No worries. We appreciate the constructive feedback.

Thanks for this helpful call-out, Jonathan. We've revised the survey to remove the reference to Reuters as it clearly does not belong in this category. We are discussing whether to include a separate category of news agencies, wires, or press associations in future studies.

The extent to which the poll aggregators produced accurate forecasts reflects the rigour and transparency of the sampling methodologies employed by most pollsters, regardless of the actual results of any individual poll.

Polling is a difficult and complex exercise—increasingly so in the digital age where modes of communication are so varied. Overall the polls were relatively accurate in predicting the election outcome. Even the larger error rates mostly fell within margins of error.

But in terms of correctly forecasting individual seats, so far @338canada.bsky.social is in the lead at 95.1% accuracy for seats not designated as toss-ups, folllowed by @voxpoplabs.bsky.social at 91.9% and @ericgrenier.bsky.social at 90.7%.

These numbers will shift as toss-ups are factored in.

While all three major poll aggregators—The Signal included—overestimated the Liberal seat share and underestimated the Conservative seat share, The Signal is once again has the lowest error.

In terms of vote share projections, @mainstresearch.bsky.social, Liaison Strategies, and @angusreidinstitute.bsky.social came came closest to the mark—as did The Signal from @voxpoplabs.bsky.social, which had the lowest error of the poll aggregators.

MAE tells you how much polls missed by, on average. RMSE accounts for bigger misses that skew overall performance.

The lower these scores, the more accurate the prediction.

A perfect poll would have an MAE and RMSE of zero.

For each major polling firm and aggregator, we calculated the two most common measures of polling error, Mean Absolute Error (MAE) and Root Mean Squared Error (RMSE).

Canadian pollsters by and large correctly predicted that the Liberal Party would win the most votes. But while most polls had relatively similar vote share projections, even marginal differences can impact outcomes in the Canadian electoral system.

For more details on our projections visit www.thestar.com/interactives...

In a few hours we will have an opportunity to see how well we've all fared in trying to model the election outcome.

But it is interesting that, despite employing vastly different methodologies, we all arrived at very similar conclusions this time around.

And it's not just on seat projections. Vote share estimates are even closer.

And it's not just on seat projections. Vote share estimates are even closer.

And it's not just on seat projections. Vote share estimates are even closer.