If the polls just closed, how can AP already declare a winner?
Control of the White House and the narrowly divided Senate and House may hang in the balance on Nov. 5, but don’t expect election night to be all nail-biters and barnburners. Chances are some winners will be declared before a single vote has been tallied, just as the polls close in those states.
So what gives?
Although determining a winner before any votes have been counted may seem counterintuitive, race calls at poll closing time in uncontested or landslide races have been a routine part of election nights for decades, even though competitive, hotly contested races that can take hours, days or even weeks to decide tend to be the most memorable and attract the most attention.
The Associated Press will consider multiple factors and analyze available data before determining whether a winner can be declared when polls close in a given state. But the AP will never declare the outcome in a competitive contest before enough votes are counted to make the winner clear.
Uncontested races
Many of the races called just as the polls close are uncontested elections where only one candidate appears on the ballot and is therefore the only possible winner of the race. Voters in some parts of the country live in multimember districts for offices such as state legislature, where more than one candidate is elected in a district. In those districts, an uncontested race is one where the number of candidates on the ballot is equal to or less than the number of seats available in that district.
In the 2024 general election, the AP will declare winners in nearly 2,000 uncontested races, compared with about 4,500 contested races.
Noncompetitive contests
Sometimes it’s possible to declare winners at poll closing time in noncompetitive races with multiple candidates in areas where one political party has an established history of lopsided victories. In these cases, the AP will analyze multiple sources of available data, including the results of AP VoteCast, a comprehensive survey of both voters and nonvoters that determines who voted, how they voted and why, to confirm the outcome.
The AP will not call a race when polls close if AP VoteCast’s results indicate a deviation from the state’s long-standing political trends and voting history. AP VoteCast results will be available for all 50 states, though only a small number will be considered as potential poll close calls. There is no AP VoteCast survey in the District of Columbia, so no contest there will be called when polls close even though the nation’s capital has a long history of overwhelming victories for Democratic candidates.
For example, the AP’s poll close calls in the 2020 presidential election included Wyoming, a state that last voted for a Democrat in 1968 and that Donald Trump won over Joe Biden by 44 points; and Massachusetts, which last voted for a Republican in 1984 and that Biden won by 34 points over Trump.
A handful of states and districts have multiple poll closing times because they are in more than one time zone. In these cases, the AP will not declare a winner before the final poll closing time in that state or district. Florida, Texas and some others begin to release vote results from most of the state shortly after polls close in the earlier time zone. Votes that are already counted from areas in the earlier time zones will also be considered in determining whether a winner can be declared at the moment when the last polls close.
Other election data the AP takes into account include an area’s voting history from recent elections, voter registration statistics and pre-Election Day polling.
When the above data points confirm the expected result in a state where either major party has a history of dominating elections, the AP may call the race as soon as voting ends.
A brief history of race calls
The AP was formed in 1846 as a newspaper cooperative. It tabulated election results for the first time two years later, when Zachary Taylor won the presidential election as a member of the Whig Party. The effort to gather the results from jurisdictions across the still-young nation relied on the telegraph, lasted 72 hours and had a then-exorbitant cost of $1,000.
In 1916, the first election broadcast aired over a small network of ham radios, according to a history written by the late CBS News Political Director Martin Plissner. The announcer closed the program by incorrectly declaring that Republican Charles Evans Hughes had won the presidency over Democrat Woodrow Wilson. The AP called the race for Wilson two days later once it was able to report results from California.
By the early 1960s, the AP and the three broadcast networks — ABC, CBS and NBC — were each conducting independent vote counts. They agreed to pool their resources in the 1964 election to compile the vote count for key races, an arrangement that would last in some form for more than 50 years and eventually expand to include exit polling of Election Day voters.
After the 2016 election, the AP left the network pool to continue its independent vote count operation and launch the AP VoteCast survey of the American electorate as an alternative to the network’s exit polls. The networks, now including CNN, remain with the pool today and receive their vote count and exit poll data from Edison Research. Fox News subscribes to AP’s vote count, as do thousands of news organizations across the United States and around the world, and partners with the AP to conduct the VoteCast survey.
What happened to ‘precincts reporting’?
It’s usually possible to determine who won an election before the vote counting is 100 percent complete. But doing so requires determining how many ballots still need to be tabulated at any point in the vote count, and that’s not as straightforward or clear-cut as you might imagine.
In fact, it’s only gotten more complicated as elections have evolved, making traditional methods of tracking vote count progress less reliable.
For years, the most common way news organizations kept track of the progress in the vote count on election night was to calculate the percentage of precincts that had reported their results. But over the years, that method has become a less and less reliable measurement that does not reflect the realities of modern elections.
For starters, precincts throughout a state are not necessarily equal in population. Saying that 10 percent of precincts have yet to report could mean different things depending on which precincts are at issue. In addition, sometimes a state or county will count a precinct as reporting even if it has only reported some of its votes.
An even bigger problem is that “precincts” traditionally refer to Election Day polling places. As voting by mail and early in-person voting has gained in popularity, the “precincts reporting” measurement could be deceiving depending on how the votes are tabulated from state to state.
Some states do not include pre-Election Day votes in the voters’ home precincts — that is, the ones where they would vote if they voted on Election Day — and opt instead to lump all of their pre-Election Day votes into one precinct. In some other states, “precincts reporting” only reflects votes cast in person on Election Day and excludes any mail or early in-person voting altogether.
These states may show that all or nearly all precincts have reported complete results, even though the actual number of ballots yet to be tabulated might be significantly larger. Alternately, they sometimes report large numbers of early and mail votes before any precincts are reported at all.
In some states where there’s relatively little voting before Election Day or where election officials strive to assign mail and early votes to their home precincts, the percentage of precincts reporting does still provide a reasonably accurate sense of the votes left to be counted.
But across the board, The Associated Press calculates an expected vote estimate to provide a more consistent and reliable measurement to track the progress of the vote count on Election Night.
Reporting the expected vote
Clearly, the ideal way to calculate the number of ballots that remain to be counted is to subtract the total votes counted so far from the total number of votes cast overall. But since the exact number of votes cast in an election isn’t immediately available on election night, the AP instead analyzes several key pieces of election data to estimate the total votes expected to be cast in any given contest.
This estimate is based on several factors, including the total votes cast in previous elections, changes in population, voter registration figures and the competitiveness of the current election. Turnout data on pre-Election Day voting has become increasingly helpful in determining the expected vote, as the share of voters casting early ballots has gone up.
Once voting and vote counting begins, the AP will evaluate the latest Election Day voter turnout figures reported by elections officials as well as actual vote returns to update and refine its expected vote estimates.
Over the course of election night, the AP will report for every contest the percentage of the total expected vote counted so far. This percentage should steadily grow as the number of votes counted gradually approaches the expected vote total. But because the expected vote estimate is continually monitored and updated as more data becomes available, it’s possible that the percentage of the vote counted may temporarily get smaller because the estimate of overall turnout got bigger.
For example, if elections officials initially tell the AP that Election Day turnout in their state was 1 million voters but later revise that number to 1.2 million, the percentage of votes counted will drop temporarily before going back up again as more votes are tabulated.
What’s so difficult about nailing down the total votes cast?
The main challenge is that it can take a while for the people running elections to figure out exactly how many people voted in any given election. It often takes local and state officials weeks to arrive at the final, official number.
That’s in part because of mail voting. A local elections office might know exactly how many mail ballots they sent out to voters and have a good estimate of how many have been returned in the days leading up to Election Day. In fact, before officials begin counting votes, it’s often easier to know the approximate number of mail ballots cast than the number of people who turned up on Election Day.
But they won’t know exactly how many of those voters actually cast a mail ballot until their deadline to receive them, which is usually on Election Day itself or even after that.
With the number of voters casting their ballots by mail each year steadily on the rise, determining the total number of mail ballots can take some time, especially if the deadline to receive them is after Election Day.
Mail voting spiked in the 2020 election and outnumbered voting at polling places on Election Day for the first time ever as voters sought alternatives to in-person voting during the COVID-19 pandemic. While the level of mail voting is expected to surpass pre-pandemic levels, in many areas, it may fall in comparison to the 2020 election, adding another complication to estimating the expected vote in this year’s elections.
