To help you stay sane over the next few weeks until Election Day, here are some tips: Don’t care about the polls. Unless you work in politics or like to bet, you’re probably looking at them to find out who will win, just like the rest of us. To know who will win, or at least to think you do. They can’t tell you that, though.
Back in 2016, Harry Enten, who was working at FiveThirtyEight at the time, figured out the final polling error for every white house race from 1968 to 2012. Most of the time, the polls were off by two percentage points.
An analysis done by the American Association for Public Opinion Research in 2016 showed that national polls were off by an average of 2.2 points, while state polls were off by an average of 5.1 points. The national polls were off by 4.5 points in 2020, and the state polls were off by 5.1 points as well.
You could picture a world where these mistakes happen by chance and cancel each other out. Maybe the number of people who back Donald Trump is three points too low in Michigan and three points too high in Wisconsin.
However, mistakes often favor one candidate over the other. In 2016 and 2020, for example, state-level polls often didn’t count enough Trump voters. When it came to 2016 and 2020, the polls were off by 3 points for Hillary Clinton and 4.3 points for Joe Biden.
If the election is a blowout, a mistake of a few points in either way doesn’t matter. Depending on the poll, Adam Schiff, a Democrat, is ahead of Steve Garvey, a Republican, by 17 to 33 points in the race for the California Senate. A mistake in the polls of up to 10 points wouldn’t change the result of the race.
Too close to call
However, that is not where the voting for president takes place. The New York Times polls average showed that Kamala Harris was three points ahead of Trump across the country as of Thursday. Of those eight states, only seven are split states. In those seven, neither candidate has a lead of more than two points.
Let’s say that the polls do better in 2024 than they did in 2016 or 2020: In the split states, they’re off by an amazing two points. Cheers! Harris should win every split state if that holds true. It also makes sense that Trump would win every swing state.
It’s not impossible for this to happen. Nate Silver’s election model says that “Harris sweeping all seven swing states” is the most likely result. And Trump winning all seven is the next most likely outcome.
In other words, the polls can’t tell you how or by how much they’ll be wrong. Now, though, that’s what counts.
A race this close is great for people who like to think about how polls are run. Last week, my coworker Nate Cohn wrote an interesting piece about how polls are different depending on whether they are “weighting on recalled vote” or not. In short, this is what it means: Pollsters want to avoid making the same mistakes they did in 2016 and 2020, when they didn’t count enough Trump voters.
Most pollsters are now asking people to remember who they voted for in 2020. They then use that information to make sure that all Trump votes are taken into account. When pollsters do that, they get numbers that look a lot like how people voted in 2020. Pollsters who don’t do that are getting very different answers.
It is known that people are bad at remembering past votes, though. Using that information might make polls even less accurate. We should know which analytical choice was right in about a month.
If you’re not a professional pollster, though, do you really need to waste your precious time on earth thinking about how to weight canceled votes until then? Tell the people you love that you love them on the phone. Take ten deep breathes and notice where your mind goes. Just about anything else.
A very stable voter
Another reason to ignore daily polling news: The polls are oddly stable.
A week before the September Harris-Trump debate, Harris lead Trump by 3 points. Trump then gave the second-worst debate performance in recent history. The July campaign rally shooting prompted another Trump assassination attempt. Then the Fed cut rates 50 basis points. Israel invaded Lebanon on land.
The vice presidential debate followed. A surprising good jobs report followed. Harris published an 82-page policy booklet, and Jack Smith, the special counsel pursuing Trump in the Jan. 6 case, filed a 165-page brief detailing Trump’s efforts to overturn the 2020 election. Harris leads Trump by …3 points after all this.
Undecided voters are usually those who pay less attention to political news and are either so uninterested in politics or so cynical about both candidates that nothing has swayed them. Many more voters have decided but may not vote by Election Day. These voters will decide the election and aren’t deciding yet.
This column is probably not read by those voters. Take a rest. Get off the emotional rollercoaster. If you want to influence the election, donate money or time in a swing state, preferably to a state party or down-ballot contest, or volunteer locally.
Call everyone you know who may be uncertain, unregistered, or unable to vote. Then release. Nothing more can be done by you or the polls.