Image: J

EMRS: Lib 49 Lab 27 Green 22 Ind 2
Interpretation: Lib 49 Lab 31 Green 18 Ind 2*
Outcome if election was held now: Liberal Majority (13-14 seats)

* (Ind may be higher depending on candidates)

A rather skimpy EMRS poll release is out. On the headline rate, Labor is up four points, the Greens down one and “Independent” down four compared with the results from May. The decline in the “Independent” rate means nothing as the “Independent” rate was inflated by contamination from the Legislative Council elections in the May sample and has now returned to a more normal level. As the Liberals have not gained anything despite four points of supposed support for “independent” returning to wherever it came from, this is actually a slightly (but not significantly) worse result for them than that in May. The Liberal vote has shown no signs of returning to the massive levels seen last year.

EMRS have thus far not released their usual electorate-by-electorate breakdowns. This is a pity as although those breakdowns are too small to mean much by themselves and are often greatly misinterpreted, they are useful both in checking for any regional bias in polling, and, more importantly, for trying to model how the results might translate into seats based on rolling averages. In my review of the May poll I noted that based on merged results of the February and May electorate samples, the Liberals would win at least 13 seats “if an election was held now” and might win up to three more. Two of the extra three were the fourth seat in Braddon and the third in Denison. The former was based largely on the strength of the February sample while the latter, while a realistic chance on the available numbers, was still pretty unlikely. On that basis, I believe that electorate-by-electorate breakdowns for the current polling, merged with the May sample, would point to 13-14 seats for the Liberals, 6-8 seats for Labor and 4-5 seats for the Greens.

No-one should be in serious doubt that the current poll would yield a majority result for the Liberals, barring very bad luck in the distribution of their vote. That the Liberals are tracking at just below 50% is no problem. As pointed out in comment 1 of a recent thread the idea that a party needs to receive more than half the primary vote in three or more electorates to win a majority (something that may occur with an overall vote of less than 50% anyway) is incorrect.

If one major party has more than a quota (i.e. 16.7+ points) lead over the other then they have excellent chances of an outright majority, with the most likely pathways to doing so being either three 3-1-1s or two 3-1-1s plus a 3-2-0 in Braddon where the Green vote is the lowest. Labor won 14 seats and very nearly 15 with a lead of 17.5 points in the 2006 election. Even with a lead of, say, 14 points, the chance of winning a majority is not that bad. Indeed, if the 1989 election had been held under the 25 seat system, the Liberals with a lead of 12.2 points over Labor would have gone extremely close to a 13th seat.

It might, however, be just a little concerning for the Liberals that their lead on the EMRS headline rate has shrunk from 33 points last August to 22 points now, the closest margin since last February. And given that that lead is probably a slight overestimate (EMRS has tended to understate Labor’s vote compared to the Liberals at election time), it would not have to close up that much more before some questions started to be asked. That is especially so since the federal picture may be inflating the margin and that picture will be different – one way or the other – by the next state election if the state government goes full term. The bar is very high for the Liberals because of proportional representation – even a vote that would be a 55-45 2PP result and a crushing win in many other states might not be enough for outright victory in Tasmania. It’s very likely the Labor/Green government will just continue should the Libs fall even one seat short, so nothing but outright victory is good enough for Hodgman’s team.

Of the media commentary generated by this poll, only Tony McCall’s comments are of any interest. McCall reckons not only that the trajectory of polling would concern the Liberals but also that “it’s very difficult, I think, in Opposition to hold your position 18 months out from an election.” My impression is that this is much more true of Federal elections than of state, especially if the government has been there too long. Looking at state polling over the past few decades, it is not easy to find examples of state governments anywhere that have come back from really horrible polling to retain office – at least, not where the polling was conducted by reputable major pollsters. There are, however, recent examples of struggling oppositions beating ageing governments out of seemingly nowhere (Vic, WA) or of oppositions holding massive leads all the way to the ballot box (Qld, NSW).

Most of the remaining commentary makes far too big a deal of the EMRS raw rates and the usual misunderstandings about EMRS’s high “undecided” rate, and the blame for this should go to the company for its prominent focus on these aspects even though the raw rates are not comparable to those used by other pollsters and claims about the “undecided” rate have been debunked so many times before. The company may say this is a sign of voters seeking leadership but voters have been doing so at fairly similar levels since at least 2004, excepting just before and after elections. The claim that a lack of change from poll to poll suggests “a settling of preferences” is unconvincing since it could easily be the case that there are drifts in different directions that are cancelling each other out, or that there was a change that was cancelled out by sample variation, or that there was no change in this period because not much that concerned the voters happened. Furthermore, when undecided voters “leaning” to a party are added, Labor is up four points (albeit from a terrible May result), which is actually significant. Far from the undecided voters “waiting to see what the major parties come up with in relation to significant policy announcements” or waiting for “a significant positive move”, it’s likely that the more hardcore undecideds will be switched off party politics until an election rolls around and they actually have to think about voting, while their softer brethren do have leanings but are just not that easily coaxed into stating them.

The ABC, Examiner, Mercury and Herald Sun reports are all pretty much the same in failing to cut through the dubious narrative provided by the pollster. Probably the only online media discussion shedding useful light on this particular poll is actually a report by Gayapolis News which reports some detailed results of a question on same-sex marriage that appears to have been conducted as part of the same polling. The poll was commissioned by Australian Marriage Equality, and overall claimed to show 61% in favour, 31% against and 8% undecided. I do not intend to comment directly on the reliability of this result itself as the poll is commissioned by an interest group and I have not seen the exact question. However the result is broadly consistent with national neutrally-conducted polls (ie those not commissioned by interest groups) and in general the survey design record of interest-group commissioned polling on this issue has been okay, despite the odd lapse (eg see comment 39 here).

What is helpful about the Gayapolis report of this polling is that it breaks down support for same-sex marriage by party – Greens voters 88% in favour, Labor 74%, Liberal 43% and undecided 59%, and in the process tells us a little about the views of “undecided” voters. It’s already clear from past elections that EMRS-undecided voters do not vote Green in any numbers, but what is useful here is that on the same-sex marriage issue, the sentiments of the “undecided” voters fall almost exactly between those of supporters of the two major parties. This finding certainly does not prove that undecided voters are, as a group, more or less evenly divided between Labor and Liberal ideologies (indeed, we do not even know if same-sex marriage is a big vote-driving issue for them) but it is at least completely consistent with it.

Some may wonder why my “interpretation” score does not distribute the points I remove from the Greens evenly to the other parties, and this is the reason for it: while EMRS distributes the undecided vote proportionally, resulting in more of it going to the Liberals than Labor when their primary vote is higher, I am not convinced that undecided voters are currently preferring the Liberals to Labor. This is why I have the gap at 18 points rather than 22.

Despite all this, the incumbent government continues to face the same problem it has since not long after the last election. While polling keeps showing that only the Liberals are anywhere near winning a majority, a major risk remains that undecided voters will swing behind the Liberals even if they do not agree with some Liberal policies, simply to prevent the Greens from getting the balance of power. Even if the Liberals entered the campaign with a lead of only ten points or so they would still be able to maintain that they and only they could win majority government. For this reason, while the gap has been closing over the past eighteen months, it has not on average been closing fast enough, and the Liberals still have an excellent chance of winning the next election outright.

First published: 2012-08-13 06:47 AM