As Canadians ponder the calling of a national election over the next week or so, they are being confronted with much professional political commentary that has reduced the election of our 44th Parliament to a single question: Will Justin Trudeau and his Liberal Party win another majority as they did in 2015 or be held to a minority as they were in 2019?
This certainty about the possible results is based on the consensus of credible public opinion polls, including those produced by Ipsos for Global News. But are the polls right this time? Experience shows they usually are but sometimes even the consensus of polls can be wrong. The last two presidential elections in the United States are examples of this.
To be clear, this isn’t about the odd rogue or outlier poll getting it wrong. Savvy political experts know a bad poll when they see it and discount it immediately. This is about all or most of the best polls missing the mark. How does this happen? And is there a risk it could happen in Canada this time?
There absolutely is potential for the polls to be missing some important campaign forces this time. But the problem usually pointed to by those who have a surface understanding of polls — demographically unrepresentative samples — isn’t it. Pollsters have a variety of ways to deal with this problem and generally do a good job of representing the demographics of the electorate in their samples.
The big problem — and it certainly was the biggest problem in the U.S. presidential misses — is under or over-representing election turnout. That is the number of people who will vote on Election Day. But the problem is more specific than just misestimating overall turnout. It is misestimating what pollsters call differential turnout. That is how motivated the supporters of each party are to show up to cast a ballot.
Why is this a particularly daunting problem for pollsters in this upcoming election? It’s because the dynamics this time around are very different from what they were in 2015 and 2019. In both 2015 and 2019, the electorate was very engaged and voter turnout was the highest we’ve seen this century. That’s because voters believed both elections were consequential, and the outcome was uncertain. They were engaged in the election and voted in record numbers.
How are voters feeling about this election? As the kids like to say, “meh.” There’s a lack of enthusiasm about any of the leaders or parties, and the ballot question is up in the air.
Are we being asked to evaluate the Government’s performance dealing with the pandemic, or will this election be about competing visions for the post-pandemic recovery? Both questions are odd and out of sorts with the public mood given that we are stuck in a pandemic cul-de-sac with an emerging fourth wave.
Add in the potential health issues associated with voting in person, and the learning curve associated with alternative methods of voting, and this election has the potential to generate a lower turnout than the last two elections.
As we saw in 2011 when Stephen Harper and the Conservatives won their majority, low turnout helps the Conservatives. The Liberals need a higher-than-typical turnout to win.
Another issue that could lead to some surprises for pollsters is correctly measuring the motivations of progressive voters. This could not only affect turnout, it could also impact the party choice that voters who have a more left-leaning orientation make.
The Liberal Party’s election strategy counts on a specific dynamic to unite progressive voters behind the Liberal Party as their preferred option. This strategy requires the Conservative Party to be seen by progressive voters as the potential winner of the election. The Liberals use this threat to marginalize other progressive parties, but especially the NDP, as viable choices for voters who want to prevent a Conservative victory.
What happens to the Liberal strategy to unite progressive voters — especially those less attached to voting at all — if the Conservatives aren’t a threat to win as they are this time? Maybe the NDP and Greens become more viable choices. Maybe, though, there will be something better for these progressive voters to do on election day other than to vote. That’s certainly what happened in B.C. in 2013 when the consensus of polls showed the NDP defeating the Liberals and the NDP vote failed to show up.
All these factors together represent serious risks for polling in this election. If the pollsters get it wrong this time you now know why.
Darrell Bricker, Ph.D., is chief executive of Ipsos Public Affairs and leads Ipsos’s political polling team for Global News.