National Post (National Edition)
Polls managed to nail result of Ontario vote
Campaign polls are frequently denounced as presenting a false view of elections as horse races, and for reflecting entrenched political bases without capturing a party’s momentum. Then, when the actual results turn out to be different than poll results, polls are criticized for simply being wrong.
According to an analysis of the polling that tracked last week’s Ontario election, however, the pollsters absolutely nailed it this time. As the governing Liberal party deflated so far that it lost official party status, the Progressive Conservatives protected a strong lead, and the NDP surged to within plausible grasp of victory before settling back into second place, the 11 polling houses that published results in Ontario all converged on results that were accurate to within barely one per cent on average.
This is the conclusion of Harold Clarke, editor of the journal Electoral Studies and professor in political economy at the University of Texas at Dallas.
“Pollsters have put a lot of effort into trying to improve their samples and so forth. It is sometimes in reaction to the criticism that they’ve taken,” Clarke said in an interview.
When the 2015 Canadian election came to a close with a Liberal majority, for example, the Conservatives and NDP were not the only losers. Public opinion pollsters took such an unusually vicious shellacking in public opinion — a terrain on which they ought to be more surefooted — that the National Post editorial board felt compelled to defend them in an editorial.
“On the same day, one poll will have one party in the lead and another in third, while another poll will have the reverse,” the editorial re- called, before settling on the conclusion that, for all their flaws, “voters have every right to consult the polls as part of their deliberations.”
This reflects top level court decisions, which have overturned previous bans on publishing polls during campaigns, for fear they will sway the outcomes. Lately, however, the real crisis for pollsters has not been a surplus of public trust, but a deficit.
Recent polling disasters include the 2015 British general election, in which Conservative support was pegged far lower than it turned out to be when they won a majority.
There were similarly inaccurate polls in B.C. in 2013 (in which Christy Clark’s victorious Liberals seemed to trail the NDP throughout), and Alberta in 2012 (when the Wildrose Party seemed a sure bet, but lost). Last year’s re-election of Calgary Mayor Naheed Nenshi was notable for the polls that suggested he had no chance, which the pollster later called a “catastrophic failure.”
One of the most dramatic was the 1992 British general election, the first after Margaret Thatcher’s resignation, when all the polls had Labour winning under Neil Kinnock, but in the end the Conservatives under John Major took a majority. More recently, in the Brexit referendum, polls were all over the place, with the telephone polls for some reason being particularly wrong in hindsight.
“The real trick is to move from votes to seats, and that’s a more difficult exercise, or in the American context to go from the popular vote through to the Electoral College. That’s where they really fell down in the most recent presidential election,” Clarke said. “That step is really complicated.”