Road to the Majority

How to read this page

What the model is saying

The probabilities shown here are posterior probabilities from a Bayesian MCMC model — meaning they already incorporate both structural fundamentals (incumbency, historical voting patterns, national environment) and current polling. When the model says Democrats have a 52% chance of winning a given seat, that is its full updated belief given all available evidence, not a starting point.

Marginal vs. conditional probability

Two numbers on this page look like they should be related but often aren't: the number of control races where the model gives Democrats an edge, and the number they typically carry in simulations where they win the majority.

Marginal probability is the model's estimate for a single race in isolation. Counting seats where Democrats are individually favored gives you a tally of those marginal probabilities above 50%.

Conditional probability is the probability of something given that a specific outcome has already occurred. Each victory of course increases the winner's chance of winning the overall majority, but the impact of each race on the conditional probability of winning the majority is not equal: Republican wins in races where Democrats are favored, for example, would indicate a national environment more favorable to Republicans than the headline forecast, and the corresponding increase to their conditional probability of a majority is higher if they win in these races.

What the impact numbers mean

Each control race shows two numbers alongside it:

    - D majority ↑ if D wins — how much the Democratic majority probability rises above its baseline if Democrats win this seat

    - R majority ↑ if R wins — how much the Republican majority probability rises above its baseline if Republicans win this seat

The combined figure is the sum of both — the total stakes for majority control riding on this single race. A district where both numbers are large is genuinely high-stakes for both parties. A district where only one is large has an asymmetric dynamic: one party's majority depends on it far more than the other's.

Districts are ranked by an internal impact score that captures how correlated each race's outcome is with overall majority control — symmetric across parties by construction.

Control races

Control races are districts that are both genuinely competitive (35–65% Democratic win probability in the posterior) and have high majority impact. These are the seats most worth following on election night — competitive enough that either outcome is plausible, and consequential enough that their results will tell you something meaningful about who controls the House.