By Tom Parkin

April 11th, 2025

BURLINGTON, ON

Voters were told to expect a significantly bigger PC majority and half of NDP MPPs defeated. Actually, each party shifted by just one seat.

The difference between the actual election results that returned Doug Ford to office and what the pollsters predicted raised more than a few eyebrows. One has to wonder if the same thing is happening with the polling being done in the current federal election.

Doctors bury their mistakes. Architects cover them with ivy. But it seems those running poll projection sites just move on to the next.

Philippe Fournier’s 338 Canada, Eric Grenier’s The Writ and Quito Maggi’s Mainstreet all got it badly wrong in the 2025 Ontario election, influencing voter behaviours and prejudicing electoral outcomes with seat projections showing the Ontario NDP would lose about half its seats and the PCs would significantly increase their majority.

In fact, the Ontario PCs came into the election with 79 MPPs and came out with 80. The NDP had 28 going in and elected 27.

Spreading bad information is prejudicial to elections

It’s impossible to know how many voters switched their vote based on inaccurate projections. Or decided not to bother voting because the PCs had it in bag, anyway.

But it’s reasonable to believe that had seat projection sites given voters accurate information, Ontario turnout would have been higher, the PC win would have been tighter and the NDP would have done better.

Seat projections are modelled on past geographic vote distribution patterns. But in the Ontario election, NDP votes shifted to become more effectively concentrated in NDP incumbent seats. And that’s very likely exactly what’s happening to federal NDP votes right now. In fact, distribution shifts happens every election.

Margins of error that would be career-ending for a pollster

If a pollster reported a party was at 13 per cent the day before 27 per cent of voters supported them, there’d be a lot of explaining to do. And some amazing Twitter fights. Media platforms would stop contracting them and reporting their results.

But not seat projection sites. On the Ontario NDP, Mainstreet’s margin of error was 80.0 per cent, The Writ was off 107.6 per cent, and 338 Canada was off 68.8 per cent. And they’re all back again.

And this time the Toronto Star has even added a new company to the mix, Vox Populi, the people who did that bizarre on-line consultation for the federal Liberal government in the wake of Justin Trudeau breaking his electoral reform promise.

Until people start calling out bad data for hurting elections, don’t expect change, but do expect some election night surprises.