Tuesday, March 10, 2015

Poll Roundup: Neither Fish Nor Fowl

2PP Aggregate: 53.1 to ALP (-0.4 since last week, -2.6 in three weeks)
ALP would comfortably win election "held now"

Last week's Ipsos (51:49 to Labor) sparked a great deal of debate.  Largely ignoring evidence that Ipsos readings have been slightly more Coalition-friendly than those of other pollsters, sources set about arguing either that the result was real and Tony Abbott had spearheaded a miraculous recovery to neck-and-neck fighting from the brink of political death, or that the result was real and that it was because voters were over Abbott and assuming Malcolm Turnbull was about to be made party leader.

A comment by Jess Elgood from Ipsos flagging the latter as possible was the subject of a pollster pile-on in the Australian ("Fairfax pollster panned by rivals"  - possibly paywalled) with heads of Galaxy, Newspoll and Essential all quoted in disagreement, as well as Mark Textor of Crosby-Textor. In last week's edition I suggested that if Elgood was right we would start to see 55:45s and worse all over again.

Well, we did see one 55, from Newspoll, a move of two points back from last fortnight's 53, and following the normal pattern that a big move in one direction in a given poll tends to be followed by a move back in the other.  (This is because big moves from one poll to the next tend on average to be at least as much bounce as signal.)  However, the primary vote moves between the two Newspolls were small (Labor gained one point at the expense of Others), making it likely that Labor benefited from rounding this time and that the real poll-to-poll 2PP move may not have been much more than a point.

Also, the post-rounding two-point 2PP move in Newspoll was countered by a 1.5-point move (on last-election preferences) from Morgan, which came down to 53.5.  Bearing in mind Morgan's tendency to skew to Labor, that might as well be around a 52, and is actually quite a mild result for the Coalition.  Moreover, the Morgan change was off a 1.5 point increase in the Coalition's primary vote, and a 2.5 point drop for Labor that was mostly absorbed by the Greens.  (I'm sure if there was a drastic difference between the two weeks of Morgan's sample they would have made some noise about it by now.)

The third pollster reporting this week, Essential, came in with its third consecutive 53 for Labor, but this was off slightly better primaries for the Coalition than last week's result.  Given William Bowe's comments about Essential's week to week samples we can reasonably infer that since the first half of the sample was strong for Labor the second half would have been nothing too drastic.

My aggregate has trimmed 0.4 points off Labor to have them at 53.1, the same level as in late November and early December last year.  However, the closing in the margin is solely because of three 57:43 results from a few weeks ago falling out of the picture.  Weekly averages adjusted for the biases of various polls have hardly moved in the last few weeks, and it's possible that the previous 53 from Newspoll was actually closer to the mark than the current 55.   (That may seem funny given that the two-point move looks like a correction from the excesses of the previous polls, but that is what the evidence from other polls is saying.)

Here is the current smoothed aggregate tracker.  It shows a flare-up lasting about four weeks and then so far supports the view that we are just back to where we were before that.  We're not yet seeing a dramatic swing back as voters realise their hopes for a quick new Turnbull dawn are dashed, and nor are we seeing a heroic recovery to polling that the Coalition should find acceptable.  What we're getting so far is nothing so exotic, just a return to blandly bad, not-quite-uncompetitive 47:53s that don't suit any media actor's narratives.  The same voting intention as late last year but with the PM even more out of favour.


Two weeks ago there was an odd dip in Bill Shorten's Newspoll netsat, from +2 to -14, which I put down to poll-to-poll bouncing.  This Newspoll sees him back at a more typical score of -3, noting that that might be a little bit kind as the Newspoll is this week's most Labor-friendly sample.  This time I find more realism in Essential's picture which had Shorten on -5 in February, and -5 again in March.

The Australian's (possibly paywalled - "Tony Abbott rises but Labor has commanding lead") take on Tony Abbott's result this week was greeted with widespread mirth and rightly so.  Abbott's approval rating of an awful 28 was greeted as a "three-month high", which isn't quite literally true, but more importantly ignores the fact that there were only two prior Newspolls in the period!  Moreover, this supposedly herculean feat of personal approval weightlifting in fact left our PM still in the lowest 6% of all historical Newspoll approval ratings.  Also, while Abbott's netsat is up eight points from -43 to -35, the latter reading is still in the worst 4% of all time, and three consecutive readings of -35 or worse is something Julia Gillard never endured.

Essential's slightly milder medicine concurred with Newspoll by having Abbott's netsat up eight points in a month, from -33 to -25, but this was still the PM's second-worst reading to date.  Both pollsters had Bill Shorten continuing to lead as better/preferred PM, by 11 points (Newspoll) and 4 (Essential), a lead that is higher than expected for the relative closeness of the 2PP and a sign of how unpopular Abbott is.

Other stuff

Essential had some polling on which groups are considered to pay too much or not enough tax.  They are largely predictable by party support and self-interest except that it's notable that both Labor and Coalition supporters are quite likely to, and equally likely to, think religious groups should pay more tax.

Essential also had a surprisingly bearish result on "sending more troops to Iraq to help train Iraqi forces fighting Islamic State insurgents", with 36% approving, 50% disapproving and scarcely anyone strongly in favour.  While this might be just down to the oddnesses of Essential's respondent base, it is in some contrast to Morgan's association of the troop deployment announcement with the Coalition's rise in their poll.

No comments:

Post a Comment