2PP Aggregate: 51.8 to Coalition (-1.1 in a week, -2.5 in five weeks)
Coalition would win election "held now" with substantially reduced majority
Just over a week ago I noted that the Coalition had an attack of the February wobbles, a common pattern of government polling tanking around this time in an election year. A shock Newspoll result this week has seen this get a lot worse, as the Turnbull government looks rattled and tired and above all confused about its own tax reform steps, and for the first time in a long time, Bill Shorten has a spring in his step. A robotic spring, true, but a spring nonetheless. Government polling seems to be in freefall and no-one really knows where the bottom might be if it doesn't snap out of its tax funk sometime soon. I've started a Not-A-Poll for anyone with a view on when or whether in this term Labor might get its nose in front.
Since the bullish ReachTEL and the bearish Ipsos discussed in the previous article we've seen the startling 50:50 2PP Newspoll (down three), a status-quo 52.5 to Coalition from Morgan (which has recently leant to the Coalition, but this seems to be diminishing) and a status-quo 52:48 from Essential. Essential hasn't moved much for a long time, coming out at 51 or 52 for almost every poll since Turnbull became PM, but most of the others seem to be coming back to it rapidly, so I've cut its house effect and could well remove it altogether in another week or two. Essential's read that voting intention has barely changed since just after Turnbull became PM is so at odds with the story from the others that I can hardly blame Mark the Ballot for saying that he doesn't aggregate it because he doesn't understand its behaviour.
The 50:50 from Newspoll is the first draw Labor has achieved from any pollster since another 50:50 Newspoll in October. It should be kept in mind that the new Galaxy-run Newspoll seems to lean to Labor by about 0.6 of a point, a lean not affected at all by the switch from Abbott to Turnbull. All the same this is its biggest deviation from the aggregate so far from a poll that has been far from bouncy. It might just be an iffy sample but if not it's a massive signal and it alone took a point off the Coalition in my aggregate. After considering the primaries and my estimate of the (rapidly fading?) house effects I counted the Newspoll at 50.1, Morgan at 51.3 and Essential at 52.7, and this all knocked the Coalition down to 51.8, down 1.1 points on the revised end-of-week figure for last week.
Here's the smoothed tracking graph. The trajectory looks alarming for the government, but we should remember that "momentum" in polling is a very unreliable thing:
Another Mistake? Newspoll This Time
I'll get onto leaderships in a moment but first I want to point out that again we have seen published polling figures that apparently just don't add up. This happened with all the leadership questions in the January ReachTEL, and now it's happened with the question of which of Malcolm Turnbull and Bill Shorten is best equipped to handle climate change in The Australian's report of Newspoll. A (more minor) recent error in another poll probably only escaped publication because I caught it first.
I did mention that errors might be hard to catch in the new Newspoll because of its opacity, but this one appears glaring. According to The Australian's report, ("Malcolm Turnbull leads Bill Shorten on Labor's turf: Newspoll") Newspoll has Malcolm Turnbull preferred for his handling of climate change 51% to 23%.
According to the party breakdowns, however, Coalition supporters prefer Turnbull on climate change 59:10 and Labor supporters like their guy 65:14. But with the Coalition on 43% and Labor on 35%, then even accounting for rounding, Shorten must therefore be preferred on this issue by at least 26% of voters even assuming no Greens or Others voters prefer him on the issue. Perhaps there's some trick here like uncommitted voters being included in the issue poll despite not being included in the voting intention poll, but in that case Turnbull's rating among Green/Other and uncommitted voters would be through the roof. My money's on the headline figure simply being wrong.
Well done to @sorceror43, who very often spots such things, for spotting this one - and rinse and repeat for my call for polls to publish full crosstabs so that poll-watchers can gauge how common this sort of thing is. Once again I will be more than happy to publish any response that explains what is going on here so keep an eye on this article for updates - it may somehow be all OK, but I doubt it.
Malcolm Turnbull's Newspoll netsat was crunched from +22 (53-31) down to a mere +10 (48-38). Previous cases of a PM losing 10 or more netsat points twice in three polls affected John Howard in 1996, 1997 and 2001, Kevin Rudd in 2008 (three times in four polls), and Julia Gillard in February 2013 and again in August-September 2013 (three times in five polls). Coming off a honeymoon was a factor in the Rudd and first Howard cases, while the last Howard and first Gillard cases were both around this time in an election year. Shorten improved slightly to -29 (28:57), and Preferred PM narrowed by five points to 55:21 in Turnbull's favour (still a huge lead for a tied 2PP).
Newspoll also had issue comparisons for Shorten and Turnbull as better leader to handle issue. As I've noted the one for climate change is suspect. Of the rest Turnbull leads on the economy (58:22), health (42:41), education (42:38), national security (51:22), likeability (71-48) and decisiveness (66-43). One bad spot for Turnbull is that he is slightly more likely to be considered arrogant (56-49). (The last three are not competitive - a voter can find both candidates likeable, or one, or neither.) In the table graphic we find also that Turnbull beats Shorten on tax reform (48:28) and cost of living (42:33) but Shorten scrapes a one-point win on industrial relations (39:40). Comparisons between the parties rather than the leaders on a lot of these issues might be more useful now.
The economy question has a long history. Leads higher than Turnbull's 36-point lead (58:22) were seen by John Howard against Simon Crean, against a newly elected Mark Latham, and against Kim Beazley in his second term. Kevin Rudd also had such a lead against Brendan Nelson shortly after being elected.
Turnbull's lead on health should be a worry for Labor since never in the previous 18 askings of this question has the Coalition leader led. Education likewise, since a Coalition leader hasn't been near leading that since Howard vs Beazley in 2005-6.
Shorten's likeability score is poor (only Keating and Abbott were lower), and his decisiveness score ties Simon Crean's as the equal second worse ever (only Alexander Downer was lower). The rest is all to be expected.
Essential's respondents again show themselves to be a bunch of pessimists in terms of where economic indicators are going. More interestingly they have a controlled experiment showing that major party voters react differently to negative gearing policies when told they are Labor's. Newspoll found Labor's approach going well at this stage (47:31).
Essential also found that Others voters are more inclined to see the Turnbull Government as worse than expected than Labor voters are. Coalition and Others voters think asylum seekers on Nauru and Manus Island are well-kept; left-leaning voters don't. Essential also unsurprisingly found that hardly anyone likes outsourcing Medicare payments.
There is plenty going on on Senate reform at the moment, but I'm yet to see a poll on it.