Exit polls are not precise, but they can be worrying

Dear ,

The Newscorp publications are running the results of exit polls they have done on early voting. It is a mixed bag with some wins and some losses, although they find that the Labor Party is up by 4.6% on primaries since the 2022 election, and the LNP is up by 2.8%. They have the minors, including the Greens, going backwards.

This doesn't seem right, as everything else we are seeing suggests a shift to the minors, at the same time as maybe a mild improvement in the ALP vote.

I was particularly interested in the two seats we are targeting - Ryan and Brisbane. Thanks to those who have donated so far. If you wish to put some money towards the campaign please click here.

You can see the relevant part of the Newscorp table above (click on it for an enlargement). The hardest thing about being a candidate is when polling data comes out because you get a false sense of a fixed future, even though it is still very fluid. I think both LNP candidates would be feeling queasy at these figures. They show who their most likely major opponent is, but apart from that, all they really say is that it is close.

First thing to know is that the exit polls conducted on this basis in the state election overstated the LNP vote, and probably made them overconfident. They are not a random sample of the whole electorate, but of those people who have chosen to vote early in selected marginal seats and they may not even be a properly randomised set of those people.

Second thing is that while around Australia they asked 4,000 people their opinion, they only asked 200 people in each electorate. What that means is that the sample error is in the region of 7% (6.9% at the 50 per cent mark).

So, on that basis, in Brisbane, the Greens vote, all other things being equal, will be between 30.5% and 16.5%, while Labor will be between 41.5% and 27.5%. That means there is a very good chance in the that electorate that Labor is ahead of the Greens.

In Ryan, the Greens vote would be between 42.3% and 28.3%, and Labor between 35.9% and 21.9%. It looks like the Greens will be ahead of Labor.

Trevor Evans' vote would be between 46% and 32%, and Maggie Forrest's between 42.3% and 28.3%.

It strikes me as unlikely that Brisbane would be in a  better position for the LNP than Ryan, given the historical margins in these seats, so there is your first sign that the poll may not be representative. (That doesn't mean that it is not indicative and can provide clues as to what is happening).

Even on those figures it is possible that both LNP candidates could win, and extremely likely that one Greens candidate will be eliminated. The Newscorp commentary includes a snippet that the LNP claims their internal polling has Ryan at 50/50. These figures do not contradict that because of the extremely large error margins.

More than that, these are snapshots of what might happen if everyone voted today, but they won't be voting until next week, and votes can and do move significantly over those periods of time.

I thought you might like some analysis of these figures. The media, and other interested parties, will be trying to spin them to their advantage. It's important not to read the wrong things into them, or to be diverted.

Regards,

GRAHAM YOUNG
EXECUTIVE DIRECTOR