The Courier-Mail is this morning touting a Galaxy Poll which purports to show Labor winning only two additional seats in Queensland – Bonner and Moreton. The poll, of 800 voters across four seats – Bonner, Moreton, Longman and Herbert – has Labor on 51% of the 2PP vote.

Psephologist Adam Carr has calculated this represents a 5.1% swing from 2004, which if replicated statewide, would place all the other Coalition seats out of range. Unhelpfully, no detail of the poll is available on the Galaxy website.

A commenter at psephological blog The Poll Bludger calculates that the margin of error in the poll would be 7% for each seat on the assumption that 200 voters were surveyed in each – a very small sample. The poll also appears to make some unwarranted assumptions about Family First preferences – which as I’ve previously reported may be directed to Labor in Queensland. Galaxy appears to assume all of the Family First vote flows back to the Coalition.

The Galaxy poll contradicts other evidence that Labor is surging in Queensland from the aggregate Newspoll and from decisions about campaign priorities across individual seats made by both Labor and Coalition strategists. Internal Labor polling shows the party streets ahead in the marginals.

Galaxy may be telling a story about a much narrower Labor lead in the marginal seats than in those which are less marginal on paper but its apparent methodological flaws suggest it should be taken with a big pinch of salt.

But it may play a role in the “media narrative” at the start of the campaign either as a “poll bounce” or a “Queensland deserts Rudd” theme – appearing to offer the government some hope when all the national polls, including this week’s Newspoll, continue to show a landslide to Labor.

But then a touch of the underdog might assist Kevin from Queensland.