Science

Stop Saying the Election Was Rigged

Trump’s win was always an option, and the theories suggesting otherwise aren’t based on facts.

President-elect Donald Trump stands outside the clubhouse following his meeting with Peter Kirsanow, attorney and member of the U.S. Commission on Civil Rights, at Trump International Golf Club, November 20, 2016 in Bedminster Township, New Jersey.

President-elect Donald Trump at Trump International Golf Club in Bedminster Township, New Jersey on Sunday.

Drew Angerer/Getty Images

They say the five stages of grief aren’t actually real, but when it comes to the results of the 2016 presidential election, many Americans still seem stuck in denialism, the first stage. That emotion is probably what’s behind the rampant sharing of two postelection articles from Bill Palmer: “You’re Not Just Imagining It: The Hillary Clinton vs Donald Trump Vote Totals Do Look Rigged” and “Rigged Election: Hillary Clinton’s Early-Voting Lead in Florida Was Mathematically Insurmountable.”

Do the election results look rigged? My quick comment upon reading those posts is that I shouldn’t have to bother to explain why they don’t: It’s the job of the professional news media to follow up on these leads and report what’s worth reporting on. Since this particular claim hasn’t bubbled up to the legitimate news, I can take it as seriously as I’d take the National Enquirer’s claim a few months ago that Hillary Clinton’s weight had ballooned to 289 pounds. On the other hand, I’m part of the professional news media, so I guess it’s part of my job to take a look at such things, and I’ve done it before.

This new claim, I don’t buy. Why? For one thing, the details that are easy to check come out wrong. A key step of the argument of the Palmer posts is that in the primaries, Hillary Clinton outperformed the polls. Palmer writes:

In any given hotly contested primary state, Donald Trump tended to perform the same as, or worse than, his polling averages. We saw it in his very first contest in Iowa, where he shockingly lost despite being favored. We saw it again in Wisconsin and other states. In contrast, Hillary Clinton tended to perform about the same as, or better than, her polling averages in most states.

Actually, though, according to FiveThirtyEight’s page on the Wisconsin primaries, Clinton was at 47 percent in the polls, and Trump was at 36 percent. In the event, Clinton got 43 percent in the Democratic primary, and Trump got 35 percent among the Republicans. So, the actual facts don’t support the theory.

Later on, there’s a discussion of the differences between early votes and Election Day votes in Florida, presuming that Election Day results should have followed the trends that came out in early voting. But early voters are different from Election Day voters; we know this. Palmer doesn’t accept this, instead writing that “while that’s mathematically possible in the theoretical sense, it’s mathematically insurmountable in any real world scenario.” That’s simply untrue. Similarly, Palmer writes that “it is statistically suspicious that in every state where Donald Trump pulled off an upset, he won it by right around 1 percent, just what he needed to win it, no more and no less. Results don’t naturally play out that way.” Except that sometimes they do, and this seems to be one of those times. Calling these outcomes “statistically suspicious” or “mathematically insurmountable” doesn’t actually mean they’re incorrect or the result of any kind of foul play.

Beyond all this, a focus on Florida, Wisconsin, and other close states distracts you from noticing that Trump performed particularly well, both compared to Mitt Romney in 2012 and to the 2016 polls, in many of the most Republican states in the nation (North Dakota, West Virginia, and Wyoming). Candidates don’t have to perform the same ways in all states, but the lack of accusations of rigging in the states where Trump won handily (and also performed better than was predicted) show that the basis for these accusations is more perceived unfairness than actual statistics.

I think we can safely set aside this particular claim regarding the election. Yes, the polls were wrong, but as I’ve written, they weren’t that wrong—they were off by about 2 percent. That said, I agree with computer scientist Ron Rivest and statistician Phil Stark that random audits should be part of our electoral security system—not just for the 2016 presidential election but more generally. Transparency in election reporting is valuable irrespective of conspiracy theories. Every year there are calls to open up the vote tallies, and it never quite seems to happen. If an Electoral College/popular vote mismatch is what it takes to get this particular ball rolling, so be it.