Poll Roundup: Ipsos Stalls Re-spill Momentum

2PP Aggregate: 53.5 to Labor (-1.1 since last week, -2.2 in two weeks)
Labor would win an election "held now" comfortably

There aren't a lot of new polls this week so far but one of the two out has been significant and much discussed.  The fourth federal monthly Fairfax-Ipsos phone poll came out with a surprisingly good result for the Coalition (by recent standards), with Labor leading only 51:49 two-party-preferred by last-election preferences and the Coalition leading 42-36 on primaries.

Prior to this poll there had been widespread and increasing speculation of a spill against Prime Minister Abbott on either Tuesday or Thursday this week, but following the poll the idea seems to have gone to ground.  Suddenly there are reports of more acceptance that Abbott should be given a few months, the NSW election and the Budget to see if the Coalition can recover.  The two events aren't automatically connected, but it looks like this single surprise result has had quite an impact on leadership speculation.  Maybe that impact is itself temporary and polls next week will spark another round of fervour, but we are approaching a point at which changing leaders becomes a really bad idea for a while because there will be not enough time to clear the air before the NSW election.

The other result this week was Essential remaining at 53:47 to Labor.  As noted before, because Essential often doesn't move when other polls do, it's not clear what we should read from this.



How important is the Ipsos poll?

The first thing to bear in mind is that Ipsos federal phone-polling results so far have been slightly more favourable to the Coalition, on average, than those of other pollsters.  This has also been the case, to a lesser and less consistent degree, in state level polling.  Quite why this is so is not obvious, and with so little data it's impossible to say yet whether it's an intrinsic feature of the pollster's methods, or a temporary divergence only.  Established phone pollsters will often have short runs of poll results that lean strongly to one side or the other, but these are taken less seriously if the pollster has a long term track record of neutral results.

Anyway, on average so far I've had Ipsos results at 1.4 points better for the Coalition than my aggregate at the time, though the difference from the current aggregate reading is one we should be a bit careful about since the voting intention picture is volatile.  There's certainly reason to suspect, until we see more evidence, that this 51 to Labor should be treated as at least a 52, and for the time being my aggregate is doing this.

But even if this is really a 52 to Labor result, that's still quite important.  As a lone result, it doesn't prove the Coalition has further recovered from last week, but it's at least inconsistent with last week's signs of recovery being overstated.  It's strongly consistent with the Coalition's awful polling from three and four weeks ago having been a temporary blowout, and not a shift to a new level of lasting hopelessness.  It might turn out that the "recovery" is fragile, but we don't know that yet.

The importance of the result is that, for this week at least, it undermines the case that Abbott would be un-reelectable.  If the party still has a chance to poll competitively under his leadership, then the case for removing him is weakened.  While some Coalition MPs might want to remove Abbott in pure disgust at internal processes (such as the perceived control of his Chief of Staff Peta Credlin and the conflict of interest with her husband Brian Loughnane being party director) for many backbenchers this is more about political survival.

It's also important to the Coalition's future chances that if they do remove Abbott, they can sell the case for doing so (as Labor failed to do with Rudd, largely because in that case the case for doing so was so weak.)  The Coalition's polling of a few weeks ago by no means proved they could not win the next election under Abbott.  But had he been removed then, there would be much acceptance (probably exaggerated acceptance, even) that the party was headed for certain defeat under him.  While that case might be debated, at least no-one would be in much of a hurry to bring him back.

But if he were removed right now, with his side seemingly on a major poll upswing, there would be an exaggerated belief that he had been cut down while on the path to a Keatingesque win.  Ideas of "momentum" in polling are usually bogus, and assessed objectively, the government's chances of winning the next election if Abbott is the leader have not changed much in the last two weeks.  However, perception is what matters, and the perception is that the Coalition is "back in the game".

Where did the blowout go?

It's nice to be able to say that event X caused polling drop Y and that this will result in outcome Z a few weeks hence, but sometimes there is no one clearly "best" explanation for what's going on.

A widespread theory is that voters are "factoring in" an expectation that Malcolm Turnbull will be PM very soon.  Deputy Prime Minister and Nationals leader Warren Truss has dismissed it as "a line that must have come straight from ALP headquarters".  (It was flagged as "possible" - which it is - by Jess Elgood from Ipsos, and that line could as easily have come from pro-Turnbull sources in the Liberal Party as from the ALP, but Truss's was an emphatic attempt to put it back in its box anyway.)  The major supporting arguments for this view are that a similar thing happened to Labor in March 2012 and that there is a sudden discrepancy between Abbott's ratings and his party's 2PP scores (seen in both Ipsos and Newspoll).

We'll know soon enough if this is true - if it is, then once it becomes clear Abbott is hanging around for more than a month or so, we will start seeing 55:45s and worse all over again.

I am actually finding the factoring-in thesis all a bit artificial.  (Yes there was a bounce to Labor in 2012 but many other such failed spill situations have generated no bounce for the party in question).  I'm more partial to the idea that we are just seeing polling go back to around where it was before the Coalition's standing was damaged by the Prince Philip debacle followed by the loss of the Queensland election, and doing so more quickly than I expected.  Even in this case, Abbott's own long-term chances don't return to what they were, as the spill motion has weakened his leadership and greatly increases the chances he will be booted later in the year compared to if it had not happened.

A third interpretation is that since Abbott is more unpopular than his government, the Opposition must be doing something dumb.  As I noted last week (Coalition Polling Improves, Abbott's Doesn't) this kind of disconnection between 2PPs and leadership figures is rather unusual, one notable prior case coming in 1992 when Paul Keating was up against very radical Liberal economic plans.  However, I just don't think there is really that much attention on the ALP at the moment, nor that they are making suddenly obvious mistakes on issues voters care about.  Also, while there is a strong correlation between leadership figures and 2PP scores, it is not that set in stone that every rating has to follow it.

Where we are now

My aggregate moved 1.1 points this week to show Labor with a 53.5:46.5 lead.  About 0.3 points of the move came from the weekly data recency reset (with all those 43s from three weeks ago no longer carrying much weight) and most of the rest was caused by Ipsos.  Here's the spiky (unsmoothed) tracking graph, just for a change:

According to it, we're where we were at the end of last year, but because of the volatility lately and because there was only one major poll this week, the estimate is quite rubbery at the moment.  It's easy enough to draw a straight line from the end of the graph and say we'll be at 50:50 in a few weeks' time, but these sharp shifts usually do not last like that.  Another thing that's notable is that the ups and downs are steeper in the last few weeks than in the Coalition's term to date.  However, this is exaggerated because of the two weeks of friendlier readings in mid-January, which were based on very little data.

Leaderships and other matters

The Ipsos poll showed Abbott's net satisfaction at -30 (32-62), up eight points in a month but still worse than Abbott got in the comparable Nielsen series.  The poll also showed some loss in support for Shorten, with his rating coming down from +8 to zero, though the former result was atypical. The poll concurred with Newspoll in finding the preferred Prime Minister gap closer (in this case 44-39 to Shorten, down from 50-34) but this lead is still much larger than expected for the current 2PP result.  Indeed, normally with a poll's 2PPs so close, the Opposition Leader wouldn't be in front at all.

Ipsos had a series of attribute questions comparing Abbott and Turnbull , on all of which Turnbull benefited from the rose-coloured glasses view of him that always shows up in polling these days.  All of Abbott's scores were poor and it was especially notable that only 21% of voters think he has the confidence of his party.  That score is not a reliable predictor of whether or when Abbott will be removed, but it does indicate that rolling him won't surprise anyone much.

Ipsos results on preferred leaders within the Liberal Party were similar to other results recently.  The pollster had Turnbull 39 and Bishop 24 leading Abbott 19 as preferred PM among all voters, with Abbott slipping from equal second with Bishop in November to third.  Among Coalition voters Abbott had slipped, but not as much as might have been expected, and continued to lead with 38 to Turnbull's 30 and Bishop's 21.  While this "lead" is sometimes compared to Julia Gillard's leads over Kevin Rudd, it's worth remembering that Gillard had only one serious opponent whereas Abbott has two, and more than half the Coalition voterbase prefer one or the other to Abbott.

An interesting aspect of the latest Ipsos poll was that voting intention for the Coalition was slightly stronger among female voters than male voters.  It is usually the other way around.

I was interested to see polling on this week's most prominent subject of left-right culture warring, the attempts to have Prof Gillian Triggs removed from her job.  Depending on who you listen to this is either a gross example of corrupt governance or an attempt to clean up a partisan Human Rights Commission.   Labor thought there was enough juice in it for two consecutive attempts to move censure motions, the second of which was shut down with a motion from Christopher Pyne that Bill Shorten be no longer heard.  Essential had polling on perceptions of the HRC.  56% actually had a view with those views breaking 34-22 in the HRC's favour, and along partisan lines but with Coalition voters only weakly opposed.  It is possible that the two previous questions (on levels of intolerance in Australia) might have influenced responses to this question in the HRC's favour, but it could also be argued that the question's preamble gave the anti-HRC side of the story.

A question on counter-terrorism found notably that Labor supporters tended to agree with Coalition supporters in supporting increased spending, and, regrettably in my view and that of Greens voters, there has been a six-point increase in the percentage of voters who are clamorous to have freedoms curtailed in the interests of stopping terrorism.  (That said, the question design could encourage the respondent to think that the freedoms are just being taken away from someone else.)  Abbott scores a +13 net rating (46-33) for his handling of terrorism, with a predictable partisan skew.

(Note also this comment about Essential's national security question - the option that current laws have already gone too far towards a security state was not included and should have been.)

Not-A-Poll: Abbott Rallies!

One of the two opt-in not-a-polls on the sidebar asks readers when they think Abbott will be removed from office.  It's not a representative sample as my reader base probably skews left.  After 178 votes, 85% think he'll be gone by the election, 10% at the election by losing it, and 5% think he will still be leader and win.  What's mildly interesting here is that none of the first 111 voters on this question had Abbott winning the next election, but since then nine out of 67 have.

Any more poll results late this week may be added in updates.