First up, please note that immediately below this post is a new entry on developments in Queensland, which include one and possibly two looming state by-elections. With that out of the way, a brief collection of polling and preselection news:
• In the wake of a contentious poll on the subject for the Institute of Public Affairs, The West Australian has published a WA-only survey on attitudes towards celebrating Australia Day on January 26, conducted by Perth market research firm Painted Dog Research. This found 65% support for maintaining the current date with 21% opposed, breaking down to 55-26 among those aged 18 to 39, 67-20 among those 40 to 59, and 78-14 among those 60 and over. Although substantial, the headline figure is narrower than the 71-11 margin recorded by the Dynata poll for the IPA, which primed respondents with two leading questions on being proud of Australia. This poll was conducted from 842 respondents drawn from an online panel, with no field work dates provided.
• Cory Bernardi has followed through on his announcement last year that he would resign to the Senate, which means his South Australian seat returns to a nominee of the Liberal Party, for which he won the seat from the top of the ticket at the 2016 double dissolution. The Australian ($) reports the matter will be decided on February 1, from a field including Morry Bailes, managing partner at Tindall Gask Bentley Lawyers and former president of the Law Council of Australia; state upper house MP Andrew McLachlan; and Michael van Dissel, former state party treasurer. Bailes has the support of conservatives including Mathias Cormann and South Australian federal MPs Tony Pasin and Nicolle Flint, which is presumably good to have.
• Heavy duty psephological pundit Mark the Ballot examines the deficiencies of polling before the May federal election, to the extent that the industry’s lack of transparency makes the matter knowable. The thrust of the analysis is that the pollsters’ models were “not complex enough to adequately overcome the sampling frame problems”, the latter reflecting the fact that surveying methods in the modern age cannot plausibly claim to produce genuinely random samples of the voting population. As well as the models by which the pollsters convert their data into vote shares, this lack of “complexity” may equally arise from herding, the unacknowledged use of smoothing techniques such as rolling averages, and over-use of the same respondents in online panels.