Tuesday, November 13, 2018

Poll Roundup: Well That Wasn't Much Of A Honeymoon

2PP Aggregate: 54.8 to Labor (+0.8 since last week) by 2016 preferences
54.2 to Labor with One Nation adjustment
Labor would win election "held now" with a very large majority 

It's been a while since my last federal poll roundup.  At that time the Coalition's polling was recovering from the shock caused by the messy and (to the public) inexplicable coup that deposed Malcolm Turnbull as Prime Minister, and it was too soon to read anything into what we were seeing.  Because the Coalition's polling was in recovery mode but the new Prime Minister was still in a polling honeymoon period it was a matter of waiting for things to settle down to get a feeling for how competitive the Coalition really was.

On my aggregate, the recovery from a post-coup low of 43.9% peaked at 46.7% after seven weeks, and since then things have been getting worse rather than better.  Furthermore, since the defeat in Wentworth, they have been getting worse faster, at least if this week's shocker Newspoll is anything to go by.  The Coalition's current position is worse than at any time with Malcolm Turnbull as Prime Minister, and also worse than all but the worst few weeks under Tony Abbott.

We haven't had much national polling data since the last roundup article.  The government's last three Newspolls had 2PP votes of 47, 46 and 45.  By last-election preferences after considering the primary votes I aggregated these as 46, 45.4 and 44.3 (Newspoll adjusts for a shift in the preferencing behaviour of One Nation voters since the 2016 election).  Essential has returned 47, 47 and 46, which I converted to 47.2, 47.1 and 45.6, and there was a lone Ipsos at 45, which I entered as 45.1.

Since the change at the top, the average Newspoll 2PP for the government has been 45.3 while the average from Essential has been 46.3.  This is even though Newspoll's preferencing method should on average make Newspoll about 0.9 points better for the Coalition than Essential is.  It is pretty common for dramatic events to produce shifts in how the house effects of different polls relate to each other (for instance, when Turnbull first became Prime Minister, Morgan polls shifted from being Labor-friendly to Coalition-friendly compared to others for a while.)  For the time being I am not making any adjustments on account of this, mainly because only two polls are regularly active, the Ipsos series being too volatile, infrequent and quirky to act as a useful third opinion.

There was also a Queensland-only Galaxy with a 50-50 2PP that was reported rather breathlessly by the Courier-Mail.  However, while that 2PP is better than one might expect from the national polls and other recent Queensland breakdowns, it is still a 4.1% swing to Labor, which would account for eight Coalition seats on a uniform swing (probably more like seven in reality).  As such, the poll shows the Coalition losing the election in Queensland alone and it is also no improvement on the same pollster's reading under Turnbull.  Various issue questions were reported in a glowing light but there is a long history of right-wing tabloids commissioning dodgily worded issues questions.  Since the Courier-Mail can't be bothered reporting the questions verbatim in its article, I can't be bothered linking to it either.  A final reason for caution (or on the Coalition side optimism) about this poll is that there is a long and curious history of Labor's performance in Queensland not living up to prior state-level polling.

The impact of Wentworth

In the Newspoll immediately following the Wentworth by-election, the Coalition lost 1 2PP point, Scott Morrison's net satisfaction dropped from +7 to -3, and Morrison's "lead" over Bill Shorten on the skewed Better PM indicator dropped from 11 points to 8.  The full impact of the Wentworth loss might not have been felt immediately because the result was not completely clear a week later.

Bad by-election results often do have polling consequences - it seems that if a party performs disappointingly in a by-election, a few of its supporters drift away.  The most comparable example is the 1992 Wills by-election in which Labor lost former Prime Minister Bob Hawke's seat, also to an independent.  In that case the damage was more severe - Paul Keating copped a 17 point netsat hit, his Better PM lead dropped by six points (taking him out of a lead that he had just acquired) and the 2PP damage was worth about two points.  Keating did eventually win but the loss of Wills was the end of a honeymoon period in which Labor's polling had been rapidly recovering.  It took five months to get back to where he was in 2PP and Better PM terms and an election win to get his personal ratings back to where they were before the loss.

However, bad by-election results that are a symptom of existing unpopularity (eg Ryan 2001) do not necessarily have any impact on polling.


Much was made of Morrison's early personal ratings as a threat to Labor by commentators keen to drum up hope among their right-leaning audiences.  However, thus far Morrison has simply confirmed the trend of new PMs' personal ratings peaking early.  He reached a netsat of +7 after six weeks, which has since slipped to -8, and a Better PM lead of 13 points after four weeks, which has since slipped to 6.  It would not be surprising to see these slip further.  Essential also found Morrison's net rating dropping sharply, from +15 to +4 in their case.

Bill Shorten's last three netsats were -16, -13 and -15.  The -13 was his best result of this term, his last result better than that being a -12 in May 2016.  His dissatisfaction rate of 50% in the last two polls has been his lowest since polling the same just after the 2016 election.  Essential also has Shorten's rating improving, but only to its best level since May.

Is the Coalition's position recoverable?

Historically in terms of the 2PP and time to go til the election, yes.  Assuming an election in May 2019, the Howard government was in roughly the same 2PP position with the same amount of time to go in the leadup to its 2001 victory.  While the Tampa affair followed by the S11 attacks greatly assisted Howard in that year, his government was recovering strongly before those events, and may well have won anyway.  However, in this case a major reason for the government's unpopularity (the leadership chaos during the term) is not rectifiable, and we are likely to again see "transactional costs" as a factor dragging the government down during the campaign - whether in the form of sitting member retirements, media distractions or poor campaign planning.  The Coalition has the added distraction of having lost its majority mid-term, which is likely to prove a continuing source of parliamentary embarrasments.  So while betting odds currently have the Coalition's chances of winning at around 26%, in my view their position is rather worse than that.

One would normally expect tightening as election day approaches and Labor's policies come under more vigorous scrutiny.  However, in making Morrison PM the Liberal Party was choosing the option most likely to end in a respectable loss, as opposed to the higher-risk Dutton option.   I could see the Coalition getting back to a 48-52ish result from their current mess, but I am not sure they will do that, and anything better than that from here would be a surprise.


  1. Considering there are state elections coming up in NSW and Victoria prior to the next Federal election, is it possible that all the angst and built up dislike of the Federal Government could be spent in those state elections? Providing a more respectable outcome for the Federal Government in those states?

    Also, historically speaking, aren't Governments that force themselves to full-term usually on the losing end of things? I've heard there is a correlation there.


    1. I don't think voters vent as much about federal issues on state oppositions as state governments, eg there isn't a correlation between federal government polling and the performance of governments of the opposite party at state level. In the case of NSW if the federal election isn't held before that, then if there is a serious venting by voters (which would cause NSW to be lost) I'd expect that to flow over into recriminations that affected the federal campaign. I think there's a fair bit to be said for the federal government going before NSW if possible, say in early March.

      Governments that go substantially over full term have tended to lose.

    2. Ah I see, I got a bit mixed up there. So it's only when the State government is the same government as the Federal government that the correlation exists? That makes more sense to be honest.

      And yeah, sorry, that's what I meant. Over full term. Which is what this government is doing, isn't it? The last possible election date is ~may when they're arguing for it. Or at least that's what I've been told.

    3. May is the last possible date for a half-Senate and a House election at the same time. However if the Government chooses to separate the two it could have a half-Senate by May and a House election as late as November. This would be foolish, but it would take something like that for the government to go beyond three years and make itself comparable to other governments that did so.