August 16, 2010
It's not as brave a pick at this stage of the game as it might have been a week ago - although the Coalition may again move back up the charts if Tony Abbott manages to get through tonight's Q and A without choking on his tongue - but a victory for the ALP, at least in terms of the two party preferred vote, seems the likely result of this peculiarly tiresome election, as the percentages on polling day come out closest to the numbers from the opinion polls conducted shortly before the election was called, which overall favoured Labor. Here's why:
Judging from Gallup Polls in the United States, the United Kingdom, and Australia, opinion often changes during an election campaign. Come election day itself, however, opinion often reverts back nearer to where it was before the campaign began... One explanation, we suggest, is that people become more responsible when stepping into the poll booth: when voting they reflect back on the government's whole time in office, rather than just responding (as when talking to pollsters) to the noise of the past few days' campaigning...A recent illustration of this was the evaporation of Cleggmania in the last UK election:
[The] recent research by Robert Goodin and James Mahmud Rice suggests that something more complicated might be going on. The polls, they reveal, don’t fluctuate in the run-up to an election because respondents are simply humouring the pollsters with the pretence that their opinions are shifting - their opinions really are shifting. Looked at over a 40-year period, general elections in Britain and Australia show that there is usually a swing of up to five percentage points between the start and the end of a month-long campaign. Then on the day of the vote opinion swings all the way back again. Polls taken at the start of the campaign are invariably closer to the final result than the ones taken at the end. So it’s not that people don’t change their minds. They change their minds twice: once in response to the excitement of the campaign, and once in the privacy of the polling booth.Presumably this goes double when "intervening excitements" are notable by their absence.
That was clearly what happened this time too. The polls were nearer to predicting the final result a month before the vote than a day before the vote, when the pollsters overstated Liberal Democrat support by three or more percentage points. There are two possible explanations for this. One is that people who said they had switched to the Lib Dems after the debates didn't in the end bother to vote (perhaps because many of them were, in the word politicians most dread, 'students'). But Goodin and Rice show that this isn't very plausible: there is no evidence that people whose opinion fluctuates are also somehow lazier (Australia, where voting is compulsory, is the test case here). The alternative explanation is what they call the '"Responsible Voter" hypothesis'. During the campaign people respond to the immediate pull of media-driven events. But when they come to vote they pause to reflect more broadly on the period since the previous election, which takes them back to the start of the campaign and puts all the intervening excitements into perspective.
Nearly always isn't always, of course; and the votes cast still have to filter their way through the lottery of our regionalised electoral system. But I don't mention this in order to shore up any optimism; to be honest, I couldn't give a toss about the major party contest, and if Labor loses my only response will be to smirk and think "serves you right for whoring yourselves to the mining lobby and a few racist bogans in marginal seats". It's just that if the Goodin and Rice thesis is confirmed again it will provide an ideal opportunity to spread the good word that everything that happens during the campaign period - every debate, every launch, every ad, every analysis of an ad, every grotesque promise, every piece of punditry and psephology and limp-lettuce-lame satire, every gaffe and spin and narrative and all the vacuous, trivial ephemera of the horse-race style coverage, every increasingly desperate attempt by the Murdoch golems to make their master's voice heard, every last bit of it - is a complete and utter waste of time.
That would be nice, is all. That and a Green balance of power, mwahahahahahahah.
Update: It turns out Wikipedia, under the heading "Minimal Effects Hypothesis", has a link to the full Goodin and Rice article (pdf). As David Runciman points out in the LRB piece quoted from above, the Goodin and Rice finding differs from the more traditional version of the Minimal Effects Hypothesis in arguing that the fluctuations in poll results during election campaigns are not illusory but merely ephemeral.