A closer look at the parties’ polling fortunes this term state-by-state, in lieu of much to go on in the way of new polling over Easter.
Easter has meant that only the regular weekly pollsters have reported this week, which means Essential Research and Morgan. The latter polls weekly but reports fortnightly, which I deal with by dividing each fortnightly result into two data points, each with half the published sample size. Neither Essential nor Morgan is radically off beam, so this week’s movements involve a correction after last week’s Greens outlier from Nielsen. This is not to say that Nielsen’s Greens surge was measuring nothing at all, the 17% result perhaps having been partly a reflection of it being the poll most proximate to the WA Senate election. In fact, both of the new results this week find the Greens at their highest level since at least the last election, and probably a good while earlier. Their 11% rating in Essential may not appear too spectacular, but it comes from what is the worst polling series for them by some distance – indeed, the only one the BludgerTrack model does not deem to be biased in their favour. Nonetheless, their rating in BludgerTrack this week comes off 1.8% on last week’s Nielsen-driven peak.
The dividend from the Greens’ loss has been divided between other parties in such a way as to produce essentially no change on two-party preferred. However, state relativities have changed in such a way as to cost Labor three seats and its projected majority, illustrating once again the sensitivity of Queensland, where a 0.8% shift has made two seats’ worth of difference. The New South Wales result has also shifted 0.6% to the Coalition, moving a third seat back into their column. Another change worth noting is a 2.4% move to Labor in Tasmania, which is down to a methodological change – namely the inclusion, for Tasmania only, of the state-level two-party preferred results that Morgan has taken to publishing. I had not been putting this data to use thus far, as the BludgerTrack model runs off primary votes and the figures in question are presumably respondent-allocated preferences besides. However, the paucity of data for Tasmania is such that I’ve decided it’s worth my while to extract modelled primary votes from Morgan’s figures, imperfect though they may be. The change has not made any difference to the seat projection, this week at least.
Finally, I’ve amused myself by producing primary vote and two-party preferred trendlines for each of the five mainland states, which you can see below. These suggest that not too much has separated New South Wales and Victoria in the changes recorded over the current term, leaving aside their very different starting points. However, whereas the Coalition has had a very gentle upward trend this year in Victoria and perhaps also New South Wales, their decline looks to have resumed lately in Queensland. Last week I noted that six successive data points I was aware of had Labor ahead on two-party preferred in Queensland, including five which are in the model and a Morgan result which is not. That’s now extended to eight with the availability of two further data points this week. The other eye-catching result in the charts below is of course from Western Australia, which clearly shows the effects of the Senate election with respect to both the Greens and Palmer United. The current gap between Labor and the Greens is such that the latter could well win lower house seats at Labor’s expense on these numbers – not that I recommend holding my breath waiting for that to happen.