Two new polls out today from Queensland, one being another of Newspoll’s composite marginal seat jobs, the other a statewide Nielsen survey of 1014 respondents. Taken together, the two continue a confounding pattern throughout this campaign of localised polling from Queensland painting a grimmer picture for Labor than polling conducted statewide. The Newspoll survey targets 800 respondents in seven of the state’s eight Labor-held seats Moreton (1.1%), Petrie (2.5%), Lilley (3.2%), Capricornia (3.7%), Blair (4.2%), Rankin (5.4%) and Oxley (5.8%) the odd man out being Kevin Rudd’s seat of Griffith (8.5%). The combined primary vote results are 38% for Labor (down from 42.4% at the 2010 election), 42% for the Coalition (up from 39.8%), 8% for the Greens (down from 11.0%) and 12% for others (up substantially from 6.8% hold that thought). On two-party preferred, the result is 51-49 in favour of the Coalition, a swing of 4.7%. Importantly though, this has been determined based on preference flows from the 2010 election. Hold that thought as well.
The Nielsen poll as published in the Fairfax papers comes with a headline two-party preferred figure of 53-47, which is at least superficially encouraging for Labor in that it suggests a swing of 2% from 2010. Unlike the Newspoll result, this comes from respondent-allocated rather than previous-election preferences (hold that thought still further). However, the real story the poll has to tell lies in the primary vote figures. Labor is at just 31%, down from 34.6% in 2010, but the Coalition is also down slightly, from 46.5% to 45%. The Greens are on 8%, down on 10.9% at the 2010 election but at the high end of what they’ve been getting generally in Queensland in recent times (perhaps reflecting an improving trend nationally which is perceptible on the BludgerTrack charts). However, the really interesting result is that the Palmer United Party is on 8%, putting into the shade Katter’s Australian Party on 4%.
This cannot dismissed as one freak result, as it has been corroborated by other polling. Roy Morgan has twice had occasion over the last week to trumpet this phenomenon going on beneath the surface of its others result. The first poll, published on Friday, had the Palmer United Party at 4% nationally and 6.5% in Queensland. The second, published yesterday, maintained the 4% national result while finding the Queensland figure up to 7.5%. I’m advised that Essential Research also had the party at 4% nationally in its polling this week and at 9% in Queensland, after it barely registered in previous weeks. In fact, the three sets of Queensland polling I have seen over the past few days have all turned in remarkably similar results for Labor, Coalition, Greens and others alike.
A clearer picture emerges if the totality of polling from Queensland is plotted out since the return of Kevin Rudd. The chart below maps out the trend from 37 such polls from seven different pollsters, with the usual BludgerTrack accuracy weightings and bias adjustments applied. Black represents the combined others vote.
The starting point is a landslip in Labor’s favour after Gillard was deposed, which appeared to consolidate for a fortnight before entering a long and steady slide. Then came the announcement of the election date at the start of August and a two-week period where Queensland appeared to buck the national trend of the time by moving to Labor. This may very well have been a dividend from the recruitment of Peter Beattie, however much media reportage and individual seat polls might have suggested that there wasn’t one.
A new phase then appeared to begin a fortnight ago with the sharp rise of the others vote. This has coincided with an onslaught of television advertising from Clive Palmer which has seemed almost to rival that of the major parties. Whereas Palmer’s earlier advertising looked like it belonged on Vine rather than network television, his current efforts appear rehearsed and properly thought out perhaps even market-researched. Most importantly, the substance of their message tax cuts which pay for themselves and pension schemes that boost the economy by $70 billion may well be striking a chord in offering voters the ever more scarce political commodity of vision, hallucinogenic though it may be in this particular case.
The other point to be noted about the surge in the others vote over the past fortnight is that it looks to be coming more at Labor’s expense than the Coalition’s. For one thing, this has significant implications for the party’s prospects of actually converting votes into seats. Mark Kenny of Fairfax’s take on the Nielsen result is that while it is almost certain Mr Palmer’s party will not win a seat in the House of Representatives, it is in with a chance of gaining a spot in the Senate. However, I’m not so sure about this on either count.
Clive Palmer himself is running in the smartly chosen Sunshine Coast seat of Fairfax, where the retirement of Alex Somlyay relieves him of the burden of having to take on a sitting member. The first task facing Palmer is to outpoll Labor, who scored 27.3% in 2010. Gouging votes directly at their expense will make that task a lot easier, as presumably will the fact the Greens (who polled a weighty 18.0% last time) are directing their preferences to him. Palmer’s next hurdle (inappropriate as athletic metaphors might be in his case) would be to overcome Liberal National Party candidate Ted O’Brien, which might not be so easy given Alex Somlyay’s 49.5% vote in 2010. Some credible seat-level polling from Fairfax would be very interesting to see. As for the Senate, lead candidate Glenn Lazarus faces the complication that James Blundell of Katter’s Australian Party has done better out of preferences, standing to directly receive (among other things) Labor’s surplus after the election of its second candidate.
The other point to be made regarding a movement from Labor to the Palmer United Party relates to the issue of deriving two-party preferred results from primary votes in opinion polls. This is always a slightly vexed question, as for most voters the act of vote choice runs no deeper than simply deciding who to vote for, be it a party or its leader. If that choice is for a minor party, the question of preference allocation secondary though it may be for the voter concerned is the thing that really matters with respect to determining the result. Since the decision is often driven by a how-to-vote card the voter does not see until they arrive at the polling booth, and is in many cases entirely arbitrary, there is limited value in an opinion pollster asking the voter what they propose to do.
For this reason, it has become standard practice over the past decade for pollsters to instead allocate minor party preferences according to how they flowed at the previous election. Only Morgan persists in favouring respondent allocation, with Nielsen conducting both measures while normally using the previous election preferences for its top-line results. Not coincidentally, the primacy of this method has emerged over a period in which the minor party landscape has remained fairly stable, with the dominant Greens being supplemented by a shifting aggregation of smaller concerns, most of them being right-wing in one way or another. However, it was always clear that the utility of the method would be undermined if substantial new minor parties emerged, particularly on the right. For example, the result of the 1996 election would have offered no guidance in allocating votes for Pauline Hanson’s One Nation when it exploded on to the scene a year later.
So it is with the Palmer United Party, at least so far as Queensland is concerned. It might have been anticipated that the party’s conservative provenance would have caused its preferences to behave much as other right-wing minor parties to emerge out of Queensland have done over the years, but the Nielsen poll throws that into doubt by finding that 62% of Palmer United Party voters (together with 55% of Katter’s Australian Party voters) intend to give their preference to Labor. It should be borne in mind here that these sub-samples are extremely small, and consequently have double-digit error margins. Eighty-six per cent of Greens voters said they would preference Labor, which is well above what’s plausible. Even so, it’s perhaps telling that the most recent national Nielsen poll, published the weekend before last, had the Coalition’s lead in Queensland at 55-45 on previous election preferences, but only 52-48 on respondent-allocated preferences an enormous difference as these things go.
Taken together with the trends observable in the primary vote chart above, it would appear that the last fortnight has seen Labor lose votes in Queensland to the Palmer United Party, and that this pool of voters contains a much larger proportion of Labor identifiers than the non-Greens minor party vote in 2010. So while the recent rise of the Palmer United Party might not be good news for Labor in absolute terms, it may cause two-party preferred projections based on the normal pattern of minor party vote behaviour to be skewed against them. This certainly applies to the BludgerTrack model in its present form, for which I might look at adding a Queensland-specific fix (with the qualification that anything I come up with will of necessity be somewhat arbitrary).
UPDATE: AMR Research has published its third online poll of federal voting intention, conducted between Friday and Monday from a sample of 1101, and it has Labor at 34%, the Coalition at 44% and the Greens at 10%.