Results Confound Pollsters Numbers Meant Little In The End Were People Lying To Them?
Saturday, 24th March, 2007
State Opposition Leader Peter Debnam, and anybody else hoping for a Liberal victory in NSW today, should get a copy of the Toronto Star of 28 June 2004 where those headlines appeared above a story describing how the Canadian Liberals won an election. As Stephanie Levitz of the Canadian Press Agency put it, "Canada's electorate appears to have confounded the pollsters. Weeks of speculation, number crunching and supper-hour phone calls to more than 25,000 Canadians over the last five weeks meant little in the end as the Liberals beat projections that they were headed for a sound thrashing in the election..."
That Canadian election was a wonderful reminder of the power of the underdog effect and a lesson in why election watchers should studiously ignore the opinion polls. For weeks the Canadian pollsters and pundits were predicting a massive decline in support for the governing Liberal Party with the opposition Conservative Party said to be in with a real chance of victory. Heading into the vote, two of Canada's leading pollsters had predicted about 32 per cent of Canadians would cast a ballot in favour of the Liberals, followed by 31 per cent voting for the Tories. So there was plenty of egg on lots of faces when the Liberals were returned to office, albeit with a decline in their number of members, having gained 37% of the national vote.
Those of us with experience in Australian elections were not surprised by this Canadian outcome. The underdog effect has had an extraordinary impact in our elections, state and federal, over the last 20 years or so - just ask Paul Keating and his true believers! The last thing a political party should want to be seen as is a winner. Just ask Jeff Kennett and Wayne Goss.
Which is a rather long way of saying that the general expectation that there is no point in waiting for the votes to be counted tomorrow night before anointing Morris Iemma as a Premier elected in his own right might yet prove premature. This election where the Sydney newspapers this morning report a massive indifference by people about who governs them for the next four years is tailor made for a few surprises.
The Glug Election Indicator, based on betting markets, might have Labor a 90% chance of winning and the polls might have Labor 10 or more points in front but the collective wisdom of the entrants in our "NSW Bragging Rights Tipping contest" is erring on the side of treating the opinion polls with caution. The average of their predictions is that Labor will win 49 of the 93 seats – quite a slight overall majority when 47 are needed for a party to govern in its own right.
Comments