Wednesday, July 1, 2015

Poll Roundup: Shorten's Struggles Aren't Shifting Votes

2PP Aggregate: 52.3 to Labor (unchanged since end of last week)
Labor would probably win election "held now" with small majority

It's been another fortnight of sound and fury, dominated by speculation over Bill Shorten's leadership (amid worsening personal polls, unflattering reflections from The Killing Season, and concerns about aspects of his union background), still more anti-terror laws and another overheated culture war about some goose let loose on QandA.  Yet in this week's voting intention polls, very little has changed.

A ReachTEL late last week came out with a 52:48 to Labor headline, followed by a 53:47 by last-election preferences (53.5:46.5 respondent-allocated) from Morgan and a 53:47 from Essential.  The latter broke the run of seven consecutive 52s, as Essential finally caught a touch of the Green surge being seen by all other pollsters.  There was no Newspoll as the brand is in transition to operation by Galaxy.  (Meanwhile former Newspoll staff have started Omnipoll, which doesn't look like it will be doing regular voting intention polling, but which will continue many of Newspoll's other operations in an online format.)

After adjustments for the primaries and Morgan's house effect, I aggregated the ReachTEL as 48.2 to Coalition, the Morgan as 48 and Essential as 47.3, and the net impact is diddly-squat: a 0.1 point improvement for the Coalition on two weeks ago:

I should note that while the different aggregators have been pretty close together in recent weeks, PhantomTrend now has it at just 51.5, from which the "model's best guess" is an Abbott-Katter coalition. Mark The Ballot has 52.2 and I'll add in BludgerTrack when it's updated (update: it's 52.0, with a projection of 77 seats for Labor).  Labor's lead might look pretty solid but when translated into seat terms, it's a sliver.


ReachTEL has been showing remarkably flat net ratings for Shorten, but not any more.  Shorten's net rating crashed over ten points down to -26.2, making this another poll in which he has recently recorded a career-low rating.  There was a seven point rise in the number of respondents rating Shorten's performance Poor or Very Poor with falls in all of Very Good, Good and Satisfactory.  (Abbott also dropped a few points to -25, with a 3.9-point increase in Very Poor.  For those who like much prettier graphs than mine, check out Mark the Ballot's review of leader polling).

ReachTEL had Shorten leading Abbott as preferred Prime Minister, 56.3% to 43.7%, but this lead is a product of ReachTEL's forcing method, and is not comparable to polls that don't use forcing (which have lately tended to have Abbott leading.) As noted before I actually prefer ReachTEL's method, but Shorten's figure would include far more soft preferences from voters who dislike both leaders than would Abbott's.  Others voters strongly prefer Shorten, my estimate being 66:34 in his favour.

Speculation about Shorten's leadership has built in the last fortnight.  The reforms brought in on the reinstatement of Kevin Rudd were supposed to solidify the party leader's position, but in fact they have undermined Rudd's successor's authority by making him the first Labor leader in history to be clearly not supported by the rank and file.  The original protections against rolling the leader (excessive in the case of Opposition Leaders in my view) are also not yet constitutionally entrenched  (Article by Troy Bramston, title "Bill Shorten's fate back in the hands of Caucus") and it remains to be seen if they'll become so.  All the same, none of the Shorten stuff seems to be having the slightest impact on voting intention.  It's quite possible that a lot of voters who disapprove of Shorten just don't care who leads Labor so long as that person beats Abbott.

Issues Polling

ReachTEL asked a question about "confidence in the Australian political system".  It's no surprise that the result is slightly more high than low, and amusing that the breakdown by party correlates so cutely with what happened in the last election.  

ReachTEL found the Coalition favoured on handling of national security, but given that this is a Coalition strength area and especially given the PM's constant hammering of the issue, the narrowness of the margin (52.6:47.4) is unflattering, even for a forced-choice question.  (With a don't know option thrown in the lead would be somewhat greater.) 

Essential showed that respondents like the idea of the federal government handling all school funding and strongly dislike the idea of it funding private schools and leaving public schools to the states.  There was also strong disapproval for paying people smugglers to return to Indonesia, with a predictable partisan breakdown.  A question on trust in journalists showed Laurie Oakes both widely recognised and far more trusted than other options, with Alan Jones dragging along the bottom.

The AFR had a useful wrap of the most recent JWS True Issues release, showing that the government rates best on issues voters aren't so prone to rate as important. Comments under the heading "obsessional culture war" are especially relevant - hardly any voters consider ABC bias an important issue - and I wonder whether the Coalition might have made more headway off Shorten's miseries had they not been preoccupied with side-issues.

Spurious Concept of the Week: The October Curse!

I'm not going to even try to link to it as it has been heavily paywalled, but Phillip Hudson in Monday's Australian had an article about the idea that first-term Labor leaders are blighted by an October Curse.  This is based around the defeats of Curtin (1937), Whitlam (1969), Hayden (1980), Beazley (1998) and Latham (2004), all first-election ALP leaders who lost in elections held in October.  The suggestion was even that Tony Abbott should be frantically scurrying for double dissolution triggers to exploit this blight, even though (as Hudson basically admits) there would probably not be time to get Senate reform implemented properly by the AEC.

The first problem with this concept is that Labor actually won an election in October once, under Scullin in 1929.  That is, however, supposedly different, because Scullin was a second-term Labor leader.  Absent of any explanation as to why that should matter, this is simply special pleading (the explaining away of a data point that doesn't fit the pattern by referring to an irrelevant difference.)  It's hardly as if Scullin's victory in 1929 was a result of a vast increase in political skill obtained in the eleven months since losing the 1928 election; indeed, at the time of his 1929 win, Scullin had been in the job for less time than most first-term losers.  Nor is the record of second-election Labor opposition leaders by any means stellar.  Rather, the main reason Scullin won big in 1929 was that the Government he opposed had collapsed.

When looking at the supposed curse on first-term Labor Opposition leaders going to the polls in October, the important thing there is that first-term Labor Opposition leaders don't do well in general.  Excluding those who had already been Prime Minister their strike rate is 2 from 12; if we add in past PMs it's 3 from 13.  It's hardly statistically significant if a month that happens to contain five such elections produces a 0-5 split. 

Furthermore, of the five first-election October Labor losers, most of them actually polled reasonably well.  Beazley and Whitlam both won the 2PP, with Curtin and Hayden breaking 49% and only Latham (47.3) really letting the side down.  If we had 2PPs for the first up losses by Tudor and Charlton it would turn out that the average 2PP of first-term Labor Opposition Leaders at October elections is no worse than at any other time.  The real problem is that coincidentally the two of them who polled the best (Whitlam in 1969 and Beazley in 1998) could not convert their good 2PP performance into enough seats, because of personal-vote legacies of the Coalition's crushing victories in the elections immediately before.  

The "October Curse" then, is just a random clustering of Labor first-term-leader defeats that has no more merit to it than the meaningless precedents that infest US election punditry.   There is really not much variation in how well governments poll at different times of the year.  

That wasn't anywhere near the worst thing I saw in psephology this week, but time is short and work is uncommonly abundant.  I'll save the other one for a separate article sometime.

Not-A-Poll Update: Another deadline has passed and Abbott is still PM; as of the end of June 2015, 140/543 (25.8%) of voters had incorrectly predicted he'd be gone by now, with the rest still in the game.  151 votes (mostly old rope from the spill-vote phase) still say he won't make the election (or it will be held this year and he'll lose it), 129 say he will lose the election next year, and 121 say he will win it.  This includes 44 votes (8.1%) who predict him to have at least as long a career as Howard.

Sunday, June 28, 2015

Singh Dumped To Fourth On Senate Ticket

Today Labor in Tasmania announced the remainder of its federal preselections.  None of these came as any surprise, but the one attracting by far the most attention is the dumping of sitting Senator Lisa Singh to the stereotypically "unwinnable" fourth position on the party's ticket.

While there was grassroots party involvement in the preselection via a state vote of members, it did not shift a long-expected outcome: the other sitting Senators Anne Urquhart and Helen Polley retained the first two positions and the Manufacturing Workers Union state secretary John Short was placed in the third position occupied by Singh at the 2010 election.  While it was widely claimed Singh was the victim of a left-right factional deal in which in return for Short being supported for the third position, the Left agreed to encourage its members to support Polley, the real cause of Singh's demise seems to be the Left supporting Short ahead of Singh, and Singh basically just not having enough primary support.  (See the detailed comment by Adam Clarke below.  I have also made some comments about the voting system.)

The reported deal attracted many appalled and bewildered reactions on social media, especially after the outcome was announced, with some commenters drawing comparisons to the electorally disastrous elevation of Joe Bullock to the top of the WA Senate ticket.  It's been seen as a typical example of the faction-hack disease that has been blighting the ALP for several elections, in which candidate quality takes a back seat to internal candidate support.  Singh appears well regarded by the media and we can expect a fairly large amount of media interest in this especially following on from the negative publicity for the party's internal workings in The Killing Season.  A case like this provides plenty of ammunition for those who want to argue that Labor has learnt precisely nothing about the damage its internal power games do to its wider standing.  That said, if the issue this time is really not the processes (again see comment by Clarke) but that party rank-and-file members also want to choose the "wrong" candidates, what can you do?

Lisa Singh's Record

Senator Singh is a former one-term Tasmanian state MHA and state Minister.  Singh polled 5760 votes in Denison at the 2006 state election, more than (for instance) Julie Collins who is now federal MHR for Franklin.  However, in 2010 Singh's vote declined as a result of the general decline in the Labor vote at that election and competition from newcomer Scott Bacon.  With Bacon's election, both Singh and fellow incumbent Graeme Sturges lost their seats as Labor dropped from three to two seats in Denison.

Singh was preselected for the Senate in third position on the ballot.  The 2010 election was the last at which Tasmania had a very high percentage of below-the-line Senate vote and Singh polled 9,132 below-the-line votes (6.7% of the ALP's total).  This was, in fact, the highest vote total for a number three Senate candidate in Tasmania since above-the-line voting began.  It was not, however, the highest percentage of a party's vote obtained by a number three candidate, as the Liberals' Guy Barnett polled 7.5% of all Liberal votes at the same election.  It did Barnett no good as he was defeated by Singh with the state returning a 4-2 left-right split (3 Labor 2 Liberal 1 Green) for the second election in a row.

Singh's appeal is especially to left-wing, young, environmentally-inclined Labor voters, and likely includes some voters who do not normally vote ALP.  At her exclusion in the 2010 state election cutup, an enormous 27.5% of Singh's vote leaked rather than flowing to Scott Bacon, with the Greens' Helen Burnet alone getting nearly a 10% leak.

During the 2006-10 state parliament, Singh notably did not support Labor's Pulp Mill Assessment Act (an attempted fast-tracking of the now moribund Bell Bay pulp mill proposal by the now insolvent timber company Gunns), unusually claiming it to be a "moral issue".  After Singh unsuccessfully requested a conscience vote on the matter, she was permitted to abstain from the vote.  While contributing to the cult appeal she currently enjoys among Tasmanian politics-watchers, this has probably not done wonders for Singh's reputation as a team player or her support among industry groups influential on pro-forestry swinging voters.

It is not easy to measure the profile of state Senators but Google News returns about almost as many hits related to Singh as fellow Tasmanian ALP Senators Anne Urquhart and Helen Polley combined.  Singh is mentioned over three times more frequently than Senator Carol Brown and about five times more often than extremely obscure Senator Catryna Bilyk, whose very existence may well be news to some of my readers.

Jobs for the boys?

One aspect of the debate I don't take too seriously is the suggestion that Singh has been pushed aside for Short because of her gender.  To the extent that gender might have been a factor, Tasmania currently has five female Labor Senators and no male, and previously had a 6-0 female-male split.  The idea that Short is being picked as a token male is hardly the whole explanation when Singh is not the only female Senator.  Of more interest in my view is the loss of ethnic diversity (Singh is Australia's first MP of Indo-Fijian ancestry), and the fact that it was Singh and not Helen Polley who was demoted.

Polley Trail Goes Silent

I freely and even proudly disclaim that I'm not usually a fan of Senator Polley's politics.  Socially reactionary union-linked types are not a group I have much politically in common with, and I think it's a shame they continue to wield so much political influence within what is supposed to be a vaguely modern party.  This problem was most acute when Julia Gillard's political debts to Shoppies boss Joe de Bruyn stymied what could have been a mandated Labor vote on same-sex marriage in the previous parliament, as a result of which Australia has avoided any chance to get there before even such notoriously religious polities as Ireland and the USA.

My own biases aside, what is interesting in the context of Singh's dumping is that at a certain time in the current Senate term, Senator Polley appeared "embattled" by more than just her antiquated views, and seemed to be on the way out.  In mid-2014 Polley came under scrutiny after spending tens of thousands of dollars in expenses on charter flights between Hobart and Launceston.   This came about six months after even more serious news that Polley's office had been investigated over bullying and harassment claims with a number of negative findings (as detailed in that link) and with three staff members taking successful action.  Senator Polley also came under scrutiny over the grand Tasmanian political tradition of employing family members in political offices.

At the time of these controversies I saw a few comments along the lines of: not to worry, she probably won't get preselected again.  Surprisingly these concerns about Polley's suitability to continue what is already a two-term career just vanished and didn't even resurface once the anti-Singh deal was struck.  I have not seen any mention of any of them in recent media coverage of the preselections, most likely because Polley's retention of the #2 position was so much a done deal that there was no point in opponents even trying to bring these matters up again.

What Is A Good Senator?

Defenders of Polley might say that despite these rough edges and reservations about some of her views, we should look at the work she does as a Senator, both for her constituents and in combating the government.  And likewise, they might say that Singh's media profile and appeal to a certain voter type is not necessarily a sign that Singh is an effective Senator for the party.

Unfortunately it is almost impossible to say who is a successful Senator by any objective measure.  At least with MHRs it is possible to look at their performance as a candidate from election to election compared to the fate of their party nationally and locally, and from this to get some idea of which candidates have strong personal votes.  But for the Senate there is no such yardstick and most voters just vote for a party without seeming to care much who its state Senators are.   This means there is really little obstacle to internal deals determining Senate positions.

What Will The Impact Be?

The impact of demoting Senator Singh on Labor's chances depends not only on the reaction but also on what electoral system is in place for the next Senate election, and we don't as yet know what that will be.

If it is the current system then there will probably not be a great impact.  While Lisa Singh polled a large personal vote at the 2010 election, the 2013 election saw a great reduction in the rate of below the line voting in Tasmania, on account of the record number of parties running for the election.  With this continuing, several of those who cast a personal vote for Singh last time would still just vote 1 Labor this time.  Even on the very plausible assumption that some portion of them defect to the Greens or left-micros in disgust, I don't think there's much likelihood of a net impact above a 1% fall in the ALP vote.

There is also no point in the current system in placing too much hope in a below-the-line rebellion to save Singh's position.  Labor voters are especially obedient of how-to-vote cards and especially averse to voting BTL.  In the 2014 WA by-election, despite the enormous controversy over Labor's placement of Joe Bullock above Louise Pratt, Labor's BTL voting rate was still lower than the average across all parties, and barely a third of the BTL rate for the Greens (who had no such candidate issues).  Pratt herself received only 2% of the state Labor vote.

It may be that several left-wing parties including the Greens and various micros all decide to preference Singh above the other Labor candidates, but if so this won't make much difference.  The Greens are unlikely to have a large (if any) surplus, and if they do then Singh will likely be excluded before their second candidate.  We don't like to say things are impossible round here, but the chance of Singh winning from fourth in the current system seems vanishingly low.

Under the proposed JSCEM model for Senate reform, things could be a deal more exciting in Tasmania.  Because of the state's strong familiarity with multi-candidate voting, it's possible that optional Senate preferencing would lead to a high BTL voting rate in Tasmania specifically.  This would create the opportunity for spurned major party candidates to try to save themselves through BTL campaigns, but even so it wouldn't be easy.  Voters who currently vote below the line in Tasmania are not only doing so because they prefer to order candidates within parties differently.  Many are doing so because they prefer to preference parties in a different order to their own party's lodged group tickets.  This type of voter would instead use the above-the-line preferencing option.

If there is lasting anger about what has happened in this case then I think the most likely impact is a small loss of ALP votes to a range of left alternatives, including the Greens.  I don't think we're likely to quite see a repeat of the WA Senate disaster as a result; not only did Joe Bullock have a lot more "form" than Senator Polley, but also that was a by-election.  There is also a long time between the announcement of this decision and the next election.

As for Labor's chances of winning a third Tasmanian seat at all, under either system that will not be easy.  They did this in 2007 (when Labor did very well nationally) and 2010 (when the Liberals ran a poor Tasmanian campaign) but those are the only times a 4-2 left-right split has happened in the state. It's not something to be confident about this far out but there seem to be good prospects for some level of rise in the Green vote, making the Greens very difficult to beat.  Three seats for Labor isn't impossible, but it is difficult enough that Singh would have been facing a difficult battle even had she just not been promoted.

I may have some more comments on this issue after seeing media reaction and whether there are any serious moves to have the decision overturned.

PS: A couple of posters on Twitter have drawn my attention to Polley being the sole survivor of what was otherwise a left-wash of the Tasmanian ALP preselections, with all the House of Reps candidates being from the left of the party.

NB: Some parts of this article have been revised following the comment from Adam Clarke below.

Another Update: James Brady at The Examiner has provided some figures! Brady gives rank-and-file votes of Urquhart 221 Polley 123 Singh 110 Short 74 (the votes for the remaining candidates are not stated) and partial figures for other vote sources.

In comments below I point out a deficiency of Labor's implementation of Hare-Clark, which is that they treat the ballot as an election for four positions, when in fact the fourth position is the "unelectable" spot and should not be treated as a win.  On this basis had what I think is the correct quota been used (25% for three winnable positions, rather than 20% for four) Polley would certainly have fallen short of quota on the first ballot of the rank and file members.  Depending on the split of Urquhart's votes between Short and Polley it is possible Short would have then overtaken Polley and been elected second, if the election had used the correct quota and only the rank and file votes.  So at this stage it is unclear to me that Polley's retention in second place is really the will of the rank and file.

Saturday, June 27, 2015

Could A UK-Style Mass Pollster Fail Happen In Australia?

Advance Summary

1. Recently all major final polls and the vast majority of analysts failed to predict the UK election correctly.

2. Random sampling error, "shy Tory effect" and late swing probably did not make major contributions to this outcome.

3. More likely causes include herding (although this has not been conclusively shown to have happened), the difficulty of estimating turnout, and the abundance of essentially non-random "panel polling" methods.

4. Australian elections are easier to poll for because almost everyone votes and because there is very little tactical voting in the Lower House.

5. Despite this there is a greater risk of high average polling errors at the next federal election because of a rapid turnover of polling ownership and methods, which will make it more difficult for pollsters to detect issues with their methods.

6. Potential for Labor support to be soft even in very late polling also appears at this very early stage to be likely to be higher than normal.

Wednesday, June 17, 2015

Poll Roundup: Hockey Gaffe Breaks The Monotony

2PP Aggregate: 52.4 to Labor (+0.6 in a week)
Labor would probably win election "held now" with small majority

Finally we have some movement at the station.  After four consecutive weeks of 51.8% to Labor my polling aggregate has finally moved on, with a modest step in Labor's direction.  The move is not large enough to be completely certain it is real, and there is enough going on to be unsure what is causing it if so, but it's most likely Treasurer Hockey's remarks on housing affordability (and some faltering and ill-judged attempts by his colleagues to put that genie back in its bottle) have contributed to a small-scale revival of perception that the government is out of touch.

This week's polls

Four polls have come out so far this week.  Fairfax-Ipsos opened procedings with a 53:47 to ALP (by 2013-election preferences, 54:46 respondent-allocated), which was interesting because Ipsos has usually thus far leaned to the Coalition by about a point.  There was some scepticism about it based on a premature perception that Ipsos is very bouncy (see below).  However the Ipsos result was more or less totally backed in by a 2.5 point move to Labor in Morgan, to 54.5 - bearing in mind that Morgan skews to Labor.  2-0 for the proposition that Hockey had really put his foot in it, but this was contradicted by a 51:49 Newspoll, while Essential did as Essential usually does (ie very little) and stayed at 52:48.

Friday, June 12, 2015

Would Proposed Senate Reforms Increase The Risk Of A Blocked Senate?

Advance Summary:

1. A previous article on this site showed that proposed Senate reforms to eliminate preference-harvesting disadvantage only preference-harvesters.

2. There apparently remain concerns that the proposed new system would lead to an increased chance of a blocked Senate with the Coalition (or the Coalition plus clearly right-wing crossbenchers) holding half the seats when Labor came to power.

3. Preventing the election of micro-parties off very small shares of the vote does increase the chance for either "side" to from time to time win exactly half of the seats.

4. However, if such a situation does happen, it would be very unlikely to persist beyond a new Labor government's first term.

5. Furthermore, it is only likely to arise in the first place in a case in which Labor is thrashed at one election then wins narrowly at the next (a situation that cannot apply to Labor if it wins the next election narrowly, because of the crossbenchers elected in 2013).

6. Based on the actual votes cast at elections, Labor would actually have had an easier road to passing legislation during its 2007-10 term under the proposed new system than under the current system.

Sunday, June 7, 2015

ReachTEL: Liberals Consolidate, But Who Should Lead Labor?

ReachTEL (state) Liberal 48.5 Labor 29.9 Green 15.8 Other/Ind 5.8
Interpretation: Liberal 48.5 Labor 32.9 Green 13.8 Other/Ind 4.8
Result based on poll taken as read: Probable Liberal majority (about 13-9-3, with 12-9-4 or 13-8-4 also possible)
Result based on adjusted interpretation: Liberal Majority (13-10-2)

Not long since the last EMRS poll suggested the Hodgman Liberal government was emerging from a period of disappointing polling, The Mercury has commissioned a large-sample ReachTEL that has recorded an even stronger reading for the party.  Indeed, this ReachTEL is not very much different from the only other one taken since the last state election (see ReachTEL: Liberals With Solid Lead).  It provides some very useful electorate-by-electorate data, the first since September, and also some very revealing polling on who should be the state Labor leader.

Wednesday, June 3, 2015

Poll Roundup: Whole Lotta Nothing Going On

2PP Aggregate: 51.8 to Labor (unchanged for last two weeks)
Labor would probably win election held "right now", in minority or with small majority

Yep, it's a no-change fortnight!  For the first time since the Abbott government was elected, my aggregate has shown exactly the same reading three Tuesdays in a row.

This week's polls were a 52:48 Newspoll, a 52:48 Essential and a 52:48 Morgan (by last-election preferences; 53-47 respondent-allocated).  The Newspoll and Morgan had effectively identical primaries (41-37 in the Coalition's favour with 13 for the Greens) while the Essential gave Labor three more points and the Greens three fewer, making the 52% 2PP look a little stingy.  (The usual explanations doubtless apply.)  I am still adjusting Morgan by one point based on evidence of its skew throughout this term (I don't yet assume that its apparent loss of most of that skew in the past few months is permanent) and so I aggregated these polls as follows: 51.1 for Morgan, 52.1 for Newspoll and 52.3 for Essential.  None of them did much by themselves, and between them they did nothing.  Here's the smoothed tracking graph:

This week's results are consistent with the Budget driving a small gain to the Coalition a few weeks back, but they don't exactly prove the Budget was the cause either.

For those who follow betting markets, there's been a move to the Government in recent weeks and months, with the probabilities implied by punters moving from more or less 50-50 to about a 62% implied chance of the Coalition being re-elected.