Are voter-intention polls hurting Canadian democracy? Polling companies badly misjudged the outcome of the B.C. general election, calling for an NDP majority government at the start of the campaign. It was the Liberals who won the majority on Tuesday, and now election-day exit polling from a major firm reveals that one possible reason for the upset was that NDP voters – perhaps convinced their party was cruising to victory – stayed home and didn't vote. "The long and the short of it was that NDP voters did not get out and fulfill their promise to vote for the party of their choice – they stayed home while Liberal voters showed up," Ipsos Reid says.
This is problematic. Polling firms also badly miscalled the outcomes of the Alberta general election in 2012 and the Quebec vote in 2011. In two of three recent elections, polls showing voter intentions at the start of the campaigns predicted an easy win by the eventual loser, or, in the case of Quebec, predicted a far stronger showing by the eventual winner than actually occurred.
The polls about voter intention that are released the day a writ is dropped tend to frame the narrative for the entire election campaign. In B.C., the NDP started the campaign with a 20-point lead in the polls. Media, as well as the political parties themselves, bought into the story of an election that was the NDP's to lose. Even in the final days of the campaign, this was still the story in the headlines, in spite of the fact some polls were showing that the election was getting closer as time went by.
Did this narrative create complacency among NDP voters? Did polls predicting a Wildrose Party victory in Alberta do the same to that party's voters, contributing the Progressive Conservatives' surprise re-election? It seems likely that they were a contributing factor.
Polling experts are now debating among themselves about whether their methods and timing are as good as they should be. Most recognize that voters' intentions change, or harden, over the course of an election campaign, especially in the era of negative advertising. There are also issues of collecting responses via landlines, cellphones and over the Internet, and whether these different technologies skew results along age and gender lines.
The bottom line is that polling companies need to track voters' intentions (perhaps including whether they actually intend cast a ballot) throughout an election campaign and make sure that the data are as up-to-date and scientifically sound as possible. There is no reason, if the sample size is large enough, the questions are properly framed and the analysis is sound, that polls can't be accurate in the final week of a campaign. Polling companies only have themselves to blame if their predictions are bogus.
The media, too, have a role to play, by breaking out of preset, poll-driven narratives and doing their own legwork. But in the end, only one thing can keep a voter at home, and that is the voter. There is no valid excuse for complacency. Every vote counts when it comes to Election Day, the only poll that matters.