How predictive are the midnight votes of six people?

While you were either asleep (East Coasters, Europeans) or still experiencing Monday (West Coasters, night owls), six people showed up at a hotel in a tiny town in northern New Hampshire and voted to make former ambassador Nikki Haley the Republican presidential nominee in 2024.

Uh, okay?, you’re probably thinking. And that is fair. Six votes in nearly any election — save, say, middle school class president — is not impressively significant. But those votes, offered in Dixville Notch, N.H., invariably attract outsized attention every four years because, one, they are the first votes cast in New Hampshire and, two, that gives something for the press to do.

And the candidates. That Haley swept those six votes led her to make a bold prediction, given former president Donald Trump’s solid position in pre-primary polling: “Do you think if this was a done deal we would have gotten any of them?”

In fact, a review of past voting in Dixville Notch shows that the town’s votes mirrored the results in New Hampshire overall more often than not. In the past two contests, though, its results have not been predictive of the eventual nominee.

Let’s first establish that your original response (if I correctly predicted it) was on the mark. Relative to total turnout in either party’s primary since 1964, the votes cast in Dixville Notch have, at best, constituted 0.03 percent of the total votes cast. (This was in the Republican primary in 1984, when President Ronald Reagan won every primary contest in a walk.) That percentage, 0.03 percent, is equivalent to 30 cents out of $1,000. It is about three-quarters of the area of Rhode Island compared to the rest of the continental United States.

Nor have the results in Dixville Notch correlated strongly to the statewide vote. For each contested primary over the past 60 years, I pulled data in both the town and the state and compared the percentages of support earned by the eventual nominee. (I chose this instead of comparing the town’s results to the winner of New Hampshire because this allows us to simultaneously evaluate the predictive power of both town and state.)

The short version: The percentage of total support the eventual nominee earned in Dixville Notch didn’t correlate strongly to the percentage statewide. So if a candidate got 30 percent in the town, that did not suggest it was likely they’d also get 30 percent statewide.

Of course, percentages aren’t a great point of comparison here, given that the median number of voters who’ve cast ballots in Dixville Notch over those decades is seven — meaning that half of the primaries had more voters and half had fewer. If the statewide result is 50-50, and 4 of 7 voters in Dixville Notch supported the eventual nominee, that’s a seven-point deviation from statewide (since 4 in 7 is 57 percent).

So I looked at it more simply. How often had the nominee’s result in Dixville Notch matched the result statewide? Because I had the numbers, I added another category, too: the eventual nominee earning a plurality of support or (in Dixville Notch) tying someone else.

Out of the 28 contests considered, Dixville Notch and New Hampshire both had the eventual nominee losing six times. The eventual nominee won both seven times. An additional eight times, the eventual nominee won a plurality, tied, or won in both the town and the state.

In other words, three-quarters of the time, the eventual nominee did the same in both the town and the state. But only 15 of 28 times did the eventual nominee win (or ties) in both the town and the state. Meaning that Dixville Notch’s ability to predict both the state and the nominee is the equivalent of a coin toss. As you’d expect.

Over the past four contested primaries, Dixville Notch’s results are not that great. In 2012, the eventual nominee tied in Dixville Notch with two votes. In 2016, it missed both eventual nominees. In 2020, the eventual nominee — Joe Biden — received zero of the four votes cast.

That is probably worth remembering as the rest of the vote in the state unfolds.

I made a complicated chart showing the support in both the town and state in primaries since 1964. It is complicated and would take too much time to explain but it also took a long time to make, so I am going to drop it here and then brush off my hands and walk away.

*brushes off hands*

*walks away*

ncG1vNJzZmivp6x7uK3SoaCnn6Sku7G70q1lnKedZL2wuMitoJyrX2d9c4COaWhoamNksarE1aKjpZ1doLuwxIynnLBlmJa6sb%2FHoqmeZaCntq6t0bJm