The Age trumpets this result as Victorian Labor “defying the national trend”. No doubt other papers are saying the same – I haven’t looked.

I’ve been arguing for a while that there is insufficient evidence to conclude that there is a national trend against Labor, and that in fact thinking about disparate polls in seven different jurisdictions with differing political histories, cultures and current circumstances as constituting a trend makes little sense. My contention for a long time has been that elections are unrepeatable and singular events and that epistemologically we can know much less about electoral behaviour and find grounds for prediction with much less certainty than we think. Political behaviour follows few laws and a lot of conclusions reached after the fact are questionable.

But there is a sort of reflexivity feedback loop built into the way we think about politics and the way polls are reported. Particularly at state level – where polls are few and far between – one poll which struggles to form a series can have a large impact on perceptions, and thus the interpretations of the public and the press and the morale of politicians and “momentum”.

Consider what would happen if we were to see a poll showing similar figures for Labor in Queensland. I don’t think that’s at all unlikely. The “political narrative” about Anna Bligh and Lawrence Springborg would instantly change. A whole range of events now in the past – Beattie’s reign, the formation of the LNP and others – would be coloured in quite differently on the interpretive palette. And this would change the reality of state politics.

The truth of polls and the interpretive work that goes into “explaining” their results is that what we are dealing with is the manufacture of social facts rather than scientific certainties. There’s a “truth effect” at work which shifts or creates a new situation. It’s worth keeping this in mind at all times when thinking about how polls affect the political landscape.