I was a bit harsh earlier this month when I said it would take the Queensland electoral commission “weeks” to produce more detailed figures from the 25 November state election. In fact preference data was released quite quickly, although it’s not easy to find: go to the list of electorate results, click on the electorate you want, then follow the link for “booth details”, and at the bottom you’ll find a “Summary of Distribution of Preferences.”
I won’t try to put a precise number on it, but it looks as if the two-party-preferred swing to Labor was at the low end of my previous estimate: about 1%, maybe even a touch under. I still think that was underwhelming, but at a time when incumbents seem to be doing it tough in most places it has to be counted as a success.
Tim Colebatch last week used the detail to produce a very interesting piece on the impact of preferences, and specifically One Nation preferences. He is arguing against the conspiracy theory, apparently promoted by News Corp (which I don’t read), that Labor’s victory was an artifact of the revival of One Nation.
Colebatch shows that preferences made almost no net difference – they “largely cancelled each other out” – and that the majority of One Nation preferences actually flowed to the Liberal National Party.
But that doesn’t dispose of the argument. For comparison, if the ALP wins a bunch of seats on Greens preferences, it doesn’t follow that the presence of the Greens benefited the ALP. What you need to know is what those voters would have done in the absence of the Greens; if they were going to vote Labor anyway, the fact that those votes instead passed through the Greens makes no difference either way.
The implicit view from the right is that One Nation’s voters are all rightly “theirs”, and that therefore any One Nation preferences that flow to the ALP (even if they’re a minority) represent a net loss to the LNP. But this is a piece of ideology, not science.
We’ve actually had this discussion before, during One Nation’s first incarnation, in the aftermath of the Western Australian and Queensland elections of 2001. If you’re at all interested, go back and read the 2004 debate between Antony Green and me on the impact of the One Nation vote.
The key point to emerge then was that the 2001 Queensland election was a good natural experiment. One Nation contested about half the seats, and the average two-party-preferred swing to Labor was the same in both groups. That strongly suggests that its presence made very little difference.
So what has changed in 16 years? This year, One Nation contested 61 of the 93 seats. Leaving aside for the moment the 27 seats where the final contest was not between ALP and LNP (all but one of which had a One Nation candidate), the mean swing to Labor was 0.8%. In those without a One Nation candidate, it was 1.7%; in those with, just 0.1%. (Taking medians instead doesn’t change it much: the three numbers then are 0.6%, 2.2% and 0.0%.)
That looks as if the presence of One Nation actually hurt Labor – the opposite of News Corp’s argument. But that would be too hasty a conclusion. The seats with One Nation candidates are nothing like a random selection; they are heavily concentrated in the non-metropolitan areas, where all commentators agree Labor did particularly badly.
So a fairer test would be to look at just the seats in the Brisbane/Gold Coast/Sunshine Coast metropolitan area.* On that basis, the effect disappears: the overall swing to Labor in the region is 1.6%; for seats with a One Nation candidate it’s 1.5%, for those without it’s 1.7%.
There’s another way, however, of looking at this, which is to not just look at seats with a traditional ALP vs LNP split, but to consider One Nation as part of the non-Labor coalition. That brings another 12 seats into the calculation, where the final contest was between One Nation and Labor; one could also add Maiwar, where it was between LNP and Greens.
Doing that gives us an average swing to Labor overall of just 0.2%; in the seats without One Nation candidates it’s 1.8%, whereas in those with them there’s a swing against Labor of 0.7%.
Again limiting ourselves to just the metropolitan south-east, the average swing to Labor is 1.2%; 0.7% with One Nation, and 1.8% without. (Medians are 1.3, 0.9% and 2.3%.) That’s a much more substantial difference – although probably still not an election winner – but unfortunately for News Corp it’s in the opposite direction to what its theory would predict.
So, depending on just how you do your calculation, it looks as if the effect of having One Nation in the field was either so small it could just about be a rounding error, or else was moderately significant in favor of the opposition. But on no story did it help Labor.
That’s on the numbers, anyway. Politically, of course, it’s quite a different matter: it’s very likely that the LNP lost votes because of its dealings with One Nation, just as it did in Western Australia. So in that sense Labor benefited from One Nation’s return.
But that’s no help to News Corp either, because the LNP came to grief precisely through following its preferred strategy, of jumping into bed with the far right.
* No two analysts are likely to define this area in exactly the same way; I’m using the map in William Bowe’s election guide as my criterion.