Now that the dust has settled after the Ontario provincial election, let's take a look at how Eyesover fared with our popular vote forecast.
We had forecast 38% PC, 35% NDP, 23% Liberal, and 4% Green. The actual results ended up at 40% (+2) PC, 34% (-1) NDP, 20% (-3) Liberal, and 5% (+1) Green.
How did we do compared to traditional polls? We won’t get into comparing numbers here, but a quick analysis of the final polls included in CBC's Poll Tracker shows Eyesover's results were definitely on par.
Just as we found in the 2016 US Election and the 2017 Alabama Special Senate Election, our methodology produces accurate public opinion data in real-time and at a fraction of the cost of traditional methods.
As the people of Ontario head to the polls tomorrow, there is a good chance that the results they are expecting are vastly different than those they predicted a month ago.
In late April, the PCs had a significant lead, and the NDP were trailing behind the Liberals in third place. But as we saw throughout the past month, there is a good reason why campaigns are held.
For our final projection, we still see the PCs in the lead with 38% of the vote, but the NDP are a close second at 35%. The Liberals have faded to third with 23% of the public's support, and the Greens hold 4%.
Our data analysis shows that with the exception of wage and education issues, little policy caught the public's attention as discussions about candidate actions or character dominated the campaign throughout. For example, comments regarding the party platforms were minuscule compared to the discussions that news of a lawsuit against Doug Ford generated.
Thursday will provide the final answers, and we'll be watching closely!
Eyesover Technologies’ analysis of the Ontario Provincial Election consists of continuously measuring mentions of each candidate and relevant issues from a daily average of approximately 20,000 comments from Twitter, Facebook, YouTube and Reddit, and translating those mentions into voter support with our proprietary algorithms and would not be considered an election survey based on recognized statistical methods.