BludgerTrack makes a fairly solid move to the Coalition this week on the back of relatively strong results for them in Newspoll and ReachTEL, to the extent that they are now ahead of Labor on the national seat projection, without going so far as to make it to a majority. Labor retains the lead on two-party preferred, but the model grants the Coalition a natural advantage in seat allocation because the decisive marginal seats will be defended by its first-term members. The change returns the two-party vote to where it was three weeks ago, before a 1.2% spike to Labor the following week. However, Labor has gone two seats backwards on the seat projection since then, because of changes in the way the votes are distributed between the states. The Coalition primary vote gain comes off the total for the Greens, which had experienced a spike over the previous fortnight, while Labor’s is essentially unchanged. Three new sets of state-level data were available to the model out of the four polls which published this week, which have caused Labor to drop two seats in Queensland and Tasmania, and one in New South Wales, Victoria, Western Australia and South Australia.
Some will be asking how Labor’s two-party vote comes to be at 50.9% when no published result has put it below 51%, for which much of the explanation lies in Newspoll’s rounded two-party numbers this week. As Kevin Bonham observes, the 51-49 result had poll watchers scratching their heads, as a crude application of 2013 preference flows to the published primary votes (Labor 34%, Coalition 41% and Greens 11% and 14% others) puts the Coalition slightly above 50-50. I don’t doubt that Newspoll has done its rounding properly the result could be explained by primary vote rounding, minor party vote shares and the poll’s internal distribution of state results but there can be little doubt that Labor was rounded upwards. Then there was Thursday’s 51-49 result from ReachTEL, a large sample poll with a good track record that the model takes seriously, but which is corrected for a slight Labor bias. The model grants Essential and Morgan together about as much weight as a single Newspoll or ReachTEL, and they had much the same results as each other after the fairly considerable Labor bias adjustment for Morgan. So the aggregate this week can roughly be seen as combining a 50-50, a 51-49 and a 52-48.
Newspoll provided a new set of results for the leadership ratings, which have unfortunately come to be dominated by the pollster since Nielsen dropped out of the game. As such, this week’s moves reflect Tony Abbott’s stronger performance in Newspoll, suggesting a second shift in his favour to supplement the one which occurred after MH17. He also widens his lead as preferred prime minister, although Bill Shorten’s net approval rating remains stable and fairly respectable, and solidly higher than Abbott’s.