Monday, April 28, 2014

Is Campbell Newman Actually In Trouble?

Updated in May and June - scroll to bottom of article.


Life isn't easy for conservative state and territory leaders anymore.  Of the five who came to power during the Rudd/Gillard years, Barry O'Farrell, Ted Baillieu and Terry Mills all failed to make it to the next election, and Colin Barnett won re-election comfortably but his polling has been on the skids (see Unpopular Premiers Have Dire Historic Fates).   Now, the future of Campbell Newman, whose government must go to an election by the middle of 2015 and is likely to go to the polls a few months earlier, is also attracting speculation.

Newman's position has always looked dicey because of the nature of the seat he holds, Ashgrove. Newman's electorate had been held by Labor for 26 of the previous 29 years at the time he won it, and it typically runs about six points above the state average for the ALP.  So, all else being equal, the Coalition would only win this seat in years when they recorded about 56% of the two-party preferred vote statewide.  

Some adjustement should be made for this because of Newman's increased public profile as incumbent Premier, and because he has replaced a sitting member who had her own personal vote, but no convincing sign of this has been seen in public polling.  Thus ReachTEL had Newman trailing soon after election (despite his party leading 60:40 in Newspoll at the time).  Another ReachTEL late last year had Newman again behind 48:52 while Newspoll had the party lead at 55:45.  More recently, the Jan-Mar Newspoll (52:48 to LNP) was matched with a Galaxy showing Newman behind 47:53. Similar figures have been seen in commissioned polling as well.  

Until early this year the story of Queensland polling was really all about Ashgrove; it was taken for granted by most that the LNP would retain government easily and that the inevitable trimming of the backbench by fifteen or so seats would only be to the party's benefit, so long as the swing-back didn't take the Premier down with it.  But the most recent public polling opens up new battle lines.  

Firstly, Campbell Newman's popularity has never matched that of his party.  On coming to office his first Newspoll yielded a lukewarm +9 netsat and since then he has polled exclusively negatives.  Until recently the negatives were modest ones (anything from about -12 or above is really nothing to worry too much about) but the January to March Newspoll saw Newman with a rather bad rating of -18 (36% satisfied, 54% dissatisfied.)  As noted in the unpopular premiers article, while 21 other premierships (including those listed and also Lara Giddings') have now recorded Newspoll netsats of -18 or worse, only two of those have so far gone on to win the election at the end of the term in which the rating was recorded. The prospects of Premiers polling around -18 are better than those polling around, say, -25, but they are still historically poor. This observation must be balanced against the size of Newman's current majority and the fact that his party still has the 2PP lead, so his chances are not likely to be as bad as the historical record suggests.  

Secondly, the LNP lead on a two-party-preferred basis has steadily shrunk.  The Newspoll for the first quarter this year is the first that does not clearly point to an LNP majority win if an election was "held right now".  And thirdly, the party's fortunes have declined at federal level in the state as well.  This is best indicated by the current Bludgertrack article showing a loss of 8.2% 2PP in Queensland since the last federal election.  Unlike in WA, in which the Labor primary vote has been gouged by the Greens and PUP, in Queensland the Labor vote appears to be at least holding, with the Green vote rising.  I suggest that the Coalition's federal vote in Queensland is being gouged by increasing dissatisfaction with the state party.

What Would a 52:48 Result Actually Mean?

It was widely reported that the 52:48 Newspoll, implying a swing of around 11% to Labor, would be enough to cost the Newman government its majority and possibly government.  Looking at the pendulum (the ABC version is good because it converts the numerous LNP vs KAP seats into LNP vs Labor) and assuming a 10.8% swing, we can mark off everything up to Whitsunday (10.7%) for a loss of 29 seats, from the 74 the LNP holds (Yeerongpilly was lost to PUP and Redcliffe at the by-election).  That's still a majority, but of one.  Those who prefer the slightly higher estimate of the LNP's result last time (see Antony Green's discussion at the ABC link - I think the truth is somewhere in the middle) might throw Burleigh as well with Mansfield line-ball, enough to move the LNP into minority territory.

However this site always warns readers to beware of unrealistic uniform-swing type readings of the Mackerras pendulum and this is a case where caution is especially warranted.  I go on about sophomore effect a lot and this election will be one of the most sophomoric (if that's a word) in Australian political history.  Of the 43 seats under 17% on the Coalition side of the pendulum, all bar Mirani were won by the LNP from Labor in the 2012 election.  In each of these, the new sitting member gets a personal vote, and in 33 of the 43, the recontesting previous sitting member's vote was reflected in Labor's result in 2012.  These 33 "double sophomore" seats should especially see muted swings to Labor compared with the state average. 

So therefore in trying to translate a 52:48 Newspoll to a likely result, the first thing that needs to be done is to adjust the seats for sophomore effects.  But there are so many seats with sophomore effects (close to half the parliament) that this also requires a hefty adjustment in Labor's favour in those seats (the third-party seats, their own seats, and the ultra-safe LNP seats) where sophomore effect does not apply.  With that done, the number of seats that Labor would be favoured to take drops from 29 to 22, leaving the LNP still with 52 intact and a majority.  

As it turns out, the LNP would be favoured to retain a number of seats by less than 1% each based on sophomore effect, and in practice it would not hold all of those.  So the 52 seats is a bit generous and should be reduced to, say, 49-50.  But it doesn't really matter: the point is that because of sophomore effect, a 52:48 2PP to Newman's government is, all else being equal, not a losing result or even a majority-losing one.

Is All Else Actually Equal?

There are a number of caveats about any projection of a Queensland 2PP result to a seat tally.  Queensland is an especially difficult state in which to conduct such projections, because while the Brisbane area tends to see conventional two-party contests, the rest of Queensland is prone to fourth-party uprisings: One Nation in 1998, KAP in 2012 and very likely PUP in 2015.  (KAP is a spent force and should be supplanted by PUP at least outside the two seats it currently holds, but the geography of this will be messy as the parties appeal to somewhat different demographics.) 

Newspoll translates primaries of LNP 40, Labor 36, Greens 8, KAP 1, Other 15 to a 52:48 2PP. Preferences are muted in Queensland because of Optional Preferential Voting, but the implied assumption that the Other preferences flow half as strongly to the Coalition as the Green ones do to Labor is difficult to credit.  That applies especially if most of the Others preferences are actually PUP - PUP did not contest in 2012 and we know that PUP preferences at federal level break only weakly to the Coalition.  The other danger in extrapolating from last-election preferences in this instance is that Labor's preference performance in 2012 probably dropped in line with their woeful primary performance, and could therefore rise accordingly.  So maybe that 52:48 is really more like 51:49.  My sophomore-effect and seat-probability-adjusted pendulum translates that to around 46 seats for the LNP - still a majority, but not at all a comfortable one.

Throw in the possibility of PUP winning LNP seats and things become even more challenging for the government if it goes to an election with such a narrow 2PP lead.  It is rather challenging for fourth parties to win seats in numbers anywhere near proportional to their vote, especially under OPV - in 1998 One Nation won 12% of the seats with 22% of the vote, and in 2012 KAP won just two seats with 11.5%.  PUP would not have won the seat of Fairfax at the federal election had preferencing been optional.  Nonetheless with state seats being much smaller than federal seats, it's likely PUP would have enough support to win a few somewhere if they can poll over 10% statewide.

Another caveat to be noted is that the typical double-sophomore situation involves a sitting member who is defeated and who then retires from politics with a new candidate taking a swing at the new incumbent.  However, in 2015 I think it's likely that Labor will try to counteract the sophomore problem by running large numbers of former incumbents.  Indeed, for purely talent-pool-related reasons, they would find it difficult to avoid doing so.  This may well mean that in many seats the advantages for the new LNP sitting members are not so great.

The final caveat is regional distribution.  The various union-commissioned polls taken through this government's term have tended to show the swing being higher in Brisbane where there are many loseable seats, and less pronounced in rural Queensland.  This is something that really needs to be corroborated by neutral polling (which so far hasn't covered the question) but that, if true, introduces the risk of a South Australia type situation in which Labor does well enough in the city that it doesn't need to win the 2PP.

There are, therefore, a number of arguments that can be advanced on both sides to vary from the published projections that the Newspoll implied a loss of 30 seats.  The one in the LNP's favour (sophomore effect) is the more concrete; those pointing the other way are more speculative but none of them are far-fetched and they do carry quite a lot of truck between them.  

So all up, if the LNP was looking at a two-party preferred vote of 51-52% at an election, they would give themselves a fair chance of retaining a narrow majority.  However, they would also have significant doubts.  A government that won the previous election so emphatically should be hoping for a great deal more.

Newspoll Corroborates "Union Push-Poll"

Queensland during this term has been notable for the massive volume of union-commissioned polling - typically ReachTELs with very large sample sizes that have been at times released and at other times alluded to, often tied in with "big data" style campaigning.  (For examples see the Pollytics Megapoll article and also Anatomy of a Modern Campaign). The Newman government is almost as aggressive towards these union polls as it is towards bikies, invariably dismissing them as "union sponsored push polls" or similar, and refusing to comment on the results.  

Although the union commissioned polls have frequently included issue-based questions, many of which occur in runs with each question likely to contaminate the results of the next, there has been absolutely nothing to substantiate the charge of "push polling". Some of the issue questions are at worst what I call skew-polling.  That means that the question design isn't primarily designed to implant information to the subject, but just might lead (intentionally or otherwise) to a slanted result. For instance,  if you ask voters a bunch of questions about the government replacing Queensland doctors with overseas doctors, and then ask if the government deserves re-election, then you've primed the respondent to be more likely to answer no.  Or if you tell someone that a report is called the Costello Report rather than, say, the Commission of Audit report, whatever it says will get the odd raspberry from someone who just doesn't like Peter Costello. That said, whatever criticisms can be made of some of the issue questions asked by the unions, these don't affect the accuracy (or otherwise) of the basic voting intention questions.

The Working For Queenslanders ReachTEL was a case in point.  This poll, taken in early April and released just before the Newspoll, was greeted with some cynicism because of its source and because it showed the LNP only four points ahead of Labor on the primary vote, which seemed a bit small.  The union's presentation of the results also used a regional breakdown to project a massive loss of government seats (I would bet with no adjustment for sophomore effect or probability, just straight off the pendulum except with regional swings.)  However the Newspoll taken over the January-March period then showed exactly the same margin between the parties, and also extremely similar primary votes across the board.  The Newspoll's leadership ratings for Newman were even worse than the union poll.  So the idea that the basic voting intention results of the union poll were wrong doesn't seem to have a leg to stand on.

Relative Deprivation

I've observed a kind of deprivation effect in the way the left generally reacts to bad polling for the right at both national and state level.  The left is so used now to years of being way behind and to massive Coalition wins at elections, that polls that merely show Labor as competitive are treated as signs that the Coalition is in big bad trouble.  We are not seeing the more realistic battle-weary attitude seen from ALP supporters in the late Howard years, in which people knew all about government resilience (especially after 2001), and realised that it was not just a lead that oppositions needed, but a big lead, before they could really get excited.

Those who like to extrapolate downwards trends would see the LNP shedding votes at the rate of about one point per month and imagine that it is not too long before the LNP is in serious strife.  However most governments in any state, historically, would have killed for 51-52% 2PP in the middle of their term - it's hardly bad as mid-term slumps go, provided that is all it is.

It may seem therefore that the LNP is only one competent charm-offensive away from a few-point lift in its 2PP and a return to a position of easy dominance.  But aside from the question of whether they can actually pull that off, the LNP are struggling with one of the forces that most strongly affects voting intention in Australian state elections: the tendency of the federal picture to contaminate state politics.  It was Labor's deep unpopularity at federal level that blew out the margins of the thumping wins for the Coalition in Queensland, New South Wales and Western Australia.  Now that there is a Coalition federal government that is not exactly covering itself in glory, and that might become really unpopular before long, it is a nervous time for state regimes of the same stripe.  Voters seem to have suddenly short memories of whatever caused them to ditch previous ALP governments - or else are just accustomed to ditching incumbents at the slightest provocation come what may.  Keeping votes taken from Labor has become hard work.

At this stage, it would be bold to suggest the LNP would actually lose one term after such a massive win, but I do think they could end up with a very hefty loss of seats, which might concern them and embolden the ALP for a little while, but in the long term is no big deal.  Even a loss of, say, two dozen seats and a seat result in the low 50s would hardly be bad, given that so many of the seat wins in 2012 were unsustainable products of Labor's long incumbency and federal malaise.  But you wouldn't want to be in the position of an LNP strategist going into an election with mediocre polling and an unpopular Premier who appears to be at risk of losing his seat, since speculation about even more unpopular alternative Premiers would be a distraction.  For all that Campbell Newman says about being unwilling to transfer to a safer seat elsewhere, he'll need to lift the party to a more clearly winning position by the time the campaign starts, for his own sake if not theirs.

First-Term Bootings: The New Black

I mentioned at the top of this article that three of the five Coalition state/territory leaders who came to office under Labor failed to make it to their first election - two resigning (one under crossbench pressure after his party lost its majority) and one being sacked.  When Kevin Rudd was dethroned as Prime Minister, this kind of tearing down of a winner by his own party, without a chance to face the people at an election, was seen as a great sacrilege by many, and even the Liberals cried many crocodile tears over it.  Yet on the other side of politics since, the main difference seems to be only that that their deposed or self-deposed first-term leaders don't come back.

In Australian political history, the situation of first-term leaders, elected from Opposition, not making it to their first election, is a very rare one.  The bulk of such cases involve parliaments that were very much hung from the outset - three-cornered parliaments in the days of the Country Party being the most common example.  Then there are cases of leaders resigning in their first term because they wanted to do something else (Barton being the classic federal example) or because they were forced to retire by party age rules.  Apart from that, what we've seen in the last four years at state and federal level is a remarkable spate of first-term leader-loss.  Dean Brown (elected Premier of SA in 1993 but then rolled by his party just before the next election) is a well known prior case but I could find just a handful of others.

I don't think Newman will be thrown under the bus (or perhaps since it is Queensland it should be the bike) in advance of the election.  However there are multiple signs that his position is a deal weaker than his party's.  Newman's own stance is that he is focused on doing what needs to be done, that the polls are ephemeral (as his infamous predecessor Joh Bjelke-Petersen had it, "They come and go as a morning mist") and that if he gets it right the electorate - including his own electorate - will come around.  But my suspicion is that the polls are not only about doing what has to be done, but also about measures taken by his government that are seen as excessive, and also about the difficulty of governing with a similar regime in Canberra.  So we'll see how he goes with all that.

Update May 25:  The answer to the article question is an emphatic "no" at party level, but a lot less clear for Newman personally, according to a new Galaxy which shows the LNP leading 55:45.   The poll was reported as showing PUP on 5% but I believe this to be incorrect (see below).  For such a hefty 2PP lead, Campbell Newman's personal ratings are again not too flash.  He had a net rating of -13 (40-53) according to Galaxy in February and I have found that his approval is down four points to 36, the same as the most recent Newspoll.  (I haven't found his new disapproval score yet).

Also while Newman's Preferred Premier lead of 48:33 over Annastacia Palaszczuk might sound decent, it's actually the sort of modest lead to be expected if the parties were much closer, given that preferred leader scores favour incumbents.  This poll came from the same run as a federal poll showing the LNP up 52:48 in the state, which is higher than all other recent results, although the Nielsen state breakdown at 51:49 also provided hints of improvement (oddly given the Coalition's national federal polling plunge).

This all comes ahead of the by-election for the seat of Stafford (date as yet unknown) which the LNP will have much difficulty (even if they are on 55% statewide) holding on a 7% margin following the resignation from parliament of Dr Chris Davis.  Prior to resigning, Dr Davis commissioned this ReachTEL of his own electorate.

The Galaxy poll results have raised some eyebrows because they seem to show a high support level for asset sales (38%) as a means to reduce state debt, but this is only when the other options given are increasing taxes (21%) or reducing services (24%). "Don't reduce debt", "reduce public sector salaries", "improve efficiency of government spending" and "other" (to give some possible examples) were apparently not options.

The other surprise, and this if correct is a more genuine one, was that the federal poll found a nearly split view (48% for, 50% against) on GP visit co-payments in the state, compared with the national Galaxy result of 35-55 a few weeks ago.

May 26: Newman's disapproval score is 55 (+2) giving him a Galaxy netsat of -19, a point worse than the last Newspoll.  The breakdown of Newman's ratings is 72% satisfied 23% dissatisfied among LNP voters, 9-87 among Labor voters, and I make it a rather horrible 9-68 among third party voters.  Although 5% was reported as the score for the Palmer United Party, it may actually be the score for KAP, with previous Galaxy state polling in Queensland having listed that ailing party but not its more dynamic populist colleague.

Annastacia Palaszczuk's personal ratings are probably of very little consequence to anything at this stage but for what it's worth she's at -4 (33-37) with relatively little polarisation: 49-24 Labor, 26-49 LNP, and my estimate is 22-33 from others.

May 30: Essential has corroborated Galaxy's result with a 53:47 result taken through May.  Essential finds a high PUP vote of 12 points.

What is interesting though is that Essential seems to confirm the LNP might have gone through a dip and then recovered, since its April sample (see full figures) was 48-52 and its January and February samples were both 50:50.  This from Essential which is notorious for the lack of volatility in its 2000-sample federal polling.

June 28: I missed the public ReachTEL a few weeks back, probably because I was running a chess tournament that weekend.  This poll showed the LNP on 40.9%, Labor on 34.1, Green 5.2, PUP 13.6, Other 6.3.  Estimating a 2PP result is a fraught exercise because of the PUP factor but Poll Bludger gave it as 53.6% using the last federal election for an assumed non-exhausting half of the PUP votes.  Possibly it would be closer, say around 53.  A more recent union ReachTEL has LNP 39.1, ALP 34.3, Green 4.6, KAP 4.3, PUP 15.1, Other 2.6, after redistributing an Undecided option which polled 6.4.  I should caution here that the inclusion of an undecided option may have some effect on results, though in this case the two polls are very similar.  The poll again points to a 2PP around 53, but the meaning of the concept becomes elusive as the PUP vote rises.  My suspicion remains that this level of PUP vote would not convert to many seats, falling mostly in safe Coalition electorates or in seats where the majors were close enough to ensure that PUP ran third.  However if it reaches the high teens and stays there til the election, things get messier.

Of the leaders, I convert Newman's ReachTEL ratings to a not-too-awful -9.3 on the Newspoll-style scale and Palaszczuk's to +3.4. The Queensland state budget secured similarly negative reactions to the federal one, but with no apparent impact on voting intention.  While the Newman government remains well placed to be returned, it is still not showing the kind of lift in results that should make the Premier more comfortable in his own seat.

An intriguing aspect of the recent union ReachTEL is that it shows the recent PUP support rise to be coming from young voters, the 18-24 group especially.

June 30: Newspoll Shocker:  A Newspoll quarterly result (April-June) has just come out with the LNP on a pathetic primary of 32%, with 34% for Labor, a generous 8% for the Greens, KAP 2%, and, well, you know where most of the other 24 has gone.  The 2PP is 51:49 to Labor.

As the above Essential, Galaxy and ReachTEL results show, a 2PP of 49 for the LNP is not exactly representative, with the average of the others coming out at around 52.5, depending on the usual shopping list of assumptions.  Also, if there has been a slight upswing through May and June then it may be that even the last month's Newspoll data would have had the LNP above 50 given a decent sample size.  But still this result is a significant dampener on the evidence that the LNP was starting to lift towards the mid-50s (for what 2PP is worth in Queensland anyway).

There is more trouble for Newman in the Newspoll in the form of a horrible net rating of -24 (33-57); his worst so far.  As documented in Unpopular Premiers Have Dire Historic Fates, for Newman to lead the LNP to victory at the next election, he will have to equal Bob Carr's record for the worst Newspoll netsat from which a Premier has recovered.  Palaszczuk's netsat is -2 with a still-high "don't know" factor (ReachTEL showed over 5% still haven't heard of her) and Newman's 39-35 better Premier lead is a miserable one given the house advantage to incumbents in that question.

No comments:

Post a Comment