Thursday, September 19, 2013
EMRS: Liberals Cruising, Giddings Polls Bad Preferred Premier Score
(Note re Tas Senate: you can follow updates here. The count is not finished, the ABC Calculator is not reliable, and the seat is too close to call.)
EMRS September: Lib 52 ALP 28 Green 15 Ind 5 Other 1
Interpretation: Lib 51 ALP 30 Green 13 Ind/Others 6
Outcome (if election was held now): Liberal Majority Government (most likely 14-8-3, or 14-9-2)
The September EMRS poll of Tasmanian voting intentions has been released, based on interviews conducted earlier this week. The results show what every other Tasmanian poll by any company since the start of 2011 has shown, which is that the Liberal Party is on track for majority government. (The trendline for EMRS can be seen here.)
The headline figures are Liberal 52, ALP 28, Green 15, Ind 5 Other 1. They are not significantly different from the May EMRS results, with the Liberals a couple of points lower. EMRS results have a history of favouring the Greens and underestimating Labor, because of the way EMRS distributes the undecided vote. Past election results have shown that on polling day the Greens only get about their raw EMRS vote rate, and not their share of the huge EMRS undecided vote. However at the last election, the EMRS projection of the Liberal vote was very accurate.
The poll is quite similar to an August ReachTEL which showed the Liberals on 54% (after removal of undecided), Labor 25.8, Green 15.6 and Others 4.4. However, the federal election showed that ReachTEL polling overestimated the Liberal vote and underestimated the Labor vote at federal level, by about four and a half points. An analysis of the ReachTEL poll on the assumption that the same was true at state level (which it may not be) projected votes of 49.2% Liberal, 28.7% Labor, 13.4% Green and 8.7% Other.
I do believe that this poll underestimates the likely state vote for the Palmer United Party. Polls that do not explicitly name fourth-party options tend to underestimate them, and EMRS had this problem in the past with Andrew Wilkie especially, while ReachTEL (which has always named him) has captured his vote much more accurately. Nonetheless if there was a groundswell for a similar Palmer vote at state level at this stage I would expect the Others score to be more like three or four than one. So I suspect that there are voters who voted for PUP federally in the knowledge the election was a no-contest, but who will not do so at state level with majority government on the line, and that at the moment PUP would get a few percent at state level, not the 6-7 of the federal poll. But voting intention may change if there is a very expensive Palmer campaign as expected.
Because of this I have projected Ind/Others at 6 rather than what would otherwise be a reading of 4. All up I think what this poll is saying, and what the ReachTEL polls have been saying, is basically the same thing. The Liberals are on track for a majority of the popular vote or nearly so, that the Greens are facing a swing of several percent, and that there is not yet enough evidence that any fourth-party or independent candidate will win (although the potential is there.)
"Undecided" Rate Down
Behind the same old story on voting intention, there are two very notable things in this poll. The first is that the EMRS "soft undecided" rate (voters who say they are undecided when not pushed), a long-term laughingstock, has gone down from 30 points to 17. I don't know whether EMRS have improved their methods to attempt to drive this figure down, or whether it is a genuine reduction. Perhaps the 30% in May was an aberration given that the figure was 20% in February. But at least the new figure makes the baseline figures much less rubbery than in past EMRS polls. On the question comparable to other pollsters (voting intention after prodding of undecided respondents) the rate is down to 12%, which is still too high compared to most other pollsters, but compares well with the 19% in May (it was 13% in February.)
Bad Rating For Premier Giddings
Unfortunately Tasmanian pollsters do not poll true approval ratings and we have to make do with the much messier and less useful "preferred premier" indicator. Nonetheless this poll has seen a massive shift in Premier Lara Giddings' fortune on that score (such as it is) with her preferred-premier score crashing from 25% in May to 18%. This is Giddings' worst PP result as Premier. She also scored a pair of 19s in 2011, but that was against a backdrop of a lower vote for her party. If this figure came from a poll that showed the Labor vote lower than it should be, then I would treat it much more cautiously, but this is not the case. A seven point difference against the backdrop of no real change of voting intention is real; the question is what is it saying.
The rating has brought back memories of the infamous 17% preferred-premier score polled by former Premier Paul Lennon shortly before he resigned. It's my view he would have soon been rolled by David Bartlett had he not thrown in the towel. However, in Lennon's case there was an abrupt shift to None Of The Above, coming from Labor voters who wanted Bartlett in the job. In this case, there is no shift anywhere in particular; there are just small pickups for McKim, Hodgman, None of the Above and rounding. Giddings is lagging well behind her party in this poll, and this is probably an expression of general dissatisfaction with Labor even among many who intend supporting it. A similar thing has been seen in ReachTEL polling, with the government polling net performance ratings that are appalling even when adjusted for the apparent skew of the ReachTEL results.
The result may throw more heat into the contest in Franklin, where it is far from certain that Premier Giddings and apparent leadership aspirant David O'Byrne will both retain their seats. Whether there is any point in leadership rumblings with six months to go and the overall and seat outcomes unlikely to change much is doubtful.
The poll was taken largely before this weeks shenanigans in Parliament on the history of the Liberals' intention of forming minority government after the 2010 election. I believe the Liberals have been quite silly on this one as it is well known that they wanted it (provided they didn't have to deal with the Greens) and there is no point denying that or moving no-confidence motions about people's recollections of phone calls. But I also believe that this sort of parliamentary theatre has no impact on voting intentions. Julia Gillard was very good at it, and it did her no good whatsoever.
I will be modelling a new Tasmanian aggregate shortly, but I believe this poll is another very typical result and that the seat picture is likely to be similar to what I have extrapolated from others. That is, the Liberals with at least 13 seats and a good chance of 14 - winning three in Bass, Braddon, Lyons and with a good chance in Franklin; the Greens losing Braddon, probably Lyons and possibly Bass; and Labor not necessarily losing anything but on balance probably dropping a seat or two. I'd nominate 14-8-3 or 14-9-2 as the two most likely outcomes based on this poll, with 13-9-3 the next most likely.
Note also William Bowe has constructed an aggregate trend of ReachTEL/EMRS polling weighted off the federal election results here. The aggregate has the Liberal and Green votes lower than my federal-election model by a few points, and Others (not graphed) presumably much higher. The trajectory of the Greens on that chart looks quite alarming for them. Even with the Liberals at 46-47 as per the end point of that graph I don't doubt they would win majority government given what we know about the behaviour of Bass/Braddon/Lyons compared to the rest of the state.