Saturday, November 29, 2014

Victorian Election: Final Aggregate And Seat Model

 Live analysis here on a new thread from 6 pm on election night. 

2PP Aggregate using 2010 preferences: 51.7 to Labor
Polls point to highly likely but not certain ALP victory, general picture of modest swing and minimal to low seat turnover
Seat Projection: ALP 47 Coalition 41
(A very small number of these seats may be won by Greens or independents, but there is not enough objective evidence to back such wins.)

Going into the main voting day for the Victorian election, I have to say it's a bit different to elections I've devoted major modelling efforts to on this site (the 2013 federal election and this year's Tasmanian state election) in that there is still some room for doubt about who will win.  Although Labor has led in almost every poll in the last year and a bit, including 15 or 16 (depends how you count 'em) of the 17 polls released in the last six weeks, their lead is not much, and sitting-member effects left over from the last election make the deck a slightly unfriendly one.

The polls suggest that probably Labor will get home with a seat tally in the high 40s, but other still-plausible scenarios (in decreasing order of likelihood) are a fairly comfortable Labor win with around 50 seats (just maybe if the polls have herded a few more), a scraped and very lucky Coalition win, or some kind of tied or otherwise hung parliament.  Anything else (eg a Coalition win where they win the 2PP as well, or a really lopsided Labor win) would be a serious surprise.

The final polls

The previous article (Endless 52:48 Edition) gave results for what are believed to be the final shots from Morgan, Essential and Galaxy, though with Morgan one never knows.  The Morgan, as noted, had a 50:50 2PP and indeed on my own calculations the primaries of that poll would normally imply a 2PP more like 50.5 to the Coalition. 

Coming from a volatile poll series by an under-tested method the Morgan wasn't taken too seriously but it had company with the release of the final Ipsos, which came in as another 52:48 by respondent-allocated preferences, but by batched last-election preferences was 50:50.  (My estimate without the batching: 50.3% to ALP).  This was off primaries of Coalition 42 (highest in any poll in the campaign bar the final Morgan), Labor 35 (lowest by anyone other than Morgan) Greens 15 and Others 8.

The Ipsos leadership scores had Napthine keeping a +9 netsat (49:40), Andrews down a little to -1 (42:43) but Andrews, as in Galaxy, very competitive on the preferred Premier question (trailing only 42:44) especially for such a close poll.   It's tempting to take Andrews' strong Preferred Premier results as a smoke signal that preferences are going to favour Labor much more strongly than in 2010, but Preferred Premier is such a messy indicator of anything that I'd rather not play with that fire.

ReachTEL came out with a 52:48 2PP based on the largest and freshest of the final samples: 2155 all on Thursday night. Coalition 39.7 Labor 38.3 Green 13.5 Other 8.5.  The 2PP was released as 52:48 (down one from their earlier 53:47) but the 2PP differences as implied by the decimalised primaries would have been trivial, with the Coalition-leaning Others losing two points of which one went to Labor and the Greens.

Napthine's Good ratings are up 4.2 and his Poor ratings down 2.2 for a net ReachTEL personal rating of -2.6 while Andrews records larger improvements yet still comes out at net -7.4.  I have been inclined to correct these ReachTEL leader scores because of the influence of the "satisfactory" option, but the need to do so seems to be reducing.  It should be noted they could be a bit on the harsh side.

What's notable is that while Others voters dislike both leaders almost equally (Napthine -24, Andrews -26.1), Greens voters like Andrews slightly (+8.3) but are even more anti-Napthine (-60.8, yikes) than Labor voters (-39.4).  All you need to know right there about how many Greens voters will pay the slightest attention should any of those naughty open Green HTVs see the light of day.

ReachTEL shows that voters generally think Labor will win.  However, Coalition voters think the Coalition will win.  I have always thought these questions need to be worded with an explicit "This question asks who you think will win, not who do you want to win?" and perhaps if the voter picks their own party while it's 55:45 behind, a little extra "Are you off your rocker?" would be helpful.

Newspoll with a good sample size came out with a 2PP of ... do I even need to say?  The primaries were 40-39 for the majors (the fourth to get exactly that) but a rather sobering 12 for the Greens.  This is the equal worst result (with one Essential) for the Greens the whole campaign, and I'll predict (confidence 70%) that it will also be the equal most accurate.

Newspoll's leader netsats were -4 (41-45) for Napthine (his worst by a point), -5 for Andrews and another strong performance for Andrews as Better Premier, trailing by only four (37-41).

There is possibly garbled word that the Newspoll tightened in the last two days of the poll, though it would be difficult to distinguish tightening from bouncing in such cases.  If it's real and I can quantify this I may tweak my poll model slightly on Saturday morning to adjust for it.  But for the time being ...


I currently have a 2PP poll aggregate of 51.7% 2PP to Labor, off primaries of Coalition 41 Labor 38.6 Greens 12.1 Others 8.3.  These are rather similar to William Bowe's current BludgerTrack model (51.8 off 41.2-38.6-12.5-7.6).

It may seem odd that when we've had two 50-50s (including Ipsos by last-election preferences) and three 52-48s in the last five polls that I end up in the high 51s and not at, say, 51.2.  However, firstly I am weighting the Galaxy, Newspoll and ReachTEL slightly more heavily than Ipsos (who are an excellent polling firm worldwide but impossible to benchmark against Australian election results at this stage) and a lot more heavily than the Morgan SMS polls.  To the extent that the Morgan SMS polls could be benchmarked against other polls based on repeated evidence they displayed large house effects to the Greens and against Labor, and also high volatility in tracking for their sample size.

The other thing is that most of the 52s in the campaign have been 52s and a bit when I convert the primaries into last-election 2PPs.  I convert the last three (ReachTEL, Galaxy and Newspoll) to 52.6, 52.7 and 52.3 to ALP.

Readers may have noticed a brief flirtation with the idea of an "expected 2PP" that differs from the aggregate.  I've decided not to do that, because at the federal election I could have done the same thing, and I didn't, and what I did then worked.  My expectation then is that if minor party preferences prove extra-friendly for the ALP then this may well be cancelled out by house effects of polling overall, greater conservative resilience to bad polling or whatever.  So we'll see how this goes.

In comparison with the federal election I've found the subjective choices to be made in aggregating all this Vic data are endless.  When using so much data from little-tried poll methods there is no way to make fully objective choices at all times.

Seat Model

The seat model I published yesterday used an "expected 2PP" concept I've decided not to use as the main model, for reasons discussed above, although I have a nagging intuitive feeling that I should have done so anyway.  So this then is the nowcast version for a 2PP of 51.7:

(Probabilities for seats with third parties in contention are conditional on them not winning.)

This model was first outlined in the late October polling article.  Seats listed on the Coalition side of the table include those that are notionally held by the Coalition based on the 2PP vote from the last election.  Five of these are seats actually occupied by Labor members that have been redistributed (hence the ALP in brackets); if there was no swing from last election in those seats the Coalition would win them.  

The table gives expected probabilities of the side of the table the seat appears on winning on the condition that the 2PP is 51.7.  The commonest path to a 47-41 seat result is that Labor holds all the seats it occupies bar Ripon (which is probably less close than the estimate suggests), and wins Carrum (much as the Liberals insist that they will hold it), Mordialloc and Frankston.  The model gives Labor more chance of losing Bellarine than is generally being expected (and it is probably wrong in this).  It also gives the Coalition more show of saving more than one of the four sandbelt seats (Carrum, Bentleigh, Mordialloc, Frankston), or picking off a Labor seat somewhere, than it gives Labor of picking up seats further up the tree.  So while 47-41 is the median, we can easily imagine what might happen if the Greens win Melbourne and the Coalition holds another seat, or Labor loses Ivanhoe and so on.  A hung parliament at least, even on this 2PP, isn't completely out of the question.

I may later regret not using the "expected 2PP" concept, and moving off 48 seats just as others move onto it (I have a habit of making my next-but-last model better than my final one), but for the record the "expected 2PP" comes out at 52.6% to Labor and 48 seats to 40, which is not so greatly different anyway.  I can make the 2PP in the model 54.6 and Labor still only get 52.

Wonky paragraph alert: I think at this stage it's also good to try an unconditional model in which the underlying 2PP is allowed to vary, so as to come up with probabilities that look more like the real chances of a party winning each seat, rather than probabilities that are pinned to a specific 2PP.  Undecided voters choosing on the day will often break one way or another compared to other voters.  To do this right involves digging through past data and then modelling variables that are under the spell of not one but two normal distributions.  I've run out of time for that sort of maths so as a crude substitute I've asked what would happen if I made the standard deviation for each seat five points instead of three.  I then get this:

I think for the closer seats this mostly gives a better rough idea of real-world priorities, though a few individual cases stand out as inaccurate (neither Mordialloc nor Ripon are probably that close, nor Bellarine for that matter).  For the moderately safe seats the model looks under-confident and I'd expect it to be (eg there is not nearly a 5% chance Denis Napthine will be unseated!) Some people like to get probabilities like this, run them through Monte Carlo simulations and report that Labor wins 95% (or whatever) of them, but such models are usually bogus because seat probabilities are not independent of each other.  Trying to get a really objective polling-based fix on Labor's chances would take a lot of historical state polling work.  I suspect such a model would still give them at least 70%.  

We should be extra-cautious about the idea of the Coalition winning at all because of the big-picture issue of federal-state drag and the messy nature of its term, which are enough to imply a likely moderate defeat despite it being a first-term government.  But I do think that Labor's campaign has been riskier than the typical opposition trying to get into power, with its stance on the East-West link hardly one from the modern playbook of "selective differentiation". 

Anyway, this is my broad expectation: a modest swing with minimal seat turnover.  The Coalition to probably gain Ripon and count themselves lucky to pick up anything else, Labor to win most (maybe all) of the sandbelt seats and not a lot (if anything) else.  My model differs from the bookies only in the seat of Bentleigh, and I won't be putting any money on that one.

As for the Greens I think we should take the evidence of the scale of doorknocking and resources thrown into the seat of Melbourne seriously.  But it's objectively going on out on a limb to say, with only one small, commissioned and unreliable seatpoll, that there is enough evidence the seat will fall against the backdrop of a modest if any increase in the Green vote. 

Friday, November 28, 2014

Victorian Poll Roundup: Endless 52-48 Edition

2PP aggregate: 51.8 (-1.4) to ALP by last-election preferences (52.6 expected preferences)
Current seat projection based on aggregate: 48-40 to Labor

This is probably the second-last of my posts on Victorian polling prior to the main day of polling for the election on Saturday.  I will be out on a field trip in the day on Friday.  Another post is expected either overnight Friday or Saturday morning - once I am satisified we have the results of all the major polls we're going to get.   Then it will be on to the live coverage on Saturday night from 6 pm.  I cannot promise you fancy graphics or dancing swing charts but I hope my live blogging will be of some interest.

(Speaking of things of some interest concerning Victorian elections, a plug for Adam Carr's page on early Victorian elections 1851-1864!)

This week's polling

Lately we've had new state polls from Morgan (twice), Essential and Galaxy.  The first Morgan SMS poll taken from 21-24 Nov and with a sample size of 1152 produced a 52:48 2PP off primaries of 39.5 to the Coalition, 33.5 to Labor, 17.5 to the Greens and 9.5 to others.  This series had favoured the Greens by an average of at least five points compared to other pollsters (indeed 17.5 was its lowest reading for them so far), so it was hard to tell whether this result signified that trend waning slightly, or that it was actually a not so good sample for the party.  The series also clearly skews against the ALP compared to other pollsters.

The Morgan was widely reported as showing a narrowing in the election race but this was mostly fallacious - the drop of three points from the previous Morgan was from a result that was abnormally high compared to other pollsters.  The Morgan showed Denis Napthine recpaturing a slim preferred premier lead from Daniel Andrews, but it is difficult to compare Morgan's readings to other pollsters who allow an undecided option.

The Essential poll represented three weeks of data rolled together (Nov 7-24) with a sample size just below 800 and had a 53:47 2PP (a correction of the 2PP to 52:48 turned out to be incorrect.)  The primaries of 40 L-NP, 39 ALP, 13 Green, 8 Others were identical to an earlier Galaxy that had a 2PP of 52:48  (the difference may be explained by rounding and/or different 2PP calculation methods).

The Galaxy of which details are just emerging also has those same primaries and again has a 52:48 2PP!  Significantly though, Denis Napthine's preferred-premier lead over Daniel Andrews sinks to a mere three points (41-38), a lower than expected lead for that set of voting intentions and a further indication that the Coalition's attempts to brand Andrews as an unsuitable, dangerous or unelectable leader have more or less totally failed. 

Finally, the last Morgan SMS threw a minor curveball with a 50-50 2PP (the first such from anyone for a very long time) off primaries of 44 Coalition, 36 Labor, 13.5 Green and 6.5 Others.  While Morgan says Labor are favoured to win off those results, any model that takes personal votes into account will show that 50-50 would probably not be enough to win.  As usual any unexpected Morgan finding is accompanied by an essay explaining why the poll respresents the absolute truth of the electorate rather than one about it being perhaps an outlier.  Morgan even claim that John Howard's appearance on the campaign trail (in a state that never much liked him) diverted voter attentions from the Abbott government.  Fortunately I am saved from having to express any original ridicule upon this point by this:

Attitude questions in the Essential poll are reported in the full report.  The first thing to notice here is that the inclusion of the Greens in the "trust to handle issues" section disadvantages Labor compared to the Coalition, since many Greens supporters would trust Labor over the Coalition on the listed issues.  The Liberals' leads on energy, public transport and support for regional Victoria are thus illusory and their other leads are closer than the poll shows them to be.  That the Liberals are preferred on the economy, roads and cops and Labor on health and education can all be filed under business and normal, so the stat that really deserves attention is Labor's lead over the Liberals on "unemployment and jobs".  So often in Australian polling lately we've seen the economy and jobs marching together in the service of the Coalition.  Not this time.

The Galaxy showed a thumping 60-29 support level for the proposed East-West link, but a tweet from Antony Green shows you all you ever need to know about the way this issue plays out:

The outer suburbs like the link but the seats where Labor is trying to fend off the Greens cannot stand it.  This probably isn't just NIMBY mode, but also that green-leaning inner city dwellers often aren't mad-keen on large-scale road projects wherever you might build them. If Labor loses, it will suddenly be immediately obvious to everyone that the East-West link was why.

Pre-Election Exit Poll Ahoy!

In an exercise I've never seen before, Galaxy have been commissioned by Trades Hall Council for an exit poll of early voters (of whom there are a hell of a lot this time around - I'll probably cover that on Saturday).  The poll of the four sandbelt seats of Bentleigh, Carrum, Frankston, Mordialloc shows a 2PP of 52-48 to Labor from a sample size of 602.  In 2010, early voting notably favoured the Coalition in both Bentleigh and Mordialloc, and might have done so by a shade more had there not been a late swing to the Liberals.  All up, the early votes in these seats in 2010 came in at 53:47 to the Liberals.  So that's an average five-point swing, if the poll is reliable. (It's a commissioned poll, hence caution, but the previous THC-commissioned Essential polls all looked OK to me.)

The pattern of early voting favouring conservatives is nothing unusual and given that this poll has been taken before the early voting period ends, it might be expected that this underestimates what Labor should get on the day.  On the other hand it might be that since more people are voting early, those doing so are more representative than usual.

In any case, there's no warrant on this poll alone for assuming Labor will win all four seats.  The sample size would only be 150 for each and it may well be there is variation between them.  Also, given the four-point MOE for the sample overall, it's still possible on this sample that the Liberals might be holding most of the sandbelt.  But, as always, it is better not to poll such adverse results than to poll them.

The other thing highlighted here is that so many votes will be in the can by polling day.  With record early voting (looking like it will settle at thirty-something percent) the impact of any very late swing will be muted.  Unless the polls are almost all getting it wrong, a quarter of the vote is already in the can for Labor and the Liberals need to get enough more on the day to counter it.

Are The Polls Herding?

In the recent US mid-terms polls displayed a strong skew to the Democrats, shown by Nate Silver to be potentially causable by herding. (Silver's article is very well worth reading in full.)  After seeing so many 52:48s in this campaign (seven of the last 14 polls!) I've been running some fake-poll Monte Carlos to see if the polls so far do look too tightly clustered.

Fortunately, I've been unable to find clear evidence of the polls being too tightly bunched together, though at times the percentage of simulations that are more closely bunched than the actual polls has been small enough for it to be close to statistically meaningful if I was looking at a random election (rather than one I was already suspicious about.)  Morgan has been volatile while Galaxy and Essential have been their usually under-dispersed selves, and really that's been about all there's been to it.  We also know from other elections that while Essential does tend to move slowly, it also goes off on tangents compared to other pollsters.

The spread of poll readings has been less than expected by random chance, but not enough to be really suspicious.


Throwing this week's polls into the mix and dropping old polls I'm now looking at the Essential 53, the Morgans 52 and 50, the Ipsos 54 and the two Galaxy 52s.  My aggregate weights Galaxy very heavily and the Ipsos and Essential are downweighted for age, with the Morgan 50 weighted higher than the 52.  (I've used my usual formula, except that I've declared it to be Saturday already.) It shouldn't thus be greatly surprising that the 2PP by 2010 election preferences comes out at 51.8, especially given my corrections to the Green vote.  This is off primaries of Coalition 41, Labor 38.8, Green 12.3 Other 7.9.  The difference between the majors looks a little high to me and will probably go down if the Newspoll and expected Ipsos pour cold water on the Morgan, which I predict that at least one of them will.  But for now that is where  have it.

The handling of the Green vote is extremely difficult.  Established pollsters like Essential, Galaxy, ReachTEL slightly overstated it in the federal election and have consistently had it about 13 leading up to this one.  Ipsos have had it much higher in two polls so far but are new, and Morgan has had it extremely high until the final poll which is conveniently close to the others.

The shift in the Morgans has been so strong that I wonder if they have herded any assumptions to correct the feral overstatement of the Green vote in their early polls, and for this reason I am reluctant to ping the 13.5 in their last poll as much as I have been pinging the others.  I would be less cynical about all this were it not for Morgan's habit of switching from Labor-skewed methods to non-ALP-skewed methods in the shadows of the post at federal elections.  Still, I guess no-one could accuse their SMS series of a pro-Labor skew!  Indeed if the change in their Green vote is due to any methods tweak then this may have skewed the 2PP of their latest poll, since their last five SMS efforts have had the lowest ALP primaries of any pollster for a very long time. 

Preference Strength Again

The other big issue is the strength of preferences.  While I have noted the 51.8 aggregated 2PP based on preferences from the 2010 state election, the general view among psephologists is that the strength of preferences to Labor will be stronger than in 2010.  I covered this in the Late October roundup, but a further development since has been preference-dealing for this election.

In a review on the Crikey site (paywalled) William Bowe noted that Labor has done preference deals for Upper House preferences with the likes of the Country Party and (in a sure sign that hell has frozen over) the DLP.  Labor is however struggling for preference flow in Frankston and a vengeful ex-member wreaking havoc in Ivanhoe. 

Minor party voters aren't strong HTV card followers generally (the follow rates are around 40-45% for the major parties, but significantly lower for micros because Greens voters tend to make up their own minds on preferences and micro-parties tend to struggle for full booth coverage).  So maybe the card games won't matter too much, but I'm expecting changes in preference flow anyhow.  To be cautious on this point, for both Greens and Others I'm assuming as my core model that we will see a preference flow change that is about 60% of that seen between the 2010 and 2013 federal elections.  That brings us to a 2PP of 52.6% (and it may well go up if the Morgan is an outlier).  If I assume preferences exactly as per the federal election (excluding preferences from voters for Independents because of the McGowan tactical voting factor) this goes up to 53.2%.

The other big preferencing story is that the Greens lodged open HTV cards in every electorate and the Liberals in the four inner-city seats eyed off by the Greens.  Lodging of a card doesn't mean that card will be handed out, but means it can be.  If these cards are handed out, the flow of Liberal preferences to Labor in the inner city seats could decrease sharply, and this could well cause the Greens to at least win Melbourne.  On the other hand, so few Greens supporters are influenced by how-to-vote cards that the damage if the Green open cards are handed out will be limited.  What exactly the Liberals are up to with their open card is a big mystery, but we will find out on the day whether this is a sneaky plan to gift Labor an unwanted crossbench, or whether it was just all mischief-making and mind-games.

Seat Model

There's not been much seat polling so my model is pretty much starved on that front.  While I am not directly including data from commissioned polls I do feel the THC exit poll justifies a slight watering down of my loading of the Galaxy seat poll of Bentleigh.

This is the current output for a 2PP of 52.6%:

It's not quite time to switch this over from "nowcast" to "forecast" mode just yet but I'm not expecting the late polls will change it greatly.  I won't be surprised if they are ALP-friendly enough to kick it up from 48 seats to 49, in which case my suspicion is that rather than Labor holding Ripon, Labor will more likely pull off some random seat from the 5-8% layer that the Liberals are apparently now trying to sandbag.  Indeed it's quite possible they could get a couple of those and miss one or two further down the tree.  Almost certainly, the underdog according to the seat model will get up in a handful of seats. 

We still don't have reliable polling information on which to judge the chances of the Greens picking up a seat, and none to judge any independent chances in Morwell (Tracie Lund) or elsewhere.  I can only stay with my previous impression: Green chances in Melbourne should be taken very seriously, but other electorates are not easy for them. 

(If we stick to last-election preferences strictly, then the model likes 47-41).

Quick Betting Watch

I don't have time to do a thorough betting roundup tonight but Labor remain heavily favoured, with the $5.50 or so on the Coalition indicating some caution about the relatively close state of the polls.  The contest between seat betting and polling models is a fizzer this election as we're all getting about the same thing.  The Sportsbet exact seat market has 47-48 Labor seats as a narrow favourite over 49-50 (bear in mind this factors in chances of losses to the Greens) and projects an average of 48.7.  The Greens are still at a mere $1.55 to get nothing, with the market still expecting their primary vote to be below 12%.  The projected 2PP has come up to 52-53.  The ALP are favourites in the same 48 seats my model has them favourites in other than Ripon which as noted will likely be an override anyway.  Tracie Lund in Morwell is at $8.  My guess is the bookies have had people working for them setting initial odds using very similar models to this one, and no-one is betting enough otherwise to push them off those. (Oh, and sadly the Mordialloc arbitrage eventually disappeared, with Labor favourite on both markets.)

Everything (external polls, reported internal polls, pseph models, historical models, betting, behaviour of the parties and commentary) is all pointing to a highly likely moderate ALP win.  Are these expectations all herded together and is the chance of something unusual happening higher than it appears?  It doesn't look like it and it's hard to be surprised at reports that some Liberals just want it all to go away.  But much will be revealed on Saturday night, and most of the rest when the early votes are counted around Monday.  Back with more Saturday morning.

(Oh, and in possibly the biggest surprise of the campaign, The Age has, albeit backhandedly, endorsed the Coalition!)

Tuesday, November 25, 2014

EMRS: Liberals Fall To Four-Year Low

EMRS: Liberal 42 (-4 since September) Labor 31 (-2) Green 19 (+3) Ind 6 (+3) PUP 2 (+1)
Interpretation: Liberal 43 Labor 34 Green 17 PUP 2 Others 4
Likely outcome based on this poll "if election held now":  Hung parliament (12-9-4) with slim chance of Liberal majority (13-9-3)
New aggregate of all polls: Likely Liberal majority government (13-9-3)

A new quarterly EMRS poll of Tasmanian state voting intentions has been released.  See also the excellent trend tracker on the EMRS site.  This poll continues a downward trend in this pollster's readings noted in my previous EMRS article (Closest Gap In Four Years) and sees the polled gap between the major parties close further and the Liberal vote at its lowest level since November 2010.  It's a rather poor poll for the new Hodgman government that suggests it has significantly lost support since winning office just eight months ago.  If this poll is to be relied on, the government would be in danger of losing its freshly minted majority in an election proverbially held now.  The next election is not until 2018 so there is not too much for opponents to get excited about yet.

As well as the usual EMRS habit of overstating the Greens vote, the vote for "Independent" in this sample (at 6 points) is too high, and this is a common issue with polls that include "Independent" as an option by itself in the readout.  (Voters may intend to vote for indies, but when there are no high-profile independents that suit their politics, most of them then don't.)  Nonetheless this figure for independents does suggest voters are looking for fourth-party alternatives.  It is the highest figure for "Independent" in EMRS polling since May 2012, and there is a history of this figure being higher in the pollster's May samples because of interference from Legislative Council elections held in that month.

The poll shows Will Hodgman still with a large (and relatively meaningless) lead over Bryan Green as Preferred Premier, 50-22, with 28% (up five) choosing neither.  Preferred Premier scores are usually skewed to the incumbent, especially against new Opposition Leaders, so there's not too much to see here.  If it's still like this in, oh, mid-2016, it might be time for Labor to be more concerned.

The poll echoes some curiously bad results for the government in two very small Morgan SMS samples taken since the last EMRS and ReachTEL polls, and suggests that those samples, while very likely skewed against the government, were not as skewed as I at first suspected.

EMRS Poll Translated to Seats
The past record of EMRS is that the pollster tends to overstate the Green vote, understate the Labor vote and get the Liberal vote more or less spot on.  Also the poll has a high undecided rate even after prodding unsure voters, in this case 11%.

With some corrections made for these tendencies, I model the votes based on this poll as Liberal 43 (-8 from election) Labor 34 (+7) Green 17 (+3) PUP 2 (-3) Other 4 (+1). A straight swing of that magnitude would result in the Liberals losing seats to Labor in both Braddon (where the Liberals remarkably won four seats in March) and Franklin.  The Liberals would also probably lose their third seat in Lyons to the Greens.  However if the swing was not quite uniformly distributed, the Liberals might hold their third seat and keep a majority, especially if the distribution between the three sitting Liberal MHAs was as uniform as it was last election.

An election being held with these kinds of primary votes is a very artificial scenario.  We have seen in the past - a good example being the Lennon Labor government in 2006 - that the party that leads in the polls can make the case that only it can win majority government.  So if the Hodgman government went into an election campaign with this sort of voting intention base, it's quite possible voters would come back to it rather than risk causing another minority government.

Morgan Was Here First (Sort-Of)
I didn't cover them in much detail at the time because I wasn't sure what to make of them as they were so different to other published polling, and as the sample sizes were extremely small.   But it is time to note that Tasmania has another state voting intention player now with the return of Roy Morgan Research after an absence of several years.  Alas, it's not a return on the scale of the pollster's old Tasmanian phone polls, but if we can make some kind of sense of it, it will be better than nothing.

Two very small 300-vote SMS samples have been included in the pollster's recent state polling releases.  The first in late September produced results of Liberal 39.5 Labor 33.5 Green 19.5 PUP 3 Ind/Other 4.5.  The second in late October produced results of Liberal 41 Labor 34 Green 19 PUP 2.5 Ind/Other 3.5.  The second is very close to the current EMRS poll.

At the time I treated these polls with vast suspicion because of their very small sample size and a history of apparent pro-Labor skew in Morgan's small Tasmanian federal samples (conducted by a combination of SMS polling and face-to-face polling.)  While Morgan's face-to-face polling is known to skew slightly to Labor (their SMS polling apparently doesn't), I suspected that the skew in Tasmania's case might be greater and that there might be issues with the Tasmanian SMS panel specifically.  This is still probably the case, but with the EMRS poll providing some evidence of the government genuinely losing support over the past few months, the Morgan samples are looking a little bit less outlandish now.

Morgan has been doing a lot of SMS sampling in the lead-up to the Victorian election and its samples have been massively over-estimating the Green vote compared to everyone else's.  At the moment I am conservatively estimating that Morgan's Victorian sampling has had the Green vote five points too high there on average, but we'll find out on the weekend.

As a first attempt at benchmarking these Morgan samples off my corrected readings of the two EMRS polls either side of them, I reckon they have the Liberal vote about four points too low and the Greens vote about four points too high.  Because of the small size of the Morgan samples these estimates may not be very accurate and will need to be fine-tuned with more evidence.  For instance if this EMRS poll is itself affected by sample error in Labor's favour, then it is possible the Morgan samples are more erratic than I've suggested.  But at least there is some basis for roughly using the Morgan data in my state aggregate now rather than throwing it away entirely.

New Aggregate
My Tasmanian aggregate was last revised in mid-September following a very large-sample-size ReachTEL that contradicted the early September EMRS and showed the Liberals still with a hefty lead.

Considering that the two Morgan samples are by a method that has never been tested in Tasmanian elections, are a smaller sample than the EMRS when combined, and are older than the EMRS, I can only give them a very small weighting (I've gone with giving the two combined one-fifth of the weighting of the EMRS, though even this is generous.)  I've weighted the new EMRS at 37.5%, the October Morgan at 4.5%, the September Morgan at 3% and the previous aggregate value at 55%.  There's no new electorate breakdown data so I've applied uniform swings to the previous electorate values, which were taken partly from the ReachTEL and partly from the state election.  Here's the aggregate:

The aggregate gives the fourth Braddon Liberal seat back to Labor (and will probably keep doing so all the way to the election).  It also now gives Labor the Liberals' third Franklin seat.  On these figures the Liberals with three evenly-matched incumbents and a healthy flow of fourth-party preferences should retain their third seat in Lyons, but it's not that far from being lost.  There are two Labor-Green battles: one in Bass and one in Lyons.  Although Labor leads the Greens by 0.1 quotas on party score in the Bass aggregate, and the Greens lead Labor by 0.1 quotas in Lyons, I suspect the result in both cases would more likely be the incumbents (Green in Bass and Labor in Lyons) retaining, because of profile and leakage issues.  Anyway one-all seems a fair division of these two contests so my new aggregate of all state polling is Liberal 13 Labor 9 Green 3.

Why is it so?
Well really, I don't know.  It's easier to explain data you expect to see than data you don't.  We still have to see more polls to see if this one is exaggerated, but I also said that last EMRS (and that one was four points higher.)  They new government has pretty much governed as it intended to do, even if it did make an abysmal mess of anti-protestor laws (finally about to pass in a greatly watered-down form after the initial draft went way too far.)

If anything, these lacklustre polling results (combined with the drubbing for the party's candidates in the Legislative Council elections) suggest the last election was just about chucking the previous mob out and that Tasmanian voters don't actually want the Liberals to keep their promises to get tough on public sector spending.  Perhaps voters just want a harmless government that's vaguely pro-development, competent and not in league with the Greens.  As seen at federal level, it doesn't seem that easy any more for new governments to surf into office and get months and months of 50+% polling and uncritical reviews.  The honeymoon period aint what it used to be.

Update: Another New Morgan Sample

Another new Morgan SMS sample is out with a sample size of a feeble 281 and figures of ALP 41 Liberal 39 Green 13 PUP 2 Ind/Other 4.5.  Not only is the sample size tiny but this poll provides still further evidence that this polling form skews to the left in Tasmania, though the low Green reading this time suggests the skew is coming on average from both Labor and the Greens, and not just the Greens as previous samples suggested.

I've redone my calculations of the aggregate above with the new Morgan included and the one two months ago removed, with appropriate re-weightings and reassigned house effects, but the impact was so small as to not be worth reposting.

I do hope any journalists reporting this Morgan "result" will take note that even if the sample had no house effects the figures would still have a margin of error of 6%.  As it is the strong evidence of house effects in these polls implies they could be even further out than that.  And the pollster's claim that Labor would "win an election easily" based on those unlikely figures is false; it would more or less certainly be a hung parliament. 

Thursday, November 20, 2014

Comments On The Lambie/PUP Split

This is what Jacqui Lambie's Facebook page looked like before she joined PUP (image: @AutumnalMonk)
In recent weeks the always edgy relationship between the Palmer "United" Party and its Tasmanian Senator Jacqui Lambie has spiralled out of control.  Lambie is now effectively an independent.  The party no longer has any influence on her Senate votes, she no longer has any influence on its Senate votes (and is currently banned from the party room), and there at present appears little prospect she'll be preselected for it again (if PUP even exists in 2019 or whenever else Lambie's seat next comes up.) Her photo has been removed from PUP's website, she has removed its branding from hers, and Lambie and party founder Clive Palmer are insulting each other publicly with no discernable restraint.  PUP is a strange party so one would not be too firm in predicting anything here, but this sure looks like an ex-relationship. 

Tuesday, November 18, 2014

Poll Roundup: G20 No Help To Government

2PP Aggregate: 53.1 to ALP (+0.8 since last week, +2.1 in five weeks, highest in four months)
ALP would easily win election "held right now"

A fairly quick update as I have a lot of other work to do, which may sometime over the weekend or so include putting this piece of pseudo-scientific garbage in its place.  This week the federal government hosted the G20 meeting in Brisbane.  This would have been expected to at least allow the government to dominate the media cycle and demonstrate its international credibility.  However, things haven't been that simple.  Climate change was seen to dominate the agenda (probably to a greater extent than it actually did), and the isolation of Vladimir Putin by various Anglosphere leaders was still an anticlimax compared to the threatened "shirtfronting".  Even when Australia signed a generally well-received (if probably not well understood) free trade agreement with China, the Prime Minister still ran into Parrot problems.

But doubtless the worst moment of the conference for the Coalition was the PM's widely slammed address in which he was seen as not only rabbiting on to world leaders about irrelevant domestic issues, but also doing so in a way that whinged about the known unpopularity of government measures such as GP co-payments.  It was a free hit for the Opposition and it was duly and effectively taken.

Saturday, November 15, 2014

Victorian Poll Roundup And Seat Betting Watch: Sandbelt Edition

2PP aggregate of recent Victorian state polling:  ALP leads 53.2-46.8
"Nowcast" seat estimate based on this 2PP: ALP 49 Coalition 39

It's taken a while for enough polling to build up to justify another Victorian pre-election roundup to follow last fortnight's, but that point has been reached with the belated release of results for the Fairfax-Ipsos poll taken last weekend.  Why it has taken the Age until Friday to get this data out there is anyone's guess, but at least we have it now.

If you followed the media comments about the unreliable mutterings of both parties about internal polling and the like, you'd have seen claims from both sides that things have tightened up through this week.  These claims always need to be treated with caution, since the side in the lead has an incentive to make them to discourage complacency, while the side trailing has an incentive to make them to discourage despair.   They might turn out to be true, but with no data fresher than Monday, we'll need to wait for more polling to be sure.

Based on where things stood early this week, there would have to be a lot of improvement to make a serious difference to the picture of a very likely change of government.  At that stage the gap was, if anything, widening slightly, but the broader picture is a lack of any major change for quite a while. 

Wednesday, November 5, 2014

Poll Roundup: Two Steps Forward, One Step Back

2PP Aggregate: 52.2 to ALP (+0.2 since last week, +1.2 in three weeks, highest in twelve weeks)
ALP would probably narrowly win election "held now" outright

Four weeks ago I declared the Abbott government's budget blowout "over", their aggregate position having come back to 49% 2PP, which was about where it had been back in April.  It had taken almost half a year to get back to that point, but assuming that the Coalition was now going to poll competitively for a while, it was easy to believe that they were on the track to re-election.

What's happened since may bring just a little smile to ALP supporters, since no sooner did the government finally get back to effective parity than its ratings turned around and started heading down again.  As the Abbott government approaches the first anniversary of losing the lead, this small shift back adds a little more interest to the question of when and under what circumstances they might actually get it back.