Did Comey Cost Clinton the Election? Why We’ll Never Know

In one of the closest presidential elections in American history, anything and everything could have been decisive. But we’ll probably never know what really was.

The letter about Hillary Clinton’s emails that the FBI director James Comey sent to Congress in October 2016 is one of these potential factors. Clinton herself told Fareed Zakaria on CNN last fall that she “would have won but for Jim Comey’s letter on October 28th.”

But as the Department of Justice concluded its report into Comey’s handling of the investigation into Clinton’s email server on Thursday, it is still unclear whether this letter cost her the presidency.

Comey’s letter came about one week after the third presidential debate and less than two weeks before Election Day. At that time, most polling averages showed Clinton ahead by around 6 percentage points in national polls. A week later, her lead had declined to 3 points.

The 3-point shift against her is often attributed to Comey’s letter — three points being more than enough to cover Donald Trump’s narrow margins in Michigan, Pennsylvania and Wisconsin. He won each state by less than 0.8 percentage points.

Late-deciding voters broke overwhelmingly for Trump, the exit polls showed, and the Comey letter and its disclosure of new information in the email investigation was a significant part of the news coverage over the last week of the election.

It’s a plausible case, but there’s a problem: Clinton’s support was probably already in decline before the Comey letter.

This decline makes it harder to claim definitively that Comey’s letter was responsible for any subsequent decline. The trends leading to Clinton’s defeat — in particular, Trump’s consolidation of hesitant Republican support — may have already been underway.

The ABC/Washington Post tracking poll, for instance, showed Clinton’s lead dwindling to just 2 points in its last poll before the letter, down from a double-digit lead after the third debate. One of our own polls — a collaboration with Siena College — showed Trump leading in Florida in the days before the letter. No live interview poll had shown Trump with such a large lead in Florida since early July.

It’s hard to be too sure of how far Clinton’s standing declined before the Comey letter because there were few polls conducted in the days leading up to it. But on average, five national daily tracking polls showed Clinton’s lead falling to 3 points in the surveys ending on Oct. 28 (but released after), according to analysis after the election by the American Association of Public Opinion Research. That was down from 6 points on Oct. 22.

The overall national poll average showed less weakening in Clinton’s standing, since they included nontracking surveys conducted at various points over the previous week. Most important, Clinton’s lead didn’t decline further after the letter, at least not in these five polls.

The American Association of Public Opinion Research analysis concluded that public data did not offer a definitive answer about the effect of Comey’s letter on the election.

An alternative narrative suggests that the Comey letter had no discernible effect at all. According to this theory, Clinton had already bottomed out around the time of the Comey letter, as post-debate coverage faded and Republican-leaning voters finally decided to back the Republican nominee.

This is possible, but there isn’t enough data to prove it either. The burden of proof is fairly high, given the size of Clinton’s post-debate lead, the number of polls that supported it, and the short amount of time between the third debate and the Comey letter.

Even if the Comey letter did affect the race, the effect might have faded in the final days of the campaign. By then, Comey’s letter was more than a week old, and he would issue a second one, stating that the new information had not altered his conclusion that Clinton should face no charges over her handling of classified information. In some accounts, the second letter may have hurt her more than the first.

Clinton’s lead in the polling average actually grew over the weekend before the election — though, again, there’s no way to know whether her support rebounded because of Comey’s second letter or something else.

None of this proves that Comey’s letter didn’t play a decisive role, even if it calls into question whether the polling evidence supports the view that it did. The election was extremely close, and public opinion polls aren’t good enough to confidently identify a net-0.8 point effect, let alone to determine the cause. The polls would be even less likely to identify a Comey effect if it accumulated over a week, rather than all at once, among undecided voters who made up their minds after the last polls were conducted or even in the voting booth.

Even if there was a real and decisive Comey effect, there simply might not be very good evidence for it.