Poll numbers can feel definitive, but every figure on this site comes with caveats that matter. This guide is a plain-language overview of how to read what you see here — what the rolling average actually averages, where margins of error come in, why two pollsters can publish different numbers in the same week, and why a small vote-share shift can swing more seats than you'd expect under Canada's electoral system.
The big bars at the top of each jurisdiction are a rolling average across recent polls, weighted toward larger samples and more recent fieldwork. A single poll is a snapshot of one pollster's sample at one moment; the rolling average is meant to smooth across the spread of pollsters so any one outlier matters less. The trend arrows reflect the seven-day change in that average, not the change in any individual poll.
Every poll has a margin of error tied to its sample size — roughly ±2 to ±3 points for a typical 1,000–2,500 respondent national poll. That margin applies to each party's number, so a one-point change between polls is often inside the noise. The Recent Polls table shows sample size for every row so you can see how thin or thick the underlying signal is for any given pollster.
Every pollster uses a slightly different methodology — random-digit-dialing vs. online panels, different weighting schemes, different question wording. These choices produce a small but persistent 'house effect' where some pollsters consistently land a few points higher for one party than the rolling average. Same-week disagreements between two firms are usually a mix of sampling noise and house effects, not one firm being right and another being wrong.
Canada uses first-past-the-post — every riding elects whoever finishes first, regardless of margin. That means a small popular-vote swing can produce a much larger swing in seats, depending on where each party's votes are concentrated. Regional concentration matters more than the national headline: a party with a thin national lead but a strong Quebec presence can take more seats than a party with a broader but flatter base.
The leader-favourability card is not the same as the party-support card. Approval and favourability scores reflect public opinion of the leader as a person and tend to move slower than vote intention. A leader's approval can rise while their party's support falls, and vice versa.
Every poll on the Recent Polls list links to the original publication. If a number on the dashboard looks surprising, opening the pollster's report is the most reliable way to understand what was actually measured and how. The dashboard summarises; the source is the record.
Last updated: 2026-06-04