Election night in Maine: When will we know? | Opinion
Read all of our coverage of Maine's 2026 election or stay up to date with the latest developments through the Maine Political Report in your inbox or text messages from politics editor Kirby Wilson.
Philip E. Mathieu is a data scientist at IDEXX Laboratories. He lectures part time at the Roux Institute and volunteers on the Executive Committee of Sierra Club Maine.
When will it be over?
For some of us, this question is almost as pressing as “who will win” the race for Maine’s Senate seat next week. But barring a blowout or a significant change in reporting practices, most major media outlets and the Associated Press probably won’t call the race until early Wednesday morning or later.
The logic is defensible: in Maine, many towns will not finish counting until Wednesday, so calling the race early risks a late “swing” as the final votes trickle in. Additionally, by nature of ranked-choice voting, a candidate has to both beat their opponent and clear the 50% threshold to “officially” win on election night, meaning that any close race is all but guaranteed to not get an official call until the following week.
For me, this uncertainty is a guarantee of anxiety on election night. So, as a data scientist trained at the Roux Institute and working full time for IDEXX, I decided to put my skills to work.
Drawing inspiration from the New York Times’ and Washington Post’s election models, I built a model that forecasts not who the winner is (the data says it’s a tossup) but when we’ll know with reasonable certainty the outcome, either way.
Advertisement
The model runs 3,000 simulations based on Maine general and primary election data from 2018-2026. For each town, it randomly selects a time for the town to report results based on previous reporting times. It also randomly assigns an actual percent for each candidate, based on current polling plus wide margins of error. This gives us 3,000 predictions spread across a range of possible results and a range of possible reporting times, which we can then use to figure out when we would know based on the final margin.
Click to expand.
In about 7 out of 10 simulations, either Jackson or Collins wins by at least 2 points, and the model will be confident — 99.5% chance or better of the result not changing — by 11 p.m. On most of those nights, media outlets still won’t call the race until a few hours later, because they wait for more of the vote to be counted.
On close nights, when we don’t have a confident prediction by 11 p.m., 6 out of 10 simulations don’t reach a confident result until Wednesday midday or, for the closest races, until ranked-choice tabulation the following week. If it’s still too close to call by my model at 11 p.m., I’m going to bed.
What about the governor’s race, you ask? The answer there is a bit simpler. Because current polling shows Hannah Pingree with a significant lead, 9 out of 10 simulations favor a called race before 11 p.m. Additionally, there’s no ranked choice tabulation in that race, so chances are we’ll know by Wednesday morning at the latest, unless it is so close that it goes to a recount.
To explore more of this analysis and some other experiments looking at election modeling in Maine, visit philipmathieu.com/election-night-2026.
Copy the Story Link
Tagged: columns, election 2026
