If you’ve been following the polls in this election campaign, you know they’ve been anything but stable. One week, the Liberals are surging. The next, the Conservatives regain ground. So how should we make sense of it all?
After spending time in Canada this year, I’ve come to appreciate just how unpredictable Canadian politics can be. Former Prime Minister Justin Trudeau’s resignation earlier this year signaled a turning point. With Pierre Poilievre’s Conservatives leading the polls for months, a change in government seemed inevitable. But now, with just two weeks left, the race is tightening.
Polling averages—like those from 338 Canada (Yes They Know It Is 343 Now) and CBC’s Poll Tracker—currently show the Liberals up by 6.2%. But dig a little deeper, and the picture gets complicated. While some firms predict a decisive Liberal win, others, like Mainstreet Research, still see the Conservatives ahead. What explains these conflicting results?
One: How They Poll
Part of the answer lies in how polling is done. Mainstreet, for example, uses automated phone calls (IVR) to reach voters—typically skewing toward older demographics. EKOS, on the other hand, combines IVR with online paneling, which can capture younger, more digital-savvy respondents. Differences in methods, sample sizes, question wording, and timing all influence the outcome. Even daily news headlines can shift public opinion overnight.
Two: Where Support’s From
It’s also worth noting where Liberal momentum came from. Recent polls suggest the party gained support from NDP and Bloc Québécois voters. But with shifting priorities and tightening margins, that support could easily drift again before election day.
The Conservative base is slightly less national, in Alberta for example, they will win some ridings with 80% of the vote. This means that even if they catch up in the overall popular vote across Canada, they may win fewer seats because of the overwhelming wins in some areas and not enough “swing riding” wins in other provinces.
So what should we make of all this?
Polls can offer valuable insight into trends, but they’re not predictions—they’re snapshots. And while they can shape narratives, they don’t determine outcomes. Only voters do that.
There’s just one poll that really matters. It happens on election day.
Want to keep digging deeper? Read on below!
Let’s compare two major independent polling companies as an example. Mainstreet Research uses IVR (integrated voice response) to collect data. This means Canadian phone numbers are called at random (thought to generally reach an older audience), played a previously recorded message, and prompted to press buttons corresponding to their voting preferences.
In contrast, EKOS research uses a combination of IVR and ProbIt paneling. In this method, thousands of verified Canadians are selected at random to become part of a group the research company gains insights from over time through phone and online surveys. We know that these different data collection methods can lead to different results (e.g., phone surveys are thought to generally reach an older sample, where online surveys tend to reach a younger group), but it isn’t the only factor at play. The way polling questions are phrased, fluctuation in voter priorities, and changes in daily news headlines can significantly sway the results of a survey.
We pulled out these two companies as examples, since when they make seat count projections, they can make very different conclusions.
The point here isn’t to make you more confused, or to say you should stop watching polls altogether. They can be a useful tool to observe changes in many Canadians’ feelings and priorities over time. But we should be reading them through the lens of their limitations, rather than taking them as gospel. After all, here at ONE, we believe good data can drive good policy!
