Because this election is seeing so much higher turnout than usual, this is not accurate.
The 338 model ranks polling by how well pollsters predicted the last elections, which had low turnout. Those pollster ranks weigh the current seat projections for this election.
With higher turnout this year that means previously unmotivated voters will be turning out and the current models don’t know how they are represented in polls.
For example we could see suppression of NDP voters and an increase in Gen Z conservatives voting, causing a sharp rightward swing vs forecasts. We also could see traditionally less active voters throw this election further left.
My point is a 338/538 style poll aggregation will be less accurate this year than prior years. DO NOT take 338 as a given.
I expect an outcome more in the tails of the distribution, meaning a swing even more conservative or liberal than even current forecasts.
Because this election is seeing so much higher turnout than usual, this is not accurate.
The 338 model ranks polling by how well pollsters predicted the last elections, which had low turnout. Those pollster ranks weigh the current seat projections for this election.
With higher turnout this year that means previously unmotivated voters will be turning out and the current models don’t know how they are represented in polls.
For example we could see suppression of NDP voters and an increase in Gen Z conservatives voting, causing a sharp rightward swing vs forecasts. We also could see traditionally less active voters throw this election further left.
My point is a 338/538 style poll aggregation will be less accurate this year than prior years. DO NOT take 338 as a given.
I expect an outcome more in the tails of the distribution, meaning a swing even more conservative or liberal than even current forecasts.