Ever wonder what researchers think about when they are working to predict election outcomes?
The team at Nanos has been at this for over 20 years when back in 2004 we did our first nightly tracking project for the Canadian federal election. Sharing nightly election tracking in partnership with Canada’s leading news organization is one of our signature capabilities. Political work is the smallest part of our practice, but we enjoy the challenge.
Some unknowns on the eve of the election:
What will be the voter turnout? High voter turnouts usually help challengers and low voter turnouts help incumbents.
How good are the Get-Out-The-Vote machines for the parties? Researcher can measure opinion but cannot measure the on-the-ground effectiveness of the campaign teams. Strong campaigns punch above their weight. Weak campaigns don’t deliver the political capital they have built.
All said, here is our prediction for the 2025 Ontario election. This prediction relates to the popular support of the parties and not the number of seats or who will form the opposition. During the course of the campaign, we did a three-day rolling average (300 interviews a day) rolled up to 900 interviews for CTV News. In the last two days before election day, we doubled the sample to 600 interviews a day. We are showing our two-day average and our one-day results of the interviews just completed at about 2100.
Our election call is based on our one-day sample because that is the wave of research closest to election day.
-Nik Nanos, Chief Data Scientist
The CTV News/Nanos nightly provincial election tracking conducted by Nanos Research surveys 900 Ontarians aged 18 years and over three days (300 interviews each day). Respondents are all randomly recruited through a dual-frame (cell- and land-line) RDD sample using live agents. One half of the sample is administered the questionnaire by telephone and one half are administered the same questionnaire online. The random sample may be weighted by age and gender according to the latest Canadian census data. Throughout the election, the interviews are compiled into a three-night rolling average of 900 interviews, with the oldest group of 300 interviews being replaced by a new group of 300 each evening. Note: On February 25th and 26th 600 interviews were conducted each night. The current data covers the two-night period ending February 26th, 2025.
A random survey of 1,246 Ontarians is accurate 2.8 percentage points, plus or minus, 19 times out of 20.
The full methodology is detailed in the technical note in this report. This research was conducted and released in accordance with the standards of the CRIC of which the firm is an accredited member.
Full data tables with weighted and unweighted number of interviews is here.
Note: Charts may not add up to 100 due to rounding.
To read the full report click here.