The Poll Bludger

The Adelaide Advertiser today brings us a survey from the Coalition’s fourth most marginal electorate, Adelaide, taken from a substantial sample of 419. Conducted the day after the announcement of Labor’s tax policy, the poll shows Liberal member Trish Worth leading Labor’s Kate Ellis 42 per cent to 35 on the primary vote and by 51 to 49 on two-party preferred.

Fending off a thorny question about the role One Nation preferences played in putting Kerry Nettle into the Senate at the 2001 election, Bob Brown has recycled the canard that Bob Menzies won the 1961 election on Communist Party preferences. This is a great story for those unready to forgive Pig Iron Bob for his crimes against the working class, but the truth is more complicated. The election in question saw Menzies returned with a one-seat majority, the Coalition’s narrowest victory being Jim Killen’s 130-vote win in the Brisbane seat of Moreton which is still with us today. Electoral historian Adam Carr, curator of the indispensible Psephos election results archive, explains it thus:

There are several myths about this famous contest. The first is that Menzies sent a message saying "Killen, you’re magnificent" (he didn’t). The second is that Killen, a fierce anti-Communist, received Communist preferences (he did not: he did benefit from the small drift of Julius’s preferences, but this was cancelled out by the bigger drift of QLP preferences to the ALP). The third is that the ALP would have won the election if it had won Moreton (it would not have: the state of the parties would then have been 61-all. There would probably have been a minority Menzies government followed by a fresh election).

The element of truth within the story is that Labor would indeed have won the seat if about 70 fewer of the 676 Communist Party voters had put Liberal ahead of Labor, but as only about 100 voters had actually done so this would have been a rather unlikely outcome. There was indeed a view among some Marxists that conservative governments were preferable as they would hasten the revolution, but Communist Party preferences were not "directed" to the Liberals on this occasion and the majority of them did in fact go to Labor. And as Carr says, even if the result had gone the other way Labor could not be said to have won the election.

This time a fortnight ago, just before the announcement of the election, visitors to the Roy Morgan website were met with the confident declaration "ALP Would Easily Win Election – Not a Conducive Time for L-NP to Call the Election". The organisation obviously expected its advice would be heeded as it went about its normal business that weekend, conducting the first half of its regular fortnightly face-to-face polling with the other half to be completed the following weekend. The results it released yesterday have thus arrived in the manner of a visitor from another planet, including many responses taken before the election was called and none from after Labor’s tax announcement on Wednesday. By any measure its findings are remarkable – Labor has 56 per cent of the two-party preferred vote despite a reasonably modest 43 per cent on primary, no doubt assisted by preferences from the Greens who have broken through the 10 per cent barrier, perhaps for the first time ever. The Coalition is on 38.5 per cent, its lowest level since early February. Gary Morgan presented the findings fairly cleverly, saying "until the bomb in Jakarta the ALP looked set to win the federal election. The bomb in Jakarta has essentially reset the clock". It’s a plausible argument that distances him from findings which run counter to those of every other polling agency.

Psephologists would much prefer it if election campaigns could be held in air-tight containers in which our theorems could operate without disturbance from the vagaries of reality. Certainly no one has been blind to the possibility that terrorism at home or abroad could make an ugly intrusion into the Australian election, but campaigns have a way of obscuring the long-term view and an event such as yesterday’s Jakarta bombing comes as a shock to those of us who had been confidently plotting the future trajectory of Labor’s tax policy bounce. The impact on Labor’s prospects is difficult to assess as there is no obvious precedent for the attack, at least from an Australian perspective. With no serious Australian casualties it will obviously be absorbed more quickly than Bali, whereas the Marriott Hotel bombing provoked less agitation as it did not strike an identifiably Australian target. Nevertheless, comparisons may be instructive. The Marriott Hotel incident of 5 August 2003 did not register in Newspoll, not too surprisingly, but the Government received a bounce following the Bali bombing of 12 October 2002 that appeared to sustain it through to the following February. However, it might be said that such an effect scaled down in proportion to yesterday’s less wounding atrocity wouldn’t last the remaining four weeks of the campign.

A variable on this occasion is that the Iraq war has taken place since the Bali bombing, and some have predicted the Government will be blamed for having made Australia a terrorist target. Others argue that the Coalition will benefit from their perceived toughness and the advantages that accrue to the incumbent in dangerous times, and this appears to carry more weight. No doubt there is a large constituency that is ready to link terrorism to the iniquities of western leaders, but it contains few swinging voters. The more important factor on this occasion is that the front pages have been cleared of election coverage just as Labor’s campaign was building a head of steam.

After taking a pounding in the polls at the end of the first week, Labor can expect the ball to bounce back their way starting tomorrow with Roy Morgan (I’m assuming their polls will now be weekly). While the family benefits issue remains a source of confusion, Mark Latham has unquestionably "cut through" with his message that workers on less than $52,000 will get hamburgers and milkshakes that will not be available to them under the Coalition. No doubt the policy is well targeted socially, but that might not make it well targeted electorally. If Labor is the party of the low income earner, won’t the dividend be wasted in safe seats?

This Australian Parliamentary Library spreadsheet, ranking electorates in order of taxable income per taxable individual, suggests otherwise. It is true that the average figure is less than $52,000 in all but the top eight seats, all of which are held securely for the Liberal Party unless you don’t count Wentworth. But the Coalition also monopolises the bottom seven, four of which are National Party seats. In reality the urban/rural divide is a more significant cleavage than income ranking, with the Coalition tending to dominate in regional areas where both incomes and the cost of living are lower.

The figures for the 24 most marginal Coalition-held seats are listed in the following table, with their ranking out of the 150 seats in brackets:

Solomon (0.1%)	38113 (43)		Hinkler (2.3%)	32599 (127)
Dobell (0.4%)	35772 (71)		Moreton (2.5%)	36581 (59)
Canning (0.4%)	34218 (99)		Longman (2.5%)	30859 (145)
Adelaide (0.6%)	43120 (21)		Gippsland (2.6%)	33577 (111)
Hindmarsh (1.1%)	36481 (61)		Page (2.8%)	30402 (147)
Parramatta (1.2%)	38886 (38)		McMillan (2.9%)	32734 (124)
Paterson (1.5%)	34827 (89)		Bowman (3.1%)	35502 (76)
Herbert (1.5%)	35498 (77)		Petrie (3.5%)	34667 (93)
Deakin (1.6%)	38319 (42)		La Trobe (3.7%)	36792 (54)
Eden-Monaro (1.7%)	34981 (86)		Makin (3.8%)	33014 (120)
Richmond (1.7%)	30772 (146)		Kalgoorlie (4.4%)	41741 (25)
McEwen (2.2%)	36192 (66)		Cowper (4.8%)	29895 (149)

Broadly speaking the very most marginal seats, most of them in cities or suburbs, are disappointingly well off from Labor’s perspective. But it’s a different story in the 2 to 3 per cent range, where there is a clump of low-income non-metropolitan electorates including three held by the National Party. They include Richmond and Page on the New South Wales north coast; Gippsland and its neighbour McMillan (held by Labor but notionally Coalition after the redistribution) in eastern Victoria; and in all important Queensland, Longman on the Sunshine Coast and the Gladstone/Bundaberg seat of Hinkler. It’s also in the former category of electorate where the Coalition’s campaigning on interest rates is likely to have its greatest impact. Unless the Coalition can find another way to sway wage earners in regional electorates, it could be here that they face the greater challenge.

The Poll Bludger, along with a number of others (Peter Coleman and Alan Jones spring to mind), could yet be made to eat his words on the folly of Peter King’s plan to run as an independent in Wentworth if Labor polling leaked to The Australian yesterday is accurate. The results were Malcolm Turnbull 30 per cent, Labor’s David Patch 27 per cent, Peter King 25 per cent and the Greens 10 per cent, a substantially better result for King than in May’s Sun-Herald Taverner poll. However King’s challenge is still to get his nose in front of Labor so he can ride over Turnbull on their preferences, which is why he is so furiously courting the Greens. However, all the evidence suggests that most Greens voters make up their own mind where to deliver their preferences and at least 70 per cent of them favour Labor over the Coalition. Where they will rank an independent Peter King in this scheme of things is an open question, but it is likely the majority will consider him a Liberal. So even on the figures quoted, King will need to have generated a substantial boost from Friday’s announcement and will have to maintain that through four weeks of fierce Liberal campaigning. Expect scare campaigning over a hung parliament to feature prominently.

Another point of interest here is the temptation for Labor to run dead and encourage "tactical voting" by their supporters, a concept familiar to voters in Britain but one rendered largely irrelevant in Australia by preferential voting. Where British Labour and Liberal Democrats supporters would often back each others candidates on the basis of who had a realistic chance of defeating the Conservatives, Labor supporters in Wentworth could face a choice between Peter King overcoming the Labor candidate and riding home on his preferences, or Labor holding firm in second place and Peter King’s preferences sealing victory for Turnbull. But to do that they would have to let go of the idea that David Patch might actually win, and with all the rancour consuming the conservative camp, he too might have enjoyed a boost of sorts from Peter King’s decision to run.

UPDATE: Malcolm Farr reported in yesterday’s Daily Telegraph (apparently not available online) that "party figures determined to see Mr Turnbull lose have urged Labor supporters to vote for Peter King, the Liberal turned independent … Mr Patch is determined to fight the poll rather than take a dive but the electorate might cotton on to the plan".

UPDATE (9/9/04): It did occur to the Poll Bludger that Labor might have suspect motives in conducting and releasing polling from a seat that would not normally be on its hit list of marginals. Prominent blog-botherer Homer Paxton has dropped a line to add support to this idea: "why on earth would the ALP waste money on polling in Wentworth? I have a fair idea of how much it would cost and they wouldn’t waste the money. I suspect this is a plant for the Libs to pour more money and more importantly people into Wentworth and less into some close-by marginals … the ALP very rarely polls only one electorate".

For crying out loud, enough freaking opinion polls already. An ACNielsen poll appears in today’s Fairfax papers (with a sample of 1415, larger than Newspoll’s) and it’s gratifying to note that its results are all but identical to those of Newspoll and Galaxy – Coalition 46 per cent, Labor 40 per cent and an even split on two-party preferred. There is also a state-by-state breakdown which I will think about a bit harder later on.

The polls conducted for the Daily Telegraph by Galaxy Research over the past two months have consistently painted an unusually sunny picture for the Coalition, and today’s is no exception. However it’s interesting how it mirrors Newspoll, putting the Coalition one point higher on 46 per cent and Labor one point lower on 39. Somehow this two-point primary vote difference transfers to four on two-party preferred, which is to say that Galaxy has the Coalition ahead 52-48.