something worth reading • news, opinion & more

Statistical modelling shouldn't be ignored even though parties love 'em when they're positive, hate 'em when they're negative

If you want to know who is going to win this federal election you should be paying attention to an astrophysics professor from a small technical college in Saint-Laurent, a borough in the heart of the island of Montreal.

Philippe J. Fournier teaches both physics and astrophysics at Cégep de Saint-Laurent. His website 338Canada.com is not a poll. Fournier isn't asking voters questions, he does something much better by aggregating all the polls conducted by name-recognized pollsters such as Angus Reid, Nanos, Abacus, Mainstreet, and Liaison.

He then weighs the polls according to their sample size and field date, pumps it into his formula, a formula that includes a variable based on a subjective editorial decision.

The formula looks like this:

In this formula, the weight of a poll (w) is proportional to the square root of its sample size (n) and diminishes as time (t) passes by. Also, a polling firm rating (f) is attributed to each poll, the variable.

"I know it can/will be a source of criticism, but I stand by it — and it is a variable, not a constant," Fournier explains.

After all that formulizing and aggregating, that most of us don't understand, Fournier's latest projection for today, April 10, 2025, has the Liberals getting close to 44 per cent of the popular vote, the Conservatives 37 per cent, the NDP eight per cent. The seat projection has the Liberals and Conservatives at 196 and 121.

Then there is Éric Grenier who runs a website and a newsletter called The Writ about elections in Canada, from byelections in the smallest provincial ridings to national campaigns. His data and aggregated polling is used by CBC in their coverage of the election.

Grenier of The Writ and Fournier of 338Canada are the number-crunching uber-nerds most people turn to for accurate forecasting in the election. They are fallible as all social scientists but their results speak for themselves.

In Grenier's Poll Tracker on CBC News, he looks at 11 pollsters numbers to come to aggregated conclusions. Each individual polls is different from the others.

For instance, Mainstreet Research's April 7-9, 2025 poll that found that if an election were held between those days, the Liberals would be projected to win 46 per cent of the popular vote compared to 39 per cent for the Conservatives, and just six per cent for the NDP.

Don't like that? Liaison Strategies' poll over those same days found different projections: Liberals at 43 per cent, Conservatives at 40 per cent, with eight per cent for the NDP.

In the graphic below, you can see those two plus nine other polls over about that same timeframe showing Liberal numbers in between that 46 as low as 42, and Conservative projections from 41 down to 36. The largest gap of any one poll is seven percentage points, the smallest is three.

All told, Grenier's aggregate polling results for April 10, 2025 comes up with a Liberal-Conservative vote projection of 44-37.

Fournier's aggregate polling for April 10, 2025 comes up with a Liberal-Conservative vote projection of 44-37.

Like or hate it, you can't argue with it.

Tory truthers

At a Conservative Party rally in Brampton, Ontario, last night (April 9, 2025), some Pierre Poilievre supporters donned sweatshirts that said "Do you believe the polls?" implying the media is lying and out to get the party. This MAGA-style attack on democracy is something that even Conservative pundits criticized.

Ken Boessenkool, one of Canada’s leading conservative strategists, Tweeted an image of the dumbassed conspiracy theorists at the Poilievre rally telling them to stop.

"We have rules," he said. "We accept outcomes based on following those rules."

Globe and Mail Ottawa bureau chief Robert Fife called it a brewing internal party controversy as it wasn't just low level supporters. Conservative campaign manager Jenni Bryne said they are "winning, don’t believe the polls, we have so many lawn signs and big rallies."

These are the same people who publicly celebrated the party's 25 per cent lead in these same polls a few months ago.

The Conservatives are, of course, not currently winning and it serves no purpose to deny this reality. The election is 18 days and a lot can and probably will change.

Attacking the media and academia and denying hard cold numbers is just stupid.

"It would have been nice to get through this campaign without a bout of Poll Trutherism," Grenier wrote in his newsletter today. "Alas, here we are."

Some people point to how the Kamala Harris-Donald Trump polling numbers were wrong. But margins of error exist, and when they were as small as they were in the U.S. election, that can happen. Grenier says that if the polls in this election were as off as they were in the U.S. campaign, the Liberals will drop from a. currently projected popular vote marging of victory of 6.6 points to 4.1 points.

"According to my model, shifting the margin by that much between the Liberals and Conservatives reduces the chance of a Liberal victory from 98 per cent to 90 per cent."

In the last two elections, the net error in projected margin by Grenier was 1.6 per cent, a number that wouldn't affect the results in Canada in 2025.

Poll denialism, however, isn't always about believing in lies, it's about bolstering support for party workers. Keep up your chin. Stay positive. Makes sense.

But if the party actually believes the media is lying and the polls are wrong internally, they are in trouble.

"It’ll be hard enough for the Conservatives to win a campaign that they are currently poised to lose with 2.5 weeks to go," Grenier says. "It’ll be nearly impossible if the Conservative brain trust believes they aren’t losing at all."

-30-

Paul J. Henderson
pauljhenderson@gmail.com

facebook.com/PaulJHendersonJournalist
instagram.com/wordsarehard_pjh
x.com/PeeJayAitch
wordsarehard-pjh.bsky.social

You’ve successfully subscribed to Paul J. Henderson
Welcome back! You’ve successfully signed in.
Great! You’ve successfully signed up.
Success! Your email is updated.
Your link has expired
Success! Check your email for magic link to sign-in.