Methodology

Where is our data from?

Our election night (and thereafter) results data is from the AP Elections API, which is generally considered to be the best source for election night results. This feed is used by most major news sources, including broadcast media and the New York Times.

The AP feed provides data on a county level for state-wide and Presidential elections, and on a district level for House elections.

The speed of data is very dependent on how quickly the relevant localities count their votes and agree to report them out. It’s very possible that at various points in the night that the counts reported on our dashboard will be ahead or behind other sources, including county webpages and secretary of state webpages. This is to be expected, and results will generally align themselves once more time has passed.

What about previous years results?

These results are from the MIT Election Lab, gathered from official results as reported in 2016/2018 and so on. Voters registered have been pulled from state websites.

Where is the early vote data from?

This data is reported to parties, campaigns, and other entities by states. We’re using a file compiled by TargetSmart, rolled up to geographies. As with all data, this is highly dependent on how the state decides to report, and will be more or less up to date by state. We’re using the best estimates we could work out, but we don’t expect this to be a perfect counting of votes cast early, particularly because ballots mailed as “early” may arrive on or after election day and still be counted, depending on local rules.

Why doesn’t early vote data show the actual results?

We don’t get records of the candidates that early votes are for, just records of the number of ballots received. The actual outcomes of these votes will be counted and released on election night. Usually, early vote results aren’t broken out from night-of results, so there’s no good way to know who early voters cast ballots for.

Will you ever vary from the AP results?

If there’s some real issue with AP results, we’re leaving open the possibility of using results gathered from other places. These will be marked as such, so there should be no confusion. This will also be a last resort if there’s some serious issue with the results.