As Fishers and Hamilton County voters head into Tuesday’s midterm election, national forecasts point to a strong night for Democrats. One of the country’s best-known forecasters says that outcome is likely, but far from certain.

Nate Silver, founder of FiveThirtyEight and now author of the Silver Bulletin newsletter, wrote last week that Democratic leads in several key U.S. Senate races are “all just a normal-sized polling error away from flipping.”

Silver’s model shows Democrats ahead by only 2.4 points in Maine, 2.5 points in Texas, 3 points in Ohio and 3.6 points in Alaska. Those margins are well within the range by which polls have missed in recent elections.

His central worry is a polling error that runs the same way in every race. If the polls are off by 3 or 4 points toward Republicans across the board, Silver said, most of those Democratic Senate leads would disappear at once. He has also noted that recent polling misses have clustered in the Midwest.

Hoosiers have reason to remember how badly polls can miss. In 2016, most major forecasts gave Hillary Clinton a strong chance of winning the presidency. FiveThirtyEight put her odds at about 71%, and The New York Times’ Upshot model put them at about 85%. Donald Trump won.

In 2024, polls described the presidential race as a toss-up. Trump swept all seven battleground states and won the Electoral College 312-226. Next door in Iowa, a final poll from the respected Des Moines Register pollster Ann Selzer showed Kamala Harris ahead by 3 points. Trump carried the state by about 13.

Silver is careful to say the error could run the other way, too. He said he could imagine Republican turnout collapsing and giving Democrats an even bigger win. He also points out that polls in the last two midterm elections, 2018 and 2022, were fairly accurate.

Still, his overall forecast favors Democrats. The House remains his strongest call. If Democrats were to lose it, he said, there would be serious questions about what is happening with the polling.

The lesson for voters is the one that never changes. Forecasts describe probabilities, not outcomes, and a 70% chance means the other result happens nearly one time in three.

The only count that matters begins when the polls close Tuesday night.