The Evidence That the Comey Letter Threw the Election Is Overwhelming

Nate Silver’s essay about the Comey effect is very rich, and there are multiple points worth discussing. But let’s start with the bottom line evaluation, which is definitive:

Clinton’s standing in the polls fell sharply. She’d led Trump by 5.9 percentage points in FiveThirtyEight’s popular vote projection at 12:01 a.m. on Oct. 28. A week later — after polls had time to fully reflect the letter — her lead had declined to 2.9 percentage points. That is to say, there was a shift of about 3 percentage points against Clinton. And it was an especially pernicious shift for Clinton because (at least according to the FiveThirtyEight model) Clinton was underperforming in swing states as compared to the country overall. In the average swing state,3 Clinton’s lead declined from 4.5 percentage points at the start of Oct. 28 to just 1.7 percentage points on Nov. 4. If the polls were off even slightly, Trump could be headed to the White House.

Is it possible this was all just a coincidence — that Clinton’s numbers went into decline for reasons other than Comey’s letter? I think there’s a decent case (which we’ll take up in a moment) that some of the decline in Clinton’s numbers reflected reversion to the mean and was bound to happen anyway.

But it’s not credible to claim that the Comey letter had no effect at all. It was the dominant story of the last 10 days of the campaign. According to the news aggregation site Memeorandum, which algorithmically tracks which stories are gaining the most traction in the mainstream media, the Comey letter was the lead story on six out of seven mornings from Oct. 29 to Nov. 4, pausing only for a half-day stretch when Mother Jones and Slate published stories alleging ties between the Trump campaign and Russia.

[…]

We also have a lot of other evidence of shifting preferences among voters in the waning days of the campaign. Exit polls showed that undecided and late-deciding voters broke toward Trump, especially in the Midwest. A panel survey conducted by FiveThirtyEight contributor Dan Hopkins and other researchers also found shifts between mid-October and the end of the campaign — an effect that would amount to a swing of about 4 percentage points against Clinton.⁵ And we know that previous email-related stories had caused trouble for Clinton in the polls. In July, when Comey said he wouldn’t recommend charges against Clinton but rebuked her handling of classified information, she lost about 2 percentage points in the polls. Periods of intense coverage of her email server had also been associated with polling declines during the Democratic primary.

[…]

So while one can debate the magnitude of the effect, there’s a reasonably clear consensus of the evidence that the Comey letter mattered⁶ — probably by enough to swing the election. This ought not be one of the more controversial facts about the 2016 campaign; the data is pretty straightforward. Why the media covered the story as it did and how to weigh the Comey letter against the other causes for Clinton’s defeat are the more complicated parts of the story.

As Silver goes on to say, the data is consistent with both a little (1-2 point) and big (3-4) Comey effect, and I agree with him that it’s prudent to assume the low end. But that was enough, so it’s not important. And given both the amount and steepness of the decline, the idea that regression to the mean explains all of it is massively implausible.

But, to me, the real kill shot is the data about the extraordinary amount of negative media coverage — without recent precedent in an election campaign — about Clinton that the letter demonstrably generated. As with arguments that the bully pulpit moves public opinion, it’s not just that the theory is inconsistent with the data, but that the theory is weak even on its face. To believe that the Comey letter didn’t change the outcome you have to believe that:

• The director of the FBI implied that one candidate was a crook
• Generating a massive wave of negative media coverage, one that wouldn’t just reach a few high-information voters
• At a time when an unusually large number of undecided voters who didn’t like either candidate were making up their minds
• But this had essentially no effect
• Even though there was a steep decline in her poll numbers in the immediate aftermath of the letter

And remember, the election was decided by less than 100,000 votes, so to conclude that the letter didn’t affect the outcome you essentially have to be saying that the coverage of the letter didn’t matter at all. This is extraordinarily implausible. You’re never going to see stronger evidence for a campaign counterfactual than this.

As we’ll get to in another post, this doesn’t mean that the Comey letter was necessarily the most important factor. It doesn’t mean that there wasn’t some choices Clinton could have made that would have overcome it. But it is nearly certain that all things being equal the coverage of the letter put Trump in the White House.