I've covered more than 100 elections in the past 40 years and have seen upsets, reversals and some big surprises. But what we saw Tuesday night in British Columbia was in a class of its own.
Our careful work and that of other reputable pollsters pointed to the almost certain defeat of Christy Clark's Liberal government. Yes, there were indications that things were tightening up. But polling three days before the election suggested that the New Democrats still enjoyed a comfortable lead.
It's easy to write off the pollsters and say Ms. Clark actually enjoyed a strong lead throughout the final stages of the contest. But that ignores the bigger lesson: Campaigning, in the classical sense of the term, matters more than ever.
In an age of almost effortless access to everything from clothing and airline tickets to newspapers and movies with the click of a button, making the effort to go out and vote requires a physical commitment.
The Liberals won this election with one of the lowest turnouts in provincial history. Polls showed an NDP train coming toward B.C. What they didn't accurately reflect was the Liberals energizing their base in resistance. The New Democrats ran an overly cautious and ambiguous campaign from the start and ultimately failed to motivate their base.
This campaign also puts to rest any doubt about the effectiveness of negative advertising. Until the dying days of the campaign, the NDP remained stubbornly on the high road, while the Liberals ran an expensive and relentless campaign of TV advertising that framed the ballot-box question – the economy – and prevented Adrian Dix from rising above Ms. Clark on the key metric of who would make the best premier.
Now, pollsters must figure out how our projections were so off. We needed to better ensure that our samples reflected demographic growth and shifts among Metro Vancouver's burgeoning ethnic communities. We needed to have communicated more clearly to reporters that polls are snapshots in time of voter intention, not iron-clad predictors of future behaviour. And we must recognize that we're moving into a new era in electoral politics in Canada – like the weather, more prone to dramatic shifts; and like hockey, more likely to produce surprise endings.
The NDP's Greg Selinger trailed the Progressive Conservatives by 20 points before Manitoba's 2011 election but fought his way back to victory. Liberal Dalton McGuinty overcame a huge pre-election deficit to win in 2011. Even more pronounced was the comeback of the Alberta Progressive Conservatives, who were significantly behind Danielle Smith's Wildrose Party until the 2012 campaign's final day. We caught the shifts in Ontario and Manitoba but missed the upset in Alberta, just like we missed it in B.C.
I don't think this spells the end of polling. The private side of the business, funded by political parties to help them hone their messages and identify supporters, is thriving. In the last U.S. election cycle, more than $180-million was spent on public-opinion surveys that were never publicized.
It's impossible to imagine elections without polls. But in a new era of instant everything, the future will see less dogmatic devotion to what they're saying. While our methodology requires review, polling models aren't inherently broken. But they are less compatible with the way parties play politics today, and the way the public mood quickly shifts. Public polling has to change with the times and find the resources to dig deeper into the more diverse – and fickle – Canadian psyche.
Angus Reid is chairman of Angus Reid Public Opinion.