The BludgerTrack poll aggregate continues to trend the Coalition’s way, to the extent that it reaches two milestones this week: parity with Labor on two-party preferred, and an absolute majority on the seat projection, albeit by the barest of margins. Three new polls were added to the national figures, those being Galaxy, the regular weekly Essential Research, and the fortnightly Morgan (fortnightly in the sense of publication, although the poll is conducted on a weekly basis). Also out this week was the Newspoll quarterly aggregates, which have been factored into the state breakdowns, along with the regular state breakdowns from Morgan (published) and Essential (unpublished). The combined effect is to add seat each to the Coalition tally in New South Wales, Western Australia and South Australia, while removing one in Victoria and Tasmania.
The quarterly Newspoll is a big deal for BludgerTrack, which is never better serviced for state data than it is immediately after being fed with three months’ worth of state-level Newspoll results. To this end, later today I will get around to publishing my own detailed quarterly state breakdowns for BludgerTrack, the previous instalment of which can be seen here.
BludgerTrack is still in the position of being slightly more favourable to the Coalition than any single published poll result, due to a variety of factors. Perhaps this could be best explained if I run through each of the pollsters:
Nielsen of course closed up shop a few months ago, which was significant in that BludgerTrack deemed it to be the most Coalition-friendly pollster, and the only one which adjusted for any substantial bias to that effect. Now that it’s gone, the model has a clear tendency to skew to the right of what a straight polling average would tell you.
Newspoll is rated as neutral by the model, but it hasn’t reported for a fortnight. When it did report, it gave Labor a 51-49 lead when the primary vote numbers looked a lot more like 50-50. It’s the primary votes that BludgerTrack goes off, so this was a 50-50 poll as far as the model was concerned. Clearly Labor got rounded up in the Newspoll result it follows that they also got rounded down in BludgerTrack.
Galaxy is taken very seriously by BludgerTrack, and receives next to no bias adjustment at all. This week it gave Labor a lead of 51-49, although putting its rounded primary votes into the model produces a result of 50.6-49.4 going off 2013 preferences (as BludgerTrack does). If not for this poll, the Coalition would have moved into the lead.
ReachTELs last poll a fortnight ago had Labor leading 51-49, and BludgerTrack adjusts this pollster slightly in favour of the Coalition.
Morgan is reckoned to have the biggest bias in the game, that being in favour of Labor. Its result on respondent-allocated preferences this week was 51.5-48.5 in favour of Labor, but the more telling point so far as BludgerTrack is concerned is that it was the Coalition’s best result since February.
Essential is noted for being slow to respond to changes, and for this reason, BludgerTrack treats its bias in a unique way, by dynamically adjusting it according to how its deviates from the model over time. Since it’s stayed stuck with Labor on the cusp of leading 52-48 or 53-47, while the other pollsters have moved to the Coalition, a Labor bias adjustment is increasingly being factored into its results.
The other development in BludgerTrack this week is that Morgan published a set of phone poll numbers on leadership ratings, and they were relatively very rosy for Tony Abbott, who wasn’t too far off parity on net approval and had a pretty solid lead on preferred prime minister. This has a pretty sharp effect on the BludgerTrack leadership ratings, which aren’t exactly spoiled for data and are always pretty sensitive to the most recent result, even if the poll in question was from a rather small sample, as was the case here.
UPDATE: As promised, here are the detailed state-level breakdowns featuring primary vote numbers and charts tracking the progress of the primary and two-party votes in each state. Crikey subscribers may enjoy my analysis of these results in today’s email, assuming it gets published.
I also promised two weeks ago that I was going to start tracking betting odds in these mid-week BludgerTrack posts, then forgot about it last week. Now that I’ve remembered again, I can inform you that there has been movement to the Coalition over the part fortnight in Centrebet’s federal election odds, with the Coalition in from $1.50 to $1.45 and Labor out from $2.55 to $2.70. Centrebet’s price on Campbell Newman being re-elected in Queensland has also shortened from $1.36 to $1.28, with Labor out from $3.15 to $3.65. There has been a very slight move to Labor for the Victorian election, with Labor in from $1.23 to $1.22 and the Coalition out from $4.00 and $4.10 which sounds a bit generous to Labor for mine. The Betfair market evidently thinks so, as it has the Coalition in from $4.10 to $3.40 and Labor out from $1.48 to $1.59.