Monday, May 18, 2015

Poll Roundup: 2015 Budget Polling

2PP Aggregate: 51.8 to Labor (-0.5 in a week, closest since last October)
Labor would probably win election held "right now", with small majority or in minority

Apologies for the boring heading.  I was going to call this piece "Poll Roundup: Budget Less Than Random Noise" but passing judgement on a week of polling with only four of the six polls released would have been risky, and thus it proved once Morgan came out.  With two polls implying a budget bounce to the Coalition and three implying no change, my aggregate moves to its best position for the government since the week of 20 October last year.

It's possible that there is really no bounce and that the two good polls are just down to sample noise, but the results are consistent with a slight Coalition gain from a budget that has been fairly well received by voters.  Still it is nothing so far as dramatic an impact as the usual storm of Budget-poll-fuelled commentary might suggest.

This isn't unusual at all; last year's shocker was the exception that proves the rule.  Budget polling is a vast source of excitable nonsense but most budgets aren't a big deal immediately to the average voter; see Mark Graph's first and second laws of budget analysis for more of this.

That's not to say that the noises surrounding the Budget are necessarily meaningless in the long term, and this week we did get some insight into possible election strategies concerning economic management.

This week's polls

With everyone in the field following the Budget we had a 53:47 for Labor from ReachTEL that looked close to rounding to a 52 (indeed the company very kindly told me that their estimate was 52.64%), a 52:48 from Galaxy that could just as easily have rounded to 53, and a 53:47 from Newspoll off primaries that would more often than not have rounded to 52.

The odd one out over the weekend was a 50:50 from Ipsos, the first poll by anyone to have the Coalition level since Essential thirteen months ago.  As with the 51:49 from Ipsos in March, this poll will create a lot of attention, but a lot of that attention will come from media consumers (or pundits) who have either not noticed that Ipsos leans to Coalition or have (in cases wilfully) not paid attention to the results in the other polls.  Dire examples of the sad disorder that I hereby christen Housepoll Myopia (assuming that the poll commissioned by your agency reflects reality utterly, while ignoring all other polls) have been seen in some Fairfax online pieces, and I refuse to link to them in case it is contagious.

The other poll attracting a lot of spurious commentary was the Galaxy, in which a turnaround from 43:57 to 48:52 for the Coalition was widely supposed in some equally myopic Murdoch tabloid reporting to be a big deal.  The previous Galaxy had been in early February with the Coalition facing a leadership crisis; the poll merely confirmed what we already knew.

The Morgan out today was 51.5% to Labor by Morgan's calculations of last-election preferences (51% respondent-allocated, which is also what the primaries would usually imply).  This is actually a very strong result for the Coalition given that the series has a long-running skew to Labor.  On that score this result has finally triggered a re-estimation of the house effect of Morgan's 2PPs (after my adjustment for the primary votes), down from 1.5 points to 1.

After adjustment for house effects and primaries, I aggregated both Newspoll and ReachTEL as 52.6 to Labor, Galaxy as 52.2, Ipsos as 51.1 and Morgan as 50.3.  Here's the current smoothed tracking graph, which I'll update tomorrow after Essential if Essential changes it:

With one little wobble a few weeks ago, the pattern is of a continuing but gradual move back to the Coalition.  It's not in the nature of these trends to reliably continue in the same direction.  However, bearing in mind that they probably only need about 49% 2PP on the day to win, it would be extremely foolish to write off this government's chances of re-election at this time - whoever its leader.


This week's leadership results continue to show Tony Abbott bouncing back from some of the worst polling an Australian PM has ever had.  Newspoll saw his net satisfaction up from -19 to -13 (39-52).  This is Abbott's least bad result since polling -11 eight months ago, and his second best since his -7 in the dubious Newspoll of early April 2014. Ipsos had Abbott at -8 (42-50), his best netsat from the combined Nielsen/Ipsos series since December and his highest approval from it since last April.  ReachTEL's harsher net ratings (which count Satisfactory as neutral) had Abbott at -21.9, up 5.3 in a month and his least worst since late November - it's also notable that Abbott's ReachTEL ratings are lately less polarised than they were.  Essential last week was perhaps the least positive for the PM, with his -18 there up seven points but only his best from them since January.  A graph posted by Mark the Ballot shows "remarkable agreement" by several pollsters in the pattern of Abbott's recent ratings.

Bill Shorten's ratings remain mediocre - stable at -9 in Essential, up five to -11 (the fifth time he's polled this) in Newspoll, up 0.3 to -15.8 in ReachTEL and down two to -4 in Ipsos.

Discordant results in preferred/better Prime Minister polling just showed the significance of different polling methods.  ReachTEL polled a forced-choice Abbott-vs-Shorten question and had Shorten leading Abbott 57.2 to 42.8. However all the other pollsters allow a don't-know option.  Of these, Newspoll had Abbott's first lead since last September (albeit only 41-40), Ipsos had Abbott ahead 43-38 (his first lead in the Ipsos/Nielsen series since last April), and Essential had Abbott up 35-32 (his first lead since November).  I can't mention these results without again banging on about what a sloppy and skewed indicator preferred-leader scores with a don't-know option are; that said, we don't really know yet if forced-choice ones are trouble-free, or just cause everyone not at least slightly enthused with the PM to prefer his/her opponent.

The Budget

This is, alas, the last of 30 budgets to be polled by Newspoll's stable phone-polling method, with stable question wordings.  Hopefully the wordings will remain the same for the new order, even if the methods won't.

This year's budget was rated good for the economy by 46% of Newspoll respondents, and bad by 28%.  20% said it was good for their own financial circumstances, and 30% bad.  The latter might sound negative, but when this year's results are added to the long-term Newspoll graph, it can be seen that this is actually a pretty good result.  Voters tend to be more negative about the impacts of Budgets on themselves than on the economy as a whole.

Newspoll budget ratings for personal outcomes and economic outcomes, 1988-present (selected years labelled)
This year's budget (the dot is the one below "2015") is the best received on both scales since the first Rudd government budget in 2008. It rates five points above the long-term average on impact on the economy and ten points above it on personal impact.  There may be a relief factor in these results after last year, but historically this looks like a fairly good but far from stellar reaction.

51% said the Opposition would have done no better, and 36% say it would have done better.  That might sound good for the government, but as I pointed out in last year's instalment (see the graph) this is an indicator that skews to the incumbent government.  Indeed for a budget with an economic net rating of +18, the expected net "Opposition no better" rating is +19, four points better than the +15 recorded here.

The partisan breakdowns were as follows:
* Good for the economy: Coalition supporters 77% yes 5% no, Labor supporters 29-46, Greens/others (inferred) 19-39
* Good for own situation: Coalition 31-12, Labor 14-41, Greens/others (inferred) 11-44
* Opposition could have done better: Coalition 5-90, Labor 68-21, Greens/others (inferred) 38-31.

These are results in Budget polling for pollsters other than Newspoll:

* Galaxy had Hockey with a preferred treasurer rating of 34%, though the rating for his opponent Chris Bowen seems not to have been published yet.

* Galaxy had 24% saying the Budget made them more likely to vote Coalition and 29% saying less likely.  Most of both probably weren't telling the truth, and these results are pretty much useless.

* Galaxy had 43% agreeing that the Budget would stimulate small business (through the small cut in small business tax and the increased range of immediately tax-deductible purchases) to 28% disagreeing.

* Galaxy have 30% believing that the Budget improves the government's economic credentials and 28% saying it shows the government to be poor economic managers.

* ReachTEL had 16.4% believing they and their family would be better off under the budget to 30.3% worse off, a very similar result to Newspoll's.  Small business owners were most likely to say better off, but were also likely to vote Coalition.

* ReachTEL had voters evenly split between Joe Hockey and Scott Morrison as the most effective promoter of the 2015 budget, with very few picking Tony Abbott.  There was an interesting partisan skew here in that Greens voters tended to join Coalition voters in preferring Hockey on this question, probably because Greens voters dislike Morrison for reasons unrelated to the budget.

* ReachTEL had an amusing question on the budget as compared to last year's.  65.8% of Coalition supporters said the budget was "strong" or "very strong" compared to last year's (cf 5.4% weak or very weak), while Labor supporters (37.6 to 12.5) and Greens supporters (40.8 to 8.0) tended to find it weak.  Here we see pathological opposition at its most blatant and illogical since ALP and Greens supporters generally detested last year's budget yet are rating this year's as weak or very weak by comparison to it.

* Ipsos had 28% saying they would be better off under the Budget and 33% worse off, but bear in mind that this is from a Coalition-leaning sample.

* According to AFR, Ipsos had 52% saying the budget is fair (33% unfair), 52% satisfied with it (35% not), 54% saying the budget would be good for Australia and 54% saying it is economically responsible.

* Ipsos found that almost everyone (81%) supports the small business tax cuts.

* For more of the Ipsos findings see their press release.

* Essential last week found 56:30 approval for removing the pension from asset-rich retirees.  Support was weakest from Greens voters.

* Essential had Joe Hockey with a -18 netsat (30-48), up six points since March.

* Essential had Hockey preferred to Bowen as Treasurer 30-22 with nearly half the sample undecided.

* Essential respondents oppose employment/looking for work tests for access to childcare and early childhood learning.

Is Labor Losing The Plot?

Beware, big-picture waffle ahead.

Overall in this Budget the Coalition basically gives up on the obsessive drive towards a quick return to surplus that made its decisions last year so unpopular.  There is evidence that voters' criticisms of the 2014 budget direction have been listened to, but at the cost of abandoning much of the ideological direction that the Coalition intended to take, and indeed of anything but pretence of being really different economically to the government before.  The government seems to have finally realised it won the 2013 election mainly because Labor self-destructed.  It will have to be enough for it to say that it tried to fix the budget deficit but wasn't able to do so.  The Senate will be scapegoated but as unrepresentative as the crossbench is, no-one should fall for it.  The real cause is voter backlash, since the Coalition could easily take blocked budget measures to a double dissolution if voters didn't dislike them even more than the crossbenchers do.

Labor too seems to have realised it was thrown out mainly for because of its internal chaos, but is risking being seen as out of touch with economic concerns while saying it - especially while pushing an economic plan that is a curious mix of populism and (rather dated) technocracy.  The policy direction sketched out by Bill Shorten's budget reply speech is that the Opposition will try to convince voters that they can have tax and revenue cuts and increased spending, and fund this just by cracking down on tax-dodging by multinationals (as if no-one ever thought of this or if it ever worked all that well before.)  Is anyone really going to buy this at the election?  Is it really so different to what we've been getting from PUP?

It is one thing to say Labor weren't really thrown out for their approach to debt and deficit issues and quite another to say that they can simply thumb their nose at the concept of economic restraint and waltz back into office without really taking it seriously.  Indeed by so doing they risk legitimising the myth that that was why they were kicked out last time.  In Shorten's reply speech and subsequent interviews I suspected signs of an Opposition that has got carried away with what is really only a modest (if very persistent) polling lead and that thinks beating Abbott will be much easier than it actually will.

Of course, it's in the nature of Abbott that he may do something outrageously silly at any moment, but if he doesn't then I'm not now convinced Labor will beat him.  They still have a lead that would win them a fictitious election "right now", but that is without the campaign that goes before it.  Governments that trail narrowly during the second half of an electoral cycle tend historically to be re-elected, and this especially applies to Coalition governments.

By the way, with my scientist hat on I found Shorten's plan of writing off student loan debts for science graduates quite misdirected.  Generally people don't seek a career in sci/tech because they figure that they will get 2.5% richer than by doing something else; they pick it because it is what interests them.  They're less likely to be bothered about a higher tax burden when they reach a substantial income than they are to be bothered about long-term job security, which Shorten's proposals do nothing about.

I'll update this article with more Budget-related polling until it is time for the next update.

Tuesday: Essential:  Essential came out with 52:48 (same as last week) but judging from the primaries it must have been very close to 53:47.  I aggregated it as 52.3, which changed nothing.  So that's four votes for no bounce vs two votes for bounce, the likelihood being that there was actually a small bounce (though it might still just be random noise, or voting intention could have drifted to the Coalition anyway).

The government scored a +1 netsat for its handling of the budget (compared to -22 last year).  Voters were neutral on whether they were more confident in the government's economic performance.  Both these questions split on predictable partisan lines, but the latter case was more of the cognitive dissonance seen in one of the ReachTEL questions, since the left voters saying they are less confident in the government's economic performance are the same ones who hated last year's budget much more than this year's.  Voters overall thought the budget was better than last year's (45:15) but Greens voters said they disagreed (albeit weakly, 28-33).

A question on the impact of the Budget on various groups found much better net ratings than last year's for its impact on all groups except the well-off and "the economy overall" (the net rating for the latter was the same as last year, +8, but with both "good" and "bad" down by ten points.  Note that last year's figures for the economy overall contrasted very starkly with Newspoll's.)  Despite the generally reasonable reception for the budget overall, when respondents were offered a series of responses to the budget they took the negative side of all of them.  Finally a question about the so-called "budget emergency"  succeeded in smoking out one-third of the Coalition support base who believe that things are grim and that this budget isn't helping.

A note about this week's Morgan: it was only taken over one weekend instead of two, and there has been some discussion on Poll Bludger suggesting that it would have had a smaller face-to-face component, and hence a smaller pro-Labor skew, than normal.  There has been so little documentation of the proportion of different ingredients in the Morgan skew that I am not inclined to make any changes to one poll reading based on this.

1 comment:

  1. Excellent stuff as usual. I have given you some publicity so my reader should look at this.