The latest weekly BludgerTrack update has been added to the sidebar, adding results from Newspoll, Galaxy and Essential Research. I have replaced my ad hoc rolling average calculation with LOESS, the polling wonk’s smoothing method of choice. As well as making my graphs look prettier, this makes for smoother trendlines and should reduce volatility from one week to the next. I’ll be applying some further methodological tinkering next week, including updating the Newspoll bias measures to account for the Western Australian election result on which more below.
First though, it should be noted that the Financial Review has published results from a JWS Research automated phone poll of 4070 respondents in 54 marginal seats, conducted on Tuesday and Wednesday. This records an aggregate swing of Labor of 9.3%, compared with a quite bad enough result of 4.8% when the exercise was last conducted in mid-January. In the aftermath of last week’s leadership chaos, it may well be significant that the shift was heavily concentrated in Queensland.
A general meeting of the ACT Liberals has voted down a motion to overturn Senator Gary Humphries’ preselection defeat at the hands of former ACT Opposition Leader Zed Seselja by a margin of 168 to 138. There had been compaints from Humphries supporters that many party members had been wrongly excluded from the February preselection vote, which Seselja won 114 to 84.
The plot has thickened in the preselection to choose a successor to Nicola Roxon in Gellibrand, with three credibly rated candidates in the field. Unidentified sources quoted by John Ferguson in The Australian suggest opposition to Tim Watts, Telstra executive and former staffer to Stephen Conroy, amounts to a test of Conroy’s influence, which is said to be starting to wane. Josh Gordon of The Age reports a 2008 factional realignment reserved the seat for the Shorten-Conroy Right sub-faction, but Roxon has entered the fray by sending a letter to local party members urging for them to support her former adviser Katie Hall. There is another prospective candidate in Kimberley Kitching, a former Melbourne councillor, current acting general manager tasked with restoring order to Health Services Union No. 1 branch, and wife of VexNews provocateur Andrew Landeryou. Gordon reports the Turkish community is emerging as a source of support for Kitching, as Conroy has roused its opposition by refusing to offer up the state seat of Footscray to the so-called Turkish bloc.
The preselection to replace Richard Torbay as Nationals candidate for New England will be held in Tamworth on April 13. Barnaby Joyce will certainly be a starter, but there have been suggestions he will or should face opposition from Nationals Farmers Federation president Alexander “Jock” Laurie, including from Calare MP John Cobb.
With the state election out of the way, WA Labor is now proceeding with federal election preselection processes, chief among which is determining its Senate election ticket and filling the casual vacancy caused by the retirement of Chris Evans. Suggestions Labor might be reduced to one Senate seat in Western Australia mean that more than prestige is at stake in ordering the top two positions on the election ticket. The two incumbents are Louise Pratt of the AMWU Left and Mark Bishop of the SDA Right, with the latter generally expected to be deposed by Joe Bullock, who succeeded him as the SDA’s state secretary. Other nominees are former state Bassendean MP Martin Whitely, a critic of party preselection processes generally and the Joe Bullock ascendancy in particular; Brett Treby, a Wanneroo councillor who ran for the state seat of Wanneroo; John Welch, secretary of the Western Australian Prison Officers’ Union; Kelly Shay, assistant state secretary of United Voice; and Sue Lines, assistant national secretary of United Voice.
Sue Lines is getting more attention for her parallel nomination to succeed Chris Evans, a position claimed by the powerful United Voice sub-faction of the Left. Lines is rated as one of two-front runners along with Sharryn Jackson, who won the lower house seat of Hasluck in 2001 and 2007 and lost it in 2004 and 2010. The aforementioned Martin Whitely, John Welch and Kelly Shea have also nominated for the Evans vacancy, together with Linda Morich and Ashburton councillor Peter Foster. Both matters are scheduled to be determined at a state executive meeting on April 15.
Finally, a review of Newspoll’s performance at the WA election and some related musings on the two-party preferred measure. The scorecard for Newspoll reads thus:
2PP ALP L-NP GRN Result 57.5? 33.1 53.2 8.4 Newspoll 59.5 32 54 8 Difference +2.0? -1.1 +0.8 -0.4
This is a very sound result, with all primary votes well within the margin of error. However, it’s worth noting that it’s the eighth pre-election Newspoll out of the last nine to shoot low on the Labor primary vote, and the seventh to do so by more than a percentage point remembering that in most cases two-party preferred ended up near the mark because support for the Greens had been overstated (although this hasn’t been evident on the two most recent occasions).
Keeping in mind that my two-party result is based on incomplete data, Newspoll’s two-party preferred result proved less accurate than the primary votes, which is largely down to an issue with two-party preferred calculations involving the Liberals and Nationals. Newspoll looks to have followed the usual method of simply combining the two and then distributing minor party preferences between the two major parties, but this doesn’t account for the fact that in three-cornered contests some Liberal and Nationals votes end up in the Labor pile when the contest is boiled down to Labor-versus-Liberal or Labor-versus-Nationals for two-party purposes. This is of little concern at federal elections, where competitive Nationals-versus-Liberal contests are uncommon. However, the WA election had no fewer than 17 three-cornered contests out of 59 seats, with both Liberal and Nationals polling strongly in most cases.
It should be noted that it makes a difference whether a Labor-versus-Liberal or Labor-versus-Nationals count is used, because Nationals voters are more likely to preference against their coalition partners than Liberal voters. This seems to be especially pronounced in those regional corners of the state where the Nationals have won a new constituency of former Labor voters who are still not keen on the Liberals. The precise result of the final two-party result will thus be influenced by the WAEC’s ruling on whether it conducts Labor-versus-Nationals or Labor-versus-Liberal counts in the eight seats where such counts remain to be published. In 2008 they went Labor-versus-Liberal in each case, which meant the Nationals were only used in seats where the final count had been between them and Labor (I believe this only applied to Pilbara, where only 7517 formal votes were cast, and that this will again be the case this time). The results were thus more favourable to Labor than they might have been if the WAEC had employed an alternative rationale, such as conducting Nationals-versus-Labor counts where the seat was won by the Nationals, as was the case in six of the eight seats with counts still outstanding.