“All the pollsters can’t be all right all the time”. Bob Dylan said something like that. So what can we predict about next Saturday’s result based on the most recent polls?

Reading the fine print is instructive, the Galaxy Poll states, “5% uncommitted or refused excluded”. Even more tantalising is the Mark Latham Hypothesis made public in the Weekend Financial Review; “growing numbers of respondents are using the polls for tactical purposes”. Specifically because so much commentary is poll-driven and respondents want to “be part of the news”. Latham concludes, “Most of the polls claim to have a maximum margin of error of 2 to 3 percent. Due to the protest vote factor, however, I believe it is now prudent to double that amount. People should factor in the possibility of errors of up to 4 or 5 percent.”

Rod Cameron, another Labor man, albeit not an embittered one, had a fortnight ago discussed a poll for the seat of Bennelong on Lateline, “I do really think that it is going to close up. I think that’s too big a margin for Maxine McKew at this stage. Labor is ahead but they are not over the line by any means, either in Bennelong or Wentworth.”

Now with these caveats in mind, examine the creative inferences that are being drawn about polls. The Galaxy Poll findings were based on a survey of 800 voters in Eden-Monaro, Page, Robertson and Wentworth. Based on what respondents told pollsters it has been inferred that primary support for the Liberal Party has slumped by more than 5% since the last election.

If this is the case and if the flow of preferences is the same as in 2004 this would result in a two-party preferred of 53% for the Labor Party in the polled marginal electorates. If one infers that this is a guide to a national swing then the Coalition will lose Bennelong and Wentworth. There’s a lot of “ifs” there. Here’s one more. If the Coalition sneaks over the line then the Mark Latham Hypothesis will have to be taken very seriously.