Thursday, February 13, 2020

Queensland 2020: Currumbin By-Election and YouGov Poll

(Now added: Bundamba, scroll down)

Queensland is heading for at least one unexpectedly interesting by-election early in another state election year.  Also, a new YouGov poll has come out that has been the subject of incorrect reporting concerning the Premier's unpopularity.  I thought it would be useful to have a post up covering these two issues in detail.

Currumbin (LNP, 3.3%)
By-election March 28

Currumbin is in Queensland's far south-eastern corner and includes the border town of Coolangatta (now a Gold Coast suburb) and surrounding southern Gold Coast suburbs and rural hinterland to the west of them.  It has been held by the retiring member, Jann Stuckey, since 2004, but before that was held by Labor's Merri Rose for 12 years.  From 1992 (when Rose first ran) until 2001 the seat was more Labor-friendly than the state average, but this ended with Rose's fall from grace and Cabinet in 2004 and since then it has reverted to being slightly LNP-leaning compared to the state average.  It is possible, as the departing incumbent Jann Stuckey suggests, that Currumbin is an electorate where perceptions of the candidate matter more than elsewhere.

An Opposition vacancy in a by-election against a Government showing plenty of wear and tear would not normally be an interesting event and in many cases Governments would be tempted to give this sort of contest a miss.  But many things about the Currumbin contest are unusual.

Firstly, Labor had an already preselected candidate, former Parents and Citizens Association President Kaylee Campradt, who was already running for the seat, ostensibly for the main election, having been announced in October.  It appears Labor had actually been tipped off that Jann Stuckey might not go full-term in the seat.  Irrespective of this, having Campradt already running for the main election Labor had no real choice but to run her for the by-election.

Secondly there has been acrimony between Stuckey and her party after Stuckey was one of three LNP MPs to vote in favour of decriminalising abortion in a 2018 conscience vote.  Stuckey has cited within-party responses to this decision as among the stress factors causing her to bring forward her intended retirement, though she has also attacked Labor.  The attacks from within on LNP members who voted for abortion reform were indeed extraordinary and smack of an attitude that the institution of the conscience vote is the preserve of moral conservatives and nobody else is authorised to use it.

Stuckey has also criticised the preselection of her replacement, and has alleged that prosecuting solicitor Laura Gerber has been chosen as a token female.  Although Gerber is clearly well qualified, Stuckey has also strongly criticised her degree of connection to the electorate and has suggested that Currumbin voters won't support a "blow-in".  Gerber has denied being a blow-in, saying she has been living in the electorate for a year and citing a childhood background in the general area.

Thirdly, the by-election is the first electoral test for Opposition Leader Deb Frecklington.  Frecklington's polling so far has been lacklustre and she was widely criticised for attacks on the Premier's image in which she, among other things, implied that the Premier was not "grounded" because the Premier wore expensive clothes didn't have children.   Social media photos of Frecklington wearing designer labels soon appeared. There is sometimes speculation about former minister David Crisafulli as an alternative leader, and in the event of a disaster on March 28, the LNP would have enough time to install a new leader before the election if it wanted to.

Fourthly, the by-election occurs against a backdrop of federal turmoil on the Nationals side of the LNP.  At this stage I do not see this as likely to be a significant factor in Currumbin, but I could be wrong, particularly if it affects the apparently already weakened popularity of the federal government as a whole.

The Greens (Sally Spain) and One Nation (Nicholas Bettany) are contesting.  The Greens have polled reasonably strongly in Currumbin recently, including 11.7% at the 2017 election, but are unlikely to be contenders in a by-election where both majors are running hard.  One Nation did not contest Currumbin in 2017.  They did contest the seat in 1998-2004 and 2015 but have never made the top two in the seat and polled below their state average there in 1998.  They also polled below their state average in the seat in last year's Senate election - figures compiled by Alex Jago showed they had a Senate 4PP of 11.8% in the seat compared to a statewide average of 16.1% (and compared to the Greens' 16.8% in Currumbin.)  Support levels for both the Greens and One Nation might be inflated at a by-election by dissatisfaction with the major parties; nonetheless, it is difficult to see either making the top two.  Perhaps surprisingly, only four candidates have nominated.

Under normal circumstances, Labor would be thinking twice about even contesting Currumbin, since by-elections normally produce swings against incumbent governments, and a large swing could be embarrassing.  But preselection disunity, the loss of a 16-year incumbent, Labor's preparedness for the by-election and Labor's recent targeting of the Gold Coast area have all led to speculation (mostly from unnamed LNP sources) that the LNP might actually lose this by-election.

Government seat gains at by-elections

The last time a government gained a seat at a state by-election I wrote a review about the history of such events.  Government gains occur in perhaps 10% of state by-elections for non-government vacancies.  Government gains are more likely to occur when the government is in its first term (often this represents the exodus of heavy hitters from the previous government) and when the federal government is not of the same party.  In this case the state government is in its second term, but it is of the opposing side to the federal government, so federal drag factors are in its favour.  (Federal drag factors against the Abbott government seem to be a major factor in the Fisher result in SA in 2014, especially since the Weatherill government was not even able to retain office at the end of its term.)

However, if we look at the 14 (including Fisher) cases where either the state government was past its first term or the state government was the same party as the federal government (or both), only half of them are government wins from oppositions rather than from crossbenchers.  The government wins from oppositions occurred under exceptional circumstances - original result voided, an ex-Premier resigning, massive federal popularity pull or push factors (including in two cases wartime), and the 1971 apartheid protest crackdowns.   Currumbin is none of this so a loss would be an extraordinary result, and LNP speculations to that effect may be just a combination of panic and expectation management in anticipation of a no-swing win.

I like to have a go at setting pre-by-election guideposts as insurance against post-result spin from parties, so here's my attempt at this (which may be modified before the by-election):

ALP or other non-LNP win (any margin): Very bad result for LNP, likely to result in serious leadership speculation if not change.
LNP win 0-3%: Bad result for LNP and good result for Labor, to some degree explainable by special factors, suggests Labor leadership not a problem in this area.
LNP win 3-8%: Not a great deal to see here, result could be explained by some combination of normal by-election swings and special factors that may have reduced them
LNP win 8-10%: Good result for LNP, suggesting state Labor may be a little on the nose, and that federal factors are not causing significant damage.
LNP win >10%: Outstanding result for LNP, terrible result for Labor, suggesting greater risk that Labor will lose the state election.

For the Currumbin by-election see also Tally Room guide, Poll Bludger guideABC guide

The same day will see the massive Brisbane City Council elections.  Brisbane Council is dominated by the LNP but any sizeable swing in any direction will be seen as a portent for the state election.  The different schedules of BCC and state elections in the past have made it more difficult to measure this rigorously but looking at recent results there is some relationship.  Perhaps it's not a super-strong one, eg, as the swings to the LNP from 2008 to 2012 were relatively modest.

YouGov Poll

The recently released YouGov poll had the parties square on 50-50 2PP, a result which with anything like a uniform swing would result in a hung parliament.  The primary votes were LNP 35 Labor 34 One Nation 15 Greens 10 KAP 3 others 3.  Except for others, none of these are significantly different to the 2017 results, and others are probably being under-polled.  However this is the second poll in a row with the government slightly below its 2017 result.  Following the size of the polling error at the 2019 federal election, not too much should be read into that difference.

The poll mostly attracted attention for the leader ratings, including a poor net rating for Premier Palaszczuk of -15 (29-44).  For instance, Sky declared that "The popularity of Queensland Premier Annastacia Palaszczuk has plummeted below those of former Premiers Campbell Newman and Anna Bligh" and the Courier-Mail had "ANNASTACIA Palaszczuk has plummeted to become the most unpopular Queensland premier in recent history".  But these comments result from a misleading comparison of results from different pollsters.  Leadership approval polling for premiers prior to Palaszczuk was conducted by the old Newspoll using live phone polling.  The current poll is conducted by YouGov using exclusively online polling.  The YouGov polling has far higher "undecided" rates than the old Newspoll, and as a result the approval and disapproval rates for leaders tend to be lower.  The undecided rate seems to have even increased compared to the previous YouGov-Galaxy polling, which is odd because when it comes to federal polling, the undecided rate for the Prime Minister has dropped sharply.

Thus while Palaszczuk's 29% positive satisfaction rating does compare poorly with Newman's worst of 33%, the latter figure was recorded alongside a dissatisfaction rating of 57%, giving Newman a net -24%.  You just can't compare an online poll with a 27% undecided rate with phone polls with undecided rates in the range 6-11%.  So the "recent history" in which Palaszczuk is the most unpopular Queensland premier is actually just her own five years in the role - she is the least popular Premier since the one before her!  When it comes to Bligh, Sky's comment isn't even accurate if positive approval is the yardstick, since Bligh at one stage sank to 24% on that measure (Oct-Dec 2010) before partly recovering following her handling of the 2010-11 floods.

To put Palaszczuk's performance in a fairer historic context, here are the worst net ratings polled by each Queensland Premier since the old Newspoll started:

Bligh -43
Bjelke-Petersen -40
Newman -24
Cooper -19
Ahern -19
Palaszczuk (so far) -15
Borbidge -13
Beattie -7
Goss +16

However, while Palaszczuk's -15 isn't historically bad, it's still bad.  It's bad because once state premiers become significantly unpopular they nearly always fail to win the next election.  See Unpopular State Premiers Have Dire Historic Fates.  Since that article was written another five Premiers who polled ratings worse than a net -10 have all gone (Baird resigned, Weatherill, Giddings, Newman and Barnett all defeated), while three of four State Premiers who never polled such ratings (Andrews, Berejiklian and Hodgman) were re-elected (Napthine lost).  Perhaps this pattern does not hold for YouGov online polling as it did for Newspoll, but I wouldn't bet too much on that being the case.  If YouGov net ratings are comparable to other polls then Palaszczuk's Premiership is probably in trouble if the LNP can get its act together.

(A footnote: Wayne Goss is one of only four Premiers in the Newspoll era to lose office at the ballot box (albeit at a by-election) without ever polling a negative net rating.  The others are John Fahey (NSW), Carmen Lawrence (WA) and Rob Kerin (SA) - there was very little polling for the last two named.)

Deb Frecklington also polled poor ratings, but it isn't totally clear how poor.  The Courier-Mail's article said she had 23% satisfied and 44% dissatisfied (33% uncommitted) for a net rating of -21.  However the graphic had the dissatisfied and uncommitted results the other way around.  I have been waiting for a detailed YouGov release (as was the case with their previous poll) to clarify this, but to this stage that has not been seen.  If it is indeed -21, then only Rob Borbidge (-45 just before the 2001 Queensland election) and Tim Nicholls (-27 before the 2017 election) have polled worse.

Palaszcuk has also increased her Better PM lead to 34-22, but there's not much to see there really. Palaszczuk led Tim Nicholls 45-31 at the previous election, and the loss of numbers to undecided on both sides is likely to be largely a result of the new polling methods.

At this stage if the polls are accurate the 2020 Queensland election is shaping up as a repeat of the 2017 election in which there was virtually no 2PP swing and relatively little seat transfer.  The difference at this stage is the lack of the exotic factors that made 2017 so uncertain (One Nation boom and partial bust, redistribution, switch to compulsory preferences.)  However polling in Queensland has been extremely sparse, the new YouGov methods make sense but have been relatively little tested, and there is still time for anything to happen.  Queensland is also a state where federal polling has been persistently badly wrong and some day state polling might be so too (for reasons other than the extreme preferencing shift seen in 2015).  At the moment both government and opposition appear to be in lacklustre shape, but nobody else is consistently taking advantage.

Added: Bundamba (ALP, 21.6%) 

There will also be a by-election for Bundamba which may also be on March 28 though that is yet to be confirmed.  This follows the resignation of 20-year incumbent Jo-Ann Miller.  While Miller's resignation had been canvassed recently, the plot there involved Miller running for Ipswich mayor, which Miller has now ruled out.  Miller has been something of a rogue MP on the Labor side in the past five years, was the only Labor vote against abortion reform, and is friendly with Pauline Hanson.

On paper, Bundamba  is so safe for Labor as to appear unloseable (it did not even fall during the 2012 wipeout) but the minor catch is that it is a little less super-safe versus One Nation than versus the LNP.  One Nation have not run against Miller since 2004 but held Labor to a 6.2% margin during their 1998 surge. 

Surrounding Ipswich seats contested by One Nation (Ipswich, Ipswich West, Jordan) were around 60-40 to Labor over One Nation in 2017, but Bundamba is somewhat stronger in terms of base Labor support as confirmed by Jago's Senate figures.  The advantage of using Senate figures to measure base support is that it is not influenced by Miller's personal vote.  A notional Labor vs One Nation figure for the seat might be in the range 65-67%.

Even throwing in the usual by-election swing and the issues surrounding Miller's exit, it would be an exceptionally bad result for Labor to actually lose Bundamba or to get within a few points of losing it.

No comments:

Post a Comment