Highlights of week two

With polling day less than a fortnight away, a journalists’ thoughts turn to preference recommendations. Today we have Greg Roberts and Michael McKinnon reporting in The Australian that the Greens will not recommend a preference to Labor in "key marginal Gold Coast, Sunshine Coast and southern Brisbane seats". This is apparently significant because "a similar strategy by the Greens in the 1995 poll helped force the Goss Labor government from office". However, this is only true because the Goss government’s fortunes were decided by one seat, and the Greens’ preference decision cost Labor Mulgrave by a margin of 0.5 per cent. An article in Saturday’s Australian Financial Review by Peter Brent of Mumble covered the familiar topic of Greens voters’ reluctance to follow how-to-vote cards, but noted that the same did not appear to be true of Family First voters – according to a South Australian Parliamentary Library analysis, their votes split 65-35 in electorates where preferences were directed to Liberal, and 39-61 where directed to Labor. By Peter’s reckoning, that could have made the difference in four seats at the federal election. Consequently, Family First’s direction of preferences to Labor in Burdekin, Glass House, Mudgeeraba, Logan and Ipswich West is likely to be more consequential than the Greens’ decision, despite their smaller field of candidates. Family First preferences will be directed to the Coalition in Gympie, Ferny Grove, Sandgate, Toowoomba North, Toowoomba South, Cunningham and Lockyer, and to independent member Dorothy Pratt in Nanango. UPDATE: Antony Green notes there are special circumstances in South Australia which partly explain Family First’s capacity to influence preferences at the state election.

A few more Campaign Updates to be added to the election guide when I get my act together:

Currumbin (Liberal 3.2%) and Mudgeeraba (Labor 1.9%): Suzanne Lappeman of the Gold Coast Bulletin writes that while Labor has "all but written off" Dianne Reilly in Mudgeeraba, the party is increasingly hopeful its 2004 defeat in Currumbin will prove to have been an aberration born of the Tugun Bypass and the troubles of defeated member Merri Rose. Dennis Atkins of the Courier-Mail also refers to "wild talk" about Labor winning Currumbin, and concurs that Labor sources describe Mudgeeraba as "next to hopeless".

Noosa (Independent 8.7%) and Kawana (Labor 1.5%): Dennis Atkins also reports from a "Labor insider" that the Sunshine Coast looms as a "wipe-out" for Labor, with Noosa returning "quite strongly" to its traditional Liberal-voting ways. The only other Labor-held seat on the Sunshine Coast is Kawana, where the government has been having ongoing troubles over the location of a new hospital. Atkins notes that "deep antagonism to the Mary River dam" is also damaging Labor in the region.

Clayfield (Labor 1.2%) and Indooroopilly (Labor 2.1%): Dennis Atkins one more time: contrary to other reports elsewhere, "Labor strategists" quoted by Atkins are pessimistic about Clayfield – "the party hasn’t bothered to poll the electorate" – and "not confident" about Indooroopilly. Labor’s other Brisbane seats are believed to be safe.

Currumbin and Robina

A poll of 749 voters in yesterday’s Gold Coast Bulletin shows the Liberals struggling to hold two of their three Gold Coast seats: Currumbin, where Labor is reportedly becoming increasingly hopeful, and Robina, which is being vacated by ousted Liberal leader Bob Quinn. However, the online article says only that the Liberals hold primary vote leads of 3 per cent and 2 per cent respectively – presumably the print edition has more.

Fatal exposure

Polling agencies move in mysterious ways, as demonstrated by their insistence on letting respondents choose out of the Nationals and Liberals in Queensland state election surveys. Given that there will be no three-cornered contests at the election, voters will in fact have to make do with one or the other depending on where the live. Nevertheless, tomorrow’s Galaxy Research figures in the Courier-Mail turn up an interesting result: support for the Liberals has fallen over two weeks from 28 per cent to 24 per cent and now 22 per cent, while the Nationals have travelled from 15 per cent to 16 per cent to 18 per cent.

Why ever could that be? One answer suggests itself if we discount the notion that "all publicity is good publicity" and do some quick word searches on the Factiva news archive, which make clear that the current Liberal leader has assumed a higher profile than his predecessor had in 2004. The following table indicates the number of references to party leaders in the first 11 days of the two campaigns, both of which started four Tuesdays before polling day.

. 13-23/1/04 15-25/8/06
Peter Beattie 760 (58.1%) 837 (54.9%)
Lawrence Springborg 385 (29.4%) 387 (25.4%)
Bob Quinn/Bruce Flegg 163 (12.5%) 300 (19.7%)

In fairness to Flegg, it should be noted that the Nationals also gained ground on the Liberals after the 2004 election was announced, presumably because respondents became aware of which candidates were running in their own seats. As for Labor, Galaxy has their support over the past two weeks progressing from 42 per cent to 45 per cent to 47 per cent, and the two-party split reversing from 49-51 to a point where it is "edging closer" to the 55.5-44.5 result from 2004.

Hold the front page (late edition)

In today’s Courier-Mail, Jamie Walker and Steven Wardill provide the first intelligence to emerge about the parties’ marginal seat strategies. Labor is reportedly pouring resources into Hinchinbrook and Burdekin, two north Queensland seats held by the Nationals. Despite the retirement of sitting member Marc Rowell, the former seems very hard to credit – the margin last time was 10.9 per cent, and this time the Nationals are unburdened by competition from One Nation (12.6 per cent in 2004) and high-profile independent Andrew Lancini (21.4 per cent). Burdekin is being contested for Labor by Steve Rodgers, who was the member from 2001 to 2004, when it was won by Rosemary Menkens of the Nationals by 4.4 per cent. Here too there is less scope for conservative vote-splitting than last time, despite the presence of Family First: in 2004, Menkens faced former One Nation member Jeff Knuth attempting to win his old seat back as an independent, as well as an official One Nation candidate. Collectively they accounted for nearly a quarter of the vote. The article quotes Coalition campaign director Geoff Greene mocking Labor’s optimism, and concedes that Labor sources "with access to the party’s internal research" believe the Nationals are "likely to retain them on preferences". But that can’t be right in the case of Hinchinbrook, because the preferences could only come from the Greens.

The Courier-Mail headline "Confidence grows" is clever, because it doesn’t say whose. Labor’s confidence gets top billing (they are also "increasingly confident of retaining the tight Brisbane marginals of Clayfield and Indooroopilly"), but further on in the article we learn that "the Coalition has its eye on seats that were previously not even on its list of possibilities". These are Mulgrave and Kallangur, respectively held by 7.7 per cent and 13.7 per cent. Kallangur is being discussed because "residents in the area are outraged over factories they believe are toxic and a chemical fire they claim has affected the health of locals"; Mulgrave because newspapers like running photos of the Nationals’ 20-year-old candidate, Krista Dunford.

Hold the front page

After an incident-packed first week of the Queensland election campaign, week two has been agreeably uneventful. I shall have to fall back on two things that caught my attention in the papers recently.

Renee Viellaris from the Courier-Mail reports: "The Coalition will adopt the previously successful Labor strategy of ‘just vote one’ on its how-to-vote cards. The aim is – as Labor did in the 2001 and 2004 elections – to harness any protest vote against the State Government to its maximum effect". Unfortunately, "just vote one" was a genie that could only be let out of the bottle once. In order to be meaningful, the Coalition’s tactic will have to influence not its own supporters, but those who vote for minor parties and independents. Peter Beattie succeeded in doing this in 2001 and 2004 because it was the first time a party leader had advertised the legitimacy of the practice. The effect was to increase the rate of "plumped" voting across the board, from around 20 per cent of the total to more than 60 per cent, preventing scattered Liberal, Nationals, One Nation, City-Country Alliance and independent votes from consolidating behind the anti-Labor front-runner. Having the Coalition leaders reiterate the point will have little further effect. It will however have a substantial impact on their own supporters, thereby denying preferences to anti-Labor independents (UPDATE 25/8/06: Former Queensland Electoral Commissioner Bob Longland on ABC Radio this morning: "I heard Mr Greene, the campaign director for the Coalition, quoted in the press this week as saying that they were going to go on a just-vote-one campaign for the first time, but of course that’s not right, they did that strongly in 2004").

Labor’s natural source of preferences remains the Greens, and as this ballot paper study from the 2004 eection makes clear, these voters are by far the most inclined to number every box. It is unlikely that this is because they are unaware of the alternative, or because they need conservative party leaders to persuade them not to give preferences. The following table indicates the rate of plumped, partial preferential and full preferential voting by party support.

. PLUMP PART FULL
Labor 69.9 4.9 25.1
Nationals 60.2 5.5 34.4
Liberal 67.2 2.2 30.6
Greens 37.9 10.1 52.0
One Nation 43.4 12.1 44.5
Others 55.3 11.2 33.5
Total 62.2 6.1 31.7

Cath Hart of The Australian reported yesterday that the election was called just before a mechanism providing for an electoral redistribution was about to be triggered. The article says: "A redistribution is automatically triggered if 30 seats deviate by more then 10 per cent for two consecutive months under the current act. Queensland Electoral Commission insiders said growth trends within the state – which attracts 1500 new people every week – meant the redistribution would have been triggered ‘by the end of the year’ had the election not been called". Enrolment ranges from 16,428 in Mount Isa to 34,580 in Murrumba, which sounds like a statistic from the bad old days of life under Joh. One measure of malapportionment, the David-Eisenberg Index, is calculated by dividing the highest electorate enrolment by the lowest. The current figure of 2.1 is precisely the same as for the old zonal system as amended by Bjelke-Petersen in 1985, going on the quota figures for the smallest and largest zones (from which a degree of variation was allowed, presumably up to 10 per cent). It should be noted however that the current figure is not entirely due to population change and is influenced by the concession to rural vote weighting that was made when one-vote one-value was introduced in 1991. This allows for geographically large electorates (more than 100,000 square kilometres) to be "weighted" with a number of phantom voters equal to 2 per cent of their total area in square kilometres. If the six electorates affected by this are discounted, the David-Eisenberg index is 1.5.

Another measure of malapportionment is the Gini index, which takes into account the overall distribution and not just the two most extreme examples. This is currently at 6.5 per cent if the weighted electorates are included and 5.2 per cent if they are not, compared with 9.6 per cent under the zoning system of 1985, 8.1 per cent under that of 1977, 8.3 per cent from 1959 and 10.9 per cent from 1949 (in each case using quota figures for each zone rather than actual enrolment). For purposes of comparison, the Australian Electoral Commission provides this measure for every federal redistribution since federation (although you will have to multiply their figures by 100 – they appear to have confused the index with the coefficient).

To give a broad idea of the direction of population change, I have divided the state into 12 zones ranging in composition from five electorates to 13. "SD" refers to the standard deviation, which indicates the degre of variance within the sample. The idea in constructing the zones has been to gather as many similarly behaving, geographically joined electorates together as possible. Obviously there are imperfections here: booming Kurwongbah (34,374 voters) is an outlier in Northern Brisbane, Woodridge (24,496) weighs down Southern Brisbane, and Urban Hinterland is pretty much ruined by below-par Nanango (24,279). The party figures refer to seats won at the 2004 election.

. MEAN SD ALP NAT LIB IND
Sunshine Coast 31,540 1,798 3 1 1 1
Gold Coast 30,796 1,346 5 0 3 0
Northern Brisbane 30,281 3,205 7 0 0 0
Urban Hinterland 29,733 3,012 1 2 0 2
Southern Brisbane 28,796 2,522 13 0 0 0
Central Qld 27,834 1,515 3 2 0 2
Inner Brisbane 27,691 1,580 13 0 0 0
Western Brisbane 27,237 1,386 5 0 1 0
Cairns/Townsville 27,584 1,951 6 0 0 0
Inland Towns 26,629 1,499 4 1 0 0
Interior 25,164 1,995 1 6 0 0
Weighted 19,146 1,931 2 3 0 0

Correct weight

Nominations for the Queensland election closed today, and a list of candidates in ballot paper order is now available from the Electoral Commission of Queensland. The Poll Bludger election guide entries have been updated accordingly. The ballot paper draw is always an eagerly anticipated event as it determines the destination of the donkey vote, reckoned to be up to 1 per cent of the total. However, Queensland’s system of optional preferential voting has a major bearing on the operation of the donkey vote, which has no doubt been the subject of a study that has escaped my attention.

The motive behind most donkey voting is presumably to fulfil the ballot paper’s instructions with the least mental effort possible: in Queensland, the relevant instruction is "place the number one in the square opposite the candidate of your choice", whereas voters in the House of Representatives are advised to "number the boxes from 1 to X in the order of your choice". Therefore, while a federal election will usually deliver a benefit to whichever of the two major party candidates appears ahead of the other, the donkey vote at a Queensland election will only benefit a candidate at the top of the ballot paper. Accordingly, today’s draw can be said to have provided a fillip for Labor in the marginal seats of Aspley, Burleigh, Hervey Bay, Kawana, Redcliffe and Redlands; for the Nationals in Bundaberg, Burnett, Gaven, Keppel, Mulgrave and Nicklin; and for the Liberals in Broadwater, Cairns, Chatsworth, Mundingburra and Townsville. The non-major party sitting member whose re-election appears to be most in doubt, Elisa Roberts in Gympie, has also drawn top position.

Meanwhile, the following electorate entries have now been expanded as part of the progressive reupholstering of the Poll Bludger election guide: Gaven, Townsville, Pumicestone, Mundingburra, Redcliffe, Chatsworth and Gympie.