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.


This week saw Newspoll give Bill Shorten another big raspberry, with his equal worst Newspoll netsat so far, of -18 (32 satisfied, 50 dissatisfied).  Shorten's netsat was in fact worse than Tony Abbott's of -15 (38-53) and that's only the third time that's happened.  The first was the late October 2013 poll (with Abbott still in honeymoon mode and Shorten not yet well established) and the second was early April 2014 (probably caused by sample error as that poll was an outlier).  This poll isn't an outlier, so as an indicator of the relative fortunes of the two, this is a big result.

Shorten's relative unpopularity seems no big deal in itself since the relationship between Opposition Leader netsats and their party's standing barely exists, but until Abbott, any OL who polled ratings this bad was invariably either removed by his party or beaten at the election.  Then again, all previous PMs who have faced a spill attempt (as Abbott has) have either not made the next election or else lost at it, so you can spin these kinds of precedents any way you like.

Essential came riding to the rescue (mostly) with attribute polling.  Despite their views of Abbott recovering since February (to about the same level as December) while their views of Shorten had him at new lows on several indicators, Shorten continued to lead Abbott on every head-to-head comparison except for a (nil-all?  28% apiece) draw on "Good in a crisis".  While Essential's measures of Shorten's ratings are generally softer than Newspoll's, the natural conclusion is that voters are disappointed with Shorten for reasons not listed by Essential - perhaps that they find him boring and over-rehearsed or doubt his ability to win.

Abbott recorded an unremarkable 41:37 lead in the incumbent-skewed "better Prime Minister" indicator - again well below the batting average for a 2PP of 48.  It was, however, Abbott's equal biggest "lead" since the 2014 budget - he also led 41:37 once in August and once in September.

Same-Sex Marriage

Same-sex marriage polling by Essential this week showed 59:30 in favour.  Given what is known about mode of interview effects concerning minority-discrimination issues, phone polling would be expected to (and in Australia does) give even stronger results than online polls like Essential or robopolls.

Even so, no real poll for many years (I of course exclude opt-ins, "elector surveys" by politicians and so on) has supported Senator Fierravanti-Wells' claims that a "silent majority" are opposed to reform.  This majority proposed by the Senator must be so silent that they run away from their phones and disconnect from their computers to avoid being polled, even if they are doing it in a way in which their views won't be noticed by anyone else!  In fact there is no more evidence for a "silent majority" against this issue than there is for Parliament House being the permanent home to nine invisible, inaudible, ballet-dancing northern white rhinoceroses.

Opponents of same-sex marriage in the lead-up to the Irish referendum also hoped, and in cases prayed, that the "silent majority" would come along and save them from the terrible non-threat facing their nation, but they were wrong.  The "yes" vote in the Irish referendum was more or less exactly the final average polled "Yes" vote for respondents classed as Yes, No or undecided, and about eight points lower than the Yes vote with undecideds excluded.  The latter is much the same modest margin as that by which SSM ballot measures have often underperformed compared to polling in the USA.

Essential showed supporters of all parties in favour of same-sex marriage: Greens supporters 93:3, Labor supporters 70:21 and Coalition supporters 49:38.  Others voters were weakly against (39:51), bearing in mind that probably 35-40% of the Others vote at the moment resides with religious-right and far-right micro-parties.

However Essential also published polling on how voters rated the issue in terms of importance to their votes, which casts some faint light on the public-opinion basis for the issue being messy at a parliamentary level. (The messiness also results from politicians being more conservative than their electorates, and from jockeying for the taking of credit).

What Essential shows is that while voters strongly give at least nominal support for change, voters who support same-sex marriage are less likely to say the issue would affect their vote, than those who oppose it.  It can be inferred from Essential's poll that of those who support same-sex marriage, that about 29% would say they are much more likely to vote for a candidate who agrees with them, 29% a little more likely and 42% wouldn't take it into account.  For opponents, however, about 50% would say they're much less likely to vote for a supporter, about 23% a little less likely, and 27% wouldn't care.  Coalition supporters who support same-sex marriage are especially unlikely to say that the issue would influence their vote.

So within the Liberal support base, it's mostly the supporters who are the silent majority while the opponents are the loud, noisy, organised minority who may indeed threaten to take their vote elsewhere, until someone asks them the obvious question: where?  To religious right micro-parties and then preferencing the Coalition anyway?  To any conservatives worried that letting this reform pass on a conscience vote might lead to the base deserting and a galling electoral loss, I have these two words: David Cameron.  (Indeed, because of our preferential system, "conservatives" have even less to worry about in supporting same-sex marriage here.)

There is also the usual problem that polling of this kind is hopeless at predicting whose votes will actually change.  Green voters indicate the issue as a big one for them, but mostly won't switch to voting for, or even preferencing, a Coalition candidate just because that candidate supports same-sex marriage. And given that 30% of voters will say that just about anything could change their vote, there's not much to see here either way in evidence for the issue moving votes.

I have seen some reports of commissioned findings on support for a conscience vote in various electorates, but this is the sort of polling where it's necessary to see the exact question before commenting.  (PS: some findings on allowing a conscience vote have surfaced.  The question design seems fine though it is not clear whether other questions were asked first.)

Seat Polls

A number of commissioned federal seat-polls have surfaced recently.  I covered the four Tasmanian AFPA polls and the ACTU Braddon poll in the wood waste article; collectively, what you make of this lot depends on whether you expect the skew to Coalition seen in ReachTEL's Tasmanian polling for the 2013 federal election to continue.

Also seen have been an ACTU ReachTEL of Hindmarsh (showing Labor struggling), while swings in Page, Corangamite, Swan and Leichhardt were typically said to be a few points against the Coalition on primaries, but with Labor not gaining anything.  However, it's clear that these polls include an Undecided option and are hence not comparable directly to election results; they surely imply at least some kind of 2PP and primary swing to the ALP, the question being how much.  (From what we can see, with the possible exception of Swan, it sounds like the swings are not that great.)

Broken record time again: seat polling in Australia is struggling badly enough without having to make sense of piecemeal reports of polling conducted by groups that don't release it even when they have no reason not to.  Why not at least release full voting intention figures?

No comments:

Post a Comment