The way the US counts votes is unique. In fact, there is no one central vote-counting system but rather tens of thousands of them in local precincts all across the country.
To help determine who has won where, the Guardian and many other newsrooms rely on the Associated Press, which has been calling US elections since 1848. (Some television networks, such as CNN, use their own analysis of results to make race calls.)
“Two years after the AP was founded, in 1848, we decided that there needed to be a trustworthy, nonpartisan source of information about who the country had elected as its new leaders,” David Scott, vice-president and head of news strategy and operations at the AP, told the Guardian.
The AP’s process for monitoring election results has undergone a few updates in the past 176 years. This year, the AP will rely on 4,000 reporters to report vote totals.
How is the AP able to predict who has won when votes remain to be counted? What is a “red mirage” and a “blue shift”? Read more of the Guardian’s interview with Scott, which has been edited for clarity and brevity:
What are the key sources of information that the AP Decision Team relies on to make its calls?
This general election, we’ll have roughly 4,000 vote-count reporters who are in county election offices, city and town election offices, Louisiana parish election offices … They’re calling into a group of about 800 people who make up our vote entry team, which is people receiving and verifying those numbers. And then there’s a whole separate quality assurance team who makes sure we’re as confident as we can be that the results are accurate.
Some states and some counties have direct feeds where we can get the votes as they are reported. Others post their results on websites, and sometimes we’re able to scrape them automatically. One of my favorites is, there’s a county out there that takes a picture of their vote-count results written on to a whiteboard and posts a picture to Facebook, so we have to go get them that way.
That’s so funny about the whiteboard. Even the reporters literally calling in the results feels pretty old school.
Yeah, I often get asked: “Well, why don’t you guys just have an app for that?” We actually like talking to those vote-count reporters. It gives us an opportunity, when someone calls a number in to say: “Hey, our system showed that you just reported there were more votes for this candidate than there are registered voters in that county. Can you go back and double-check that?”
We’re able to just do that verification in real time, and it helps avoid fat-finger mistakes. So it is a little analog, but we feel it’s the process that still works the best for us.
One question that comes up every election cycle is: how can the AP know the winner of a race when there are still so many votes left to be counted?
The question we ask ourselves before we declare a winner is: is there any chance the trailing candidate can catch the leader? If there’s a chance of that, then we don’t call it. We wait.
Sometimes it’s really obvious. People have very large leads, a lead so large that there’s not enough votes left to be counted to to make it up, and that allows us to move forward. Other times, races are very closely contested, and there’s just no more votes to count.
And how does your team account for the “red mirage” or “blue shift”, when the first batch of reported votes might skew more toward one party?
It’s probably best to illustrate this with an example. In a state like Virginia, generally the votes that are reported first will come from the more rural areas of the state, and those votes tend to favor Republicans. The votes that come last tend to come from northern Virginia and the suburbs of Washington DC, and those votes tend to favor Democrats.
That is absolutely something that we account for when thinking about whether a winner can be declared. States are unique, and counties are different within states, so we have to account for that geographic diversity.
Why are some places slower than others to report results? Could you give an example?
Florida, after 2000, really remade their election system, and it’s very fast. It’s very efficient across all of their counties, and so they count really quickly, and that was something that they prioritized. California prioritizes maximum participation. They allow mail votes to arrive well after election day, as long as they’re postmarked by election day, and they send everyone a mail ballot.
Both of those choices are completely valid, but those choices have an impact on how quickly we know the results in both states.
In 2020, we were waiting on the results in Pennsylvania for days. Do you anticipate that happening again, or is there another state that could lag in counting this year?
I do not make political predictions. All I’ll say is, how long it takes us to know who the next president will be is entirely dependent on how close the race is, and that’s up to the voters.
