2022 Ontario Forecast

by Eliza Pallas

Progressive Conservative Party (Doug Ford) – 73

Liberal Party (Stephen Del Duca) – 25

New Democratic Party (Andrea Horwath) – 24

Green Party (Mike Schreiner) – 1

Ontario Party (Derek Sloan) – 1

New Blue Party (Jim Karahalios) – 0

63 seats needed to form a majority

Last Updated: June 1, 2022

Based on a weighted average of the latest polling, Canada Compass predicts that the governing Progressive Conservative Party will win a majority with 73 seats, while the formerly third-placed Liberal Party will form the Official Opposition with 25 seats, the New Democratic Party will fall into third place with 24 seats, the Green Party will win 1 seat, and the Ontario Party will win their 1st seat. 63 seats are needed to form a majority in the Legislative Assembly of Ontario, and because the Progressive Conservatives have a 76% chance of winning the tipping point riding for a majority, the race is Lean PC Majority.

Is this accurate? How is it made?

Canada Compass uses a polling model created by Eliza Pallas, who hails from Victoria, British Columbia, Canada. She covers Canadian politics as a contributor for Congress Compass.

The Canada Compass parliamentary model would have accurately forecast the winner of the last three general elections in Ontario. In 2018 the model would have correctly predicted that Doug Ford’s Progressive Conservatives would win a majority. In 2014 the model would have correctly predicted that Kathleen Wynne’s Liberals would win re-election, and in 2011 the model would have correctly predicted that Dalton McGuinty’s Liberals would lose their majority but still win a minority.

The average districtwide error across the 2011, 2014, and 2018 general elections is 4.0%, and 284 out of 338 districts (84.0%) would have been called correctly. Out of the 36 misses, just 16 were outside the standard error, only 4.7% of the total races.

The Canada Compass model is a weighted polling average that gives more weight to polls if they were conducted closer to the election, or have a larger sample size, using regional breakdowns to estimate the swing by riding.