THE POLLS ARE NOT WRONG

David Parker, April 5 2025

Denying reality does not make it disappear. All over social media, in private conversations, and in messages from friends, I keep hearing the same phrase: “I don’t trust the polls.” This isn’t a serious position—it’s the slogan of a campaign in denial. It’s the last refuge of people who have run out of arguments and are now retreating into wishful thinking. Rejecting reality because it’s uncomfortable is not a strategy; it’s a delusion. Ironically, it mirrors the very mindset we criticize in progressives—those who deny biological truth, economic consequences, or the moral order because they find those realities inconvenient. If we are to win, we must be more grounded than that. We must be rooted in facts, not feelings.

No one is saying you should blindly trust a single poll. Any individual survey can be flawed. A pollster may use a poor sample, apply biased weighting, or simply get it wrong. But rejecting the entire polling industry because one poll offends you is like refusing to believe in thermometers because one brand gave you a slightly off reading. It’s silly at best—and dangerously self-deceptive at worst. Polling is not a tool of the mainstream media. It is a scientific method, based in statistics and probability. While imperfect, it remains the most reliable instrument we have for measuring public sentiment in the lead-up to elections.

Let’s look at the data. The polling aggregator 338Canada has been tracking elections for years. In the last 18 general elections—provincial and federal—their model made 2,039 individual riding predictions. Out of those, it correctly projected the winner in 1,821 races. That’s an accuracy rate of 89.3%. Of the remaining 218 ridings, 132 were within the model’s margin of error. Only 85 predictions—just 4.2%—were total misses. That level of precision is remarkable. If someone could forecast the stock market with that level of accuracy, they would be wealthier than Warren Buffett. This isn’t a guesswork industry—it’s math. It’s probability. And the numbers speak for themselves.

People often claim that Donald Trump “proved the polls wrong” in 2016 and again in 2024. But that simply isn’t true. In 2016, he lost the popular vote by millions of ballots, just as the polls predicted. He won the presidency through narrow victories in key swing states—victories that were within the margin of error. In 2024, polling was even more precise, forecasting a near dead heat in the electoral college. The idea that Trump’s victories represent some kind of polling failure is a myth—one created by people who don’t understand how margins of error work.

Yes, there are bad polls. Outliers exist. There are firms that skew Liberal or Conservative. Frank Graves of EKOS, for example, recently released a poll showing the Liberals over 40% in Alberta—an absurd result. And there are newer firms, like Liaison Strategies, that are still establishing their track records. But this is why poll aggregators exist. They weigh pollsters according to historical accuracy, adjust for house effects, and build models based on rigorous data. Websites like 338Canada, Eric Grenier’s Poll Tracker, or Bryan Breguet’s Too Close to Call consistently produce results that closely match actual election outcomes.

To dismiss polling simply because you don’t like what it says is not analytical—it’s emotional. It’s no different than progressives being unable to define what a woman is because their worldview is detached from physical and biological truth. If we are to defeat the left, we must be better than them—not just morally, but intellectually. That means engaging with data, even when it challenges us. That means seeing reality clearly and preparing accordingly. The polls are not our enemy. Denial is.