Learn
Short, plain-English explainers on how to read the numbers we publish. No statistics background required. Election data is full of small leads, noisy polls, and labels that sound more certain than they are — these pages explain what the numbers can and can't tell you.
What an average is (and isn’t), why it moves, and how to think about margin of error.
Safe, Likely, Lean, and Toss-Up in plain English — and why a rating is a description, not a prediction.
Mode effects, likely-voter screens, weighting choices, house effects, and timing — why two good polls can show different numbers.
Why Decision Labs doesn’t publish win probabilities, what the simulator is for, and how to read a two-point lead.
Why some primaries don’t end on election night — runoffs and top-two primaries, and what “Runoff pending” means on a race page.
Go Deeper
These pages cover the same ground with the full detail — exact weights, formulas, and the standards we hold ourselves to.
- Polling Methodology — The exact formula behind our averages and ratings — every weight and constant.
- Race Ratings — Our live competitiveness ratings for every 2026 Senate and Governor race.
- Editorial Standards — How we stay nonpartisan, how we report race calls, and how we source data.