As we move into what may very well be an election year, the BludgerTrack voting intention poll aggregate is finally cranked back into action. The model, which naturally picked a fairly comfortable Labor win on the eve of the 2019 election, is not quite what it used to be: there are dramatically fewer data points and less depth available in terms of breakdowns (pollsters have promised more rather than less transparency on this score, but thus far largely failed to deliver), which means there’s no point attempting state-level trends and seat projections as was done before.
Nonetheless, and for what it’s worth, you can now see voting intention trends on the sidebar, and in greater detail here. The lodestar for the model is Newspoll/YouGov: the results of the other pollsters, which really just means Essential Research and the occasional Morgan, are adjusted for bias as measured by the extent of their deviation from a Newspoll trend measure. As it happens though, these adjustments don’t amount to much: over time, none of the three pollsters has shown any particular tendency to favour any one party more than the others.
The trend shows a consistently close race through the current term, somewhat in defiance of media narratives, with Labor poking in front on two-party preferred in the wake of last summer’s bushfires but the Coalition maintaining a lead of around 51-49 for most of this year. This pattern is equally evident in the cruder but probably no less effective aggregate that Kevin Bonham knocked together for his comprehensive view of the year in polling. Part of this may be related to the fact that the new YouGov-administered Newspoll has maintained the pollster’s curious habit of being more consistent than the vagaries of random sampling should theoretically lead us to expect.
I’ve also gone the extra mile on the poll data archive, which now includes all of the expanded breakdown data that Newspoll is now providing in its quarterly aggregates (education, income, language and religion, on top of the traditional state, age and gender) and such two-party state breakdowns as Morgan has provided us, right down to two tiny-sample readings for Tasmania. The leadership ratings trends are still in business, though I’ve bumped them in favour of the voting intention trends on the sidebar.