then the Coalition cannot win anyway.

The marginal seats argument appeared in a truly ludicrous form in an article by the usually interesting Christian Kerr headed Donkey votes to go to Coalition in key marginal Labor seats“Donkey Votes to go to Coalition in key marginal Labor seats”.  It’s hard to know where to start with this shocker, but firstly the donkey-vote estimate of 2% given in the article is certainly inflated, since in many federal seats an obscure candidate drawn top of the ticket polls less than that as a total of genuine plus donkey votes.  In some cases, it’s less than 1%.  Secondly, even if the donkey vote was worth 2%, that doesn’t mean it could swing all seats on a margin of less than that, since some of those would be seats where the loser had the donkey-vote advantage last time and hence if the donkey vote switched it would increase the margin, not negate it.  Thirdly if you read the article carefully you can see that it lists ten supposedly key marginals where the Liberals have the donkey vote, then says “the draws haven’t all gone the Opposition’s way” and gives a list for Labor which also totals, er, ten.  By the end of the article sixteen seats where Labor has the donkey vote have been mentioned compared to just thirteen for the Coalition and one for the Greens.  And still there’s more – the article bizarrely omits five marginals on a less than 1% margin, and Labor has the donkey vote in three of those.  Labor also has it in three of the four closest to lineball on current betting.  Unfortunately this article is systematic of the ingrained tendency of The Australian to publish pieces that bolster the chances of the Coalition whatever the facts, a tendency that is grossly demeaning and degrading the newspaper with the best-resourced and most detailed political coverage in the land.  For a much better view of donkey-voting trivia see Charles Richardson here.

This election is being contested following a major economic upheaval and with both parties having undergone recent leadership change.  Labor’s leadership change – the dumping of an incumbent PM just months out from an election – is unprecedented.  For these reasons we have to be a little cautious in applying the lessons of past election and assuming that Labor must win from this position.  But I am inclined to think that much that has happened in the last several months will have far less impact on voter behaviour than might be expected, simply because so little of it relates to the hip pockets of the average voter.  I therefore see the order of likelihood of plausible results as roughly follows: a close Labor outright victory, a fairly comfortable Labor outright victory with few if any net losses, a hung parliament, and a close Coalition victory.  An increase in Labor’s majority is faintly possible but would be a remarkable result after everything, and if the Coalition does win outright I doubt they can do any more than just fall over that line.  I currently expect Labor to just win outright, with about 78 seats, but it is always hard to forecast seat totals accurately.

The Greens vote is bound to increase given that neither major party wants to make more than token noises on climate change, both are attempting to sound hard-headed on border control, and the standard of campaigning by both is so feeble (not that the Greens have been all that inspiring themselves).  But polls showing Greens support nationally in the mid-teens just cannot be taken seriously, and show that most of the pollsters have systematic method problems in the way that they assess the vote for minor parties.  I expect the Green House of Reps vote to be around 10 or 11 points. 

For the seat of Melbourne, Greens candidate Adam Bandt is currently a very strong favourite in the betting.  He does have an excellent chance based on the assumption of a swing to the Greens in general, given that swings to the Greens tend to magnify in seats where they are already strong, and given the retirement of a legendary sitting member.  But Labor has done well by selecting a candidate well to the left, both because this takes votes from the Greens and because it gives votes to the Liberals, and the best thing Labor can do in Melbourne is give away as much of its vote to the Liberals as it can and attempt to thus relegate Bandt to third place.  It is well known that when a well-regarded long term sitting member retires, up to five points of their party’s vote goes out the window.  But in Tanner’s case, will that vote go to the Liberals or the Greens?  I would like to see some polling of this electorate to see what might be really going on here, because while there’s a strong objective case that Bandt should just scrape over the line if he has campaigned well enough, $1.42 does seem way too short given the paucity of data.

In the Senate, Family First will lose their Victorian seat, and the Greens are very likely to make gains, but they are not likely to pull off the seven-seat clean sweep predicted by Roy Morgan’s perennially unreliable Senate polls.  Especially, I doubt they will win NSW, and several others depend on how the ball bounces in terms of the distribution of the other parties.

Tasmania

There has been less polling in Tasmania than normal, with only one full EMRS poll to date; I am unsure if there will be any more.  Peter Tucker has summarised and analysed the EMRS polling, which is also available in the Monday edition of the Examiner.

I think the EMRS figures are perhaps a little more positive for Labor than the likely reality.  We know that EMRS tends to overcook the Greens vote except when it confines results to firm support, and redistributing undecided votes to the Greens (then passing them mainly to Labor as preferences) is unrealistic.  Thus, while EMRS gives the state margin as 60-40, I think 59-41 or 58-42 is a more likely interpretation.

Nevertheless, there is very strong evidence in the poll, combined with other known polls, that Labor is on track for a 5-seat clean sweep.  There should be a swing to the Greens from their 13.5% result in 2007, simply because that result lags so far behind what they got at the most recent state election, and I think this swing should be about three points.

The following are my rough guesses of how the seats will pan out at the moment.  They will probably not be very accurate as they are based on very limited polling information.

Bass  

Polling continues to show no evidence that the Liberals will win this seat and continues to suggest, as I have argued before, that Bass is artificially super-marginal and may not be quite as close as it looks.  In my view Geoff Lyons will very probably win this seat with a substantial increase over Jodie Campbell’s margin (since the Liberals no longer have the Ferguson effect) but I also don’t think it will blow out quite as far as the 57-43 that has been showing in some polls.  I predict Labor 41 Liberal 41 Greens 17 CEC

<1 and a 2PP of about 54-46. As previously noticed, why Labor has spent much of the campaign at $1.60s and $1.70s in this electorate but almost as short in Braddon as in Franklin is pretty much beyond me.

Braddon  

After a brief flutter of interest at the height of the Rudd mining tax debacle, there’s not much sign of life in this one either, and in this case Labor have the incumbency advantage back and have it in the hands of a veteran local member.  Sid Sidebottom will very likely hold this seat and also pad his margin out a little, but I also don’t think this one will go our as far as the 56-44 claimed by EMRS.  I predict Labor 46 Lib 43 Green 11 and a 2PP of again about 54-46.

Lyons

Dick Adams will win this seat with probably not much change to his margin.  There should be a large swing to the Greens, whose vote did not pick up as much surge in 2007 as it would have done because of competition from anti-pulp-mill Lib-turned-independent Ben Quin.  The 2007 figures after the exclusion of Quin – Labor 47 Liberal 36 Green 17 and a 2PP of 59-41 will do me fine as a lazy projection for this one, except that we should allow the Secular Party candidate at least one point (and maybe two, although I doubt it) in view of him having the donkey vote, so I’ll assume that point goes half off each of Labor and Green and then pretty much returns to sender from there.

Franklin

Franklin was an oddity last time with the shambolic replacement of Harry Quick and a strong campaign from Vanessa Goodwin resulting in the largest swing to the Liberals in the nation.  With a new Labor incumbent Julie Collins now entrenched, albeit hardly setting the world on fire, and against hardly the most dangerous Liberal opponent in Jane Howlett, the swing will return to sender and then some.  With Franklin recording a huge Green vote of 27% in the state election there is some expectation of a similar result in the federal election.  It won’t happen, but with an experienced well-established candidate with a good electoral record the Greens should do much better than their 14 points from last time. I’m trying Labor 43 Liberal 38 Green 18 Ind 1 for a 2PP of around 57-43.

Denison

The Liberals’ decision to preference Labor ahead of the Greens has pretty much killed the latter’s chances in this electorate, since they have no preference source to catch up to Labor to.  The Liberals have exactly the same problem and while their candidate has impressed me on the campaign trail, his belief that he might still scrape over the line somehow is simply not realistic.  Only Andrew Wilkie has the preference-getting potential pose a serious challenge to Labor’s well-managed if amazingly low-profile transition from Duncan Kerr to Jonathan “Who?” Jackson, and it has been interesting to see that Wilkie is now the second-favourite for the seat at $17.  Indeed, Wilkie is one of the shortest-price non-incumbent independents in the nation (John Clements in Parkes is a fairly serious threat to a sitting National and there are a few roughies in Calare, but that’s about all).  However I just don’t think Wilkie’s making the kind of massive and innovative campaigning effort it would take to milk the strategic voting possibilities in Denison for all that they are worth and seriously threaten to win, and I see no evidence that he can get within a bull’s roar of the third on primaries that he needs. 

Labor should lose several points because of Kerr’s long incumbency, perhaps more than normal because he was so good at keeping the Green vote down, and maybe a point or two to Wilkie as well, but they should still win easily.  The Greens will gain votes from Kerr, lose votes to Wilkie, and gain votes from a general swing and purple patch at state level, so should go up a bit.  The Libs could lose as much to Wilkie as they gain from Kerr, and should stay somewhere near the same.  It’s really hard to get all this to add up to just 100 but let’s say Labor 40, Liberal 29 Green 20 Wilkie 10 and SA 1.  The big uncertainty is Wilkie though; he surprised in the state election, and if the idea of tactically voting for Wilkie to try to get rid of Labor really catches on at the last moment he could conceivably tear strips off both the Green and Liberal totals.  I don’t think this will happen; I think Wilkie will be stuck in fourth place although in a straight contest with Jackson he’d go fairly close to winning, and I expect the Wilkie prefs to not break strongly between the major parties and the 2PP to be about 61-39.

Senate

Eric Abetz, Stephen Parry, Helen Polley, Anne Urquhart and Christine Milne will all be elected, and the Greens should have no trouble getting a quota in their own right this time.  The real contest is between Guy Barnett (Lib) and Lisa Singh (Labor) for the final seat. 

In analysing this contest we can basically line up the parties on two sides based on their preferencing decisions, and assuming below-the-line leakage won’t make that much difference.  On the left side are Labor, the Greens, Senator On-Line, the Democrats and the Secular Party, and on the right side are the Liberals, Family First, the Climate Sceptics, the DLP and the Shooters and Fishers.  Dino Ottavi is a wildcard as he cannot be voted for above-the-line this election, but he won’t get many votes in any case.  Basically, Guy Barnett gets back in if the “right side” gets three quotas (a little under 43%) and otherwise he’s done for and a recently defeated State MHA will become Senator Singh. 

With the Greens vote likely to be in the high teens, it’s worth considering the chance of a 2-2-2 result, but basically, there isn’t much.  Not only does this require the left side winning at least four quotas, but it requires the difference between the Greens’ tally (which would itself need to be a bare minimum of 20 points) and Labor’s tally to be not much over one quota (14.3 points).  Perhaps the gap could be as high as 16 points, because the left-preferencing micro-parties have all preferenced the Greens ahead of Labor, but that is still a bridge too far, since at the last election the same gap was 22.  It could come down, but not that much.

So it is really between 3-2-1 and 2-3-1.  I must say I’ve been a little sceptical of Labor’s chances given that the Liberal team contains three well established Senators while Labor’s contains only one (and Helen Polley isn’t all that high-profile compared with Abetz and Barnett). Also this is the same Liberal ticket that took advantage of the Latham forestry farce to poll 46.1% in 2004, fully four points above their party’s Reps vote in a year when the party wrested two Reps seats from Labor incumbents.  The gap is partly explained by the incumbency advantages Labor held in 2004 in the Reps seats, and these are lower this time given that they have new candidates in two electorates and a one-term MHA recontesting in another. 

But assuming there will still be some difference then with my projected Liberal primary of about 37.5 in the Reps, a Liberal primary of close to 40 in the Senate is still possible, and that puts the right side tantalisingly close to their three quotas, if we can assume a point or two from Family First and a little bit of shrapnel from the others.  However, if I am projecting the Liberal Reps primary a bit too high and EMRS are right about the strength of Labor’s vote in Bass and Braddon, then it is hard to see the Liberal Senate primary getting high enough. 

I also tried modelling the outcome on the assumption that the Morgan senate poll for Tasmania (Labor 42 Lib 29 Green 21.5 FF 2.5 Others supposedly 5 – not likely) was wrong by about the same amount it was wrong by last time, which was large. For this exercise I assumed that the Family First and Others totals would be the same as they were in 2007 - 2 and 2.3 respectively - and allocated the 2.3 points for Other as 1 point to the three left micros (ah, the amusement of referring to the Democrats as a micro-party) and 1.3 to the three right micros.  I then applied the errors for Liberal, Green and Labor votes in 2007 to the 2010 Morgan polling, and split the leftover votes between the majors.  This produced an estimate of 38.8 Labor, 38 Liberal, 19.1 Green, 2 FF, 1.3 right others, 1 left others.  On this model, the Liberals miss out by about 1.8 points.  All up, it does seem like Guy Barnett is in big trouble this time, but it should not be taken for granted that he will actually lose.  Many commentators have Labor very strong favourites to dislodge him but in my view it is not much more than a 50-50 chance.

We may not know the result of the House of Reps election on Saturday night, or alternatively it may be clear early enough that it is really a low-energy election after all with not a lot of seats changing hands in either direction.  Either way I intend to send comments through here on a regular basis as the count unfolds.  If there are late developments that affect my view of the likely outcome in the final few days, I will update these in the comments section.