As my colleague William Bowe (the Poll Bludger) reports, the Australian Electoral Commission yesterday published the final two-party-preferred results from last September’s federal election. They show the Coalition recording 53.5% to Labor’s 46.5%. Antony Green also has a post on the figures, as does Peter Brent at Mumble.

The reason this happens so long after the election is that it requires the Commission to go back and recount preferences from the seats in which the top two candidates were something other than Labor and Coalition – what it calls “non-classic divisions”. This is evidently not something it treats as a high priority.

This year there were eleven such divisions: four in which the final contest was Coalition vs a rural independent or minor party, three in which it was Liberal vs National, three in which it was Labor vs Greens and one Labor vs independent.

There are a number of reasons why these seats are particularly worth looking at (one of them, Fairfax, was the closest seat in the country, with Clive Palmer winning by just 53 votes), but to my mind the interesting thing is the fact that they are the only places in which we see major party preferences being distributed. And since the vast majority of Australians vote for major parties, any information about how those people think is worth having.

Not surprisingly, when Labor voters are given a choice between a Coalition candidate and someone else, they vote overwhelmingly for the someone else. Palmer got 81.2% of Labor preferences, Bob Katter got 84.6% in Kennedy, Cathy McGowan got 88.7% to beat Sophie Mirabella in Indi, and Rob Taber got 75.4% in an unsuccessful attempt to beat Barnaby Joyce in New England.

Choosing between Liberal and National candidates isn’t so easy. Only 52.8% followed the how-to-vote card and preferenced the Liberals in Mallee, while 57.3% and 65.9% went to the Nationals in Durack and O’Connor respectively (the latter would have been inflated by the donkey vote). All three are large rural electorates, so distribution of Labor how-to-vote cards would have been patchy at best.

Most interesting is what Liberal voters do. In Denison they were told to preference independent Andrew Wilkie, and they did, to the tune of 83.6% – also inflated by the donkey vote.

In the other three, Liberal voters were told to preference Labor ahead of the Greens. They did, but much less enthusiastically: 71.3% in Wills, 67.4% in Batman and just 66.3% in Melbourne.

In 2010, by contrast, when Liberal preferences were directed to the Greens, they flowed much more tightly: 73.5% in Grayndler, 80.0% in Melbourne and 80.9% in Batman. That confirms a few years’ worth of data from Victorian state elections that show the same pattern. It’s harder to get Liberal voters to preference Labor than to preference the Greens.

So does that mean that ordinary Liberal supporters disagree with their party’s recent decisions, both federal and state, to preference against the Greens? Not necessarily. For a start, Liberal voters in the inner city are not representative of the whole body of Liberal voters; it just happens that they are the only ones we have the information for. It’s plausible to think that preferences in the middle or outer suburbs might tell a different story.

In any case, even if Liberal voters as a whole would rather preference the Greens, I doubt that reflects any serious assessment of the relative merits of the Greens and the ALP.

Like the Labor preferences in Coalition vs independent seats, what it shows is that voters tend to have a two-party mentality. The main game is Coalition vs Labor, and anyone else is at least a potential ally when matched against the real opponent. It’s not that Liberals actively disagree with the Tony Abbott/News Ltd view that the Greens are more philosophically alien than Labor, rather that they think it’s beside the point.

And, of course, most of the time that attitude is an accurate reflection of what’s really going on in our political system. In only a handful of seats is there any reason for voters to think about “non-classic” preference issues. While the Greens and Palmer have both made some inroads into the two-party setup, they will need to go a lot further to challenge its hold on the way voters think.

Abbott’s directive to preference against the Greens, while in one sense it might seem to undermine that two-party system, is designed in the long run to reinforce it by reducing the relevance of a competitor. In that sense at least, he may well be doing what his voters want.