Voters cast their ballots on Election Day in Elyria, Ohio. (Aaron Josefczyk/Reuters)

On Nov. 2, six days before the general election, the political world paused to watch a live stream from the campus of Marquette Law School in Milwaukee. Charles Franklin, director of the school's election poll, talked through the release of its final numbers. After days of awful headlines about the FBI probe of her email server, Hillary Clinton had held onto her lead, and enjoyed a 44-37 percent advantage over Donald Trump.

“Concern about Clinton’s use of a private email system does not appear to have shifted much in the wake of the FBI news,” Franklin said.

That, it seemed, was that. The poll was the “gold standard” for Wisconsin, having nailed the results of the 2012 and 2014 elections. A planned Trump rally in Wisconsin was canceled; the candidate stumped in Minnesota instead. On Election Day, FiveThirtyEight's poll-based review of Wisconsin gave Trump only a 16.5 percent chance of winning the state.

When Trump did win it, Franklin was not in Wisconsin. He was in New York, aiding ABC News on its election decision desk, and he watched the entire Midwest swing toward Donald Trump by a greater margin than any poll had suggested. In most of Wisconsin, Trump was outperforming the poll by six points; in the Milwaukee suburbs, where he was supposed to be unusually weak, he ran ahead of the poll by 10 points.

“No one will ever say the Marquette poll is 'never wrong' ever again,” Franklin said in an interview this week. “We've now been wrong. It's that simple.”

The 2016 election, which rewarded the media's love of hyperbole, made fools of almost every prognosticator. Nate Silver's FiveThirtyEight, which had given Hillary Clinton a 71.4 percent chance of victory, came out better than most; Sam Wang, director of the Princeton Election Consortium, blew it so badly that he made good on a promise to eat a bug live on TV. (He pronounced it “nutty.")

But when the national popular vote was certified, the major national pollsters were nearly redeemed. The final Washington Post/ABC News tracking poll put Clinton three points ahead of Trump. She won the popular vote by 2.1 percentage points, about the same as Jimmy Carter's 1976 margin over Gerald Ford. No national pollster was as badly burned by 2016 as by 2012, when those projecting a tie between President Obama and Republican challenger Mitt Romney had to explain a clear Obama majority.

It was a different story in the states, where a half-dozen pollsters, seen as rock-solid for their command of the local numbers, saw a Clinton win that never materialized. In Ohio, which seemed to have slipped away from Clinton in the summer, the Columbus Dispatch's unique mail poll — praised by Silver as the country's most accurate — seemed to find late movement her way. Its final numbers, released on the Sunday before the election, found Clinton and Trump deadlocked. Two days later, Trump triumphed in Ohio by eight points, the biggest Republican victory in the state since 1988.

“I realized that our poll, which showed Clinton leading by one point the weekend before the election, was going to be 'wrong' as I was monitoring results from across Ohio showing her underperforming in urban counties and Trump rolling up extraordinary margins in more rural areas,” said Darrel Rowland, the Dispatch reporter who runs the poll. “Our presidential poll has never been that far off in my memory nor that of my predecessor, who started at the Dispatch in the early 1970s.”

In Michigan, which went red for the first time in 28 years, the final EPIC/MRA poll found Clinton clinging to a four-point lead over Trump. Unlike in Wisconsin, neither Trump's nor Clinton's campaign responded like the state was deadlocked. Nonetheless, that poll — and all but one Michigan poll conducted in 2016 — pointed to a Clinton victory that never happened.

Like Marquette Law's poll, the EPIC/MRA survey had concluded after a few bad Clinton news cycles but before the campaign's final days. “We had an 11-point lead after the debates, it dropped down to seven after WikiLeaks, then after the [FBI Director James B.] Comey letter it dropped down to four points,” said Bernie Porn, the poll's director. “The only thing I wish we'd done is take one more poll the weekend before the election. I assumed, based on the numbers we found, that Clinton would win unless she wasn't able to turn out enough blacks in Detroit and Flint. And that was what happened.”

In Pennsylvania, two reliable state pollsters — Susquehanna and Morning Call/Muhlenberg College — released final numbers that suggested Clinton was going to repeat recent Democratic history, with a narrow win. The latter poll, which suggested Clinton was ahead by six points — outside the margin of error — also surmised a seven-point Democratic registration advantage. On Election Day, said Muhlenberg College pollster Chris Borick, the advantage was only three points.

“Our methodology seemed to work well in traditional races but not this race,” he said. “The idea of the 'silent Trump voter' seemed to fit well with this. But who is that voter, and will he or she be back? I’m worried about an overreaction, that we try something vastly different than what worked in the past, and by adjusting too much.”

Jim Lee, the owner of the Susquehanna poll, noted that its final numbers — a two-point Clinton lead — fit within the margin of error. But he also suggested it was too early to tell whether the old model of Pennsylvania elections had been bent for Trump, or totally shattered by Trump.

“We've always told Pennsylvania Republicans that you can't win statewide if you're badly losing in the southeast and Philadelphia,” he said. “This year, even Chester County went for Clinton. She came out of the suburbs with exactly the margin that Democrats usually need to win. And she didn't, because of losses everywhere else. That's a huge transformation.”

Both polls had, indeed, found Trump underperforming in the Philadelphia suburbs and overperforming in rural areas and Appalachia. What belatedly occurred to Lee was that the polling that relied on robo-calling, instead of live interviewers, found people more willing to admit they were backing Trump. “Half of the surveys were done by live interviewers and that group found Clinton winning by eight,” he said. “The automated interviews had Trump winning by two.”

Still, in conversations with state pollsters, none suggested that the playbook needed to be shredded. The media, they said, played two roles in leaving the impression that Clinton was wrapping up the election along the “blue wall.” First, across the industry, each poll's top-line number was reported as definitive. A pollster could explain why a two-point Clinton lead with a three-point margin of error allowed for a Trump win. But that was rarely, if ever, how the horse race was reported.

Second, the cutbacks that have affected newsrooms around the country affected pollsters, too. There were fewer surveys of swing states in general, and almost no budget for polls in the final days before the election, when exit polls found Trump eking out a winning margin.

“We have conducted as many as seven [polls] in a year that included primaries,” said the Columbus Dispatch's Rowland. “This year we were down to one, published the Sunday before Election Day. I had to personally assume numerous additional tasks because so many people had been let go by our company.”

Each pollster agreed that more timely polling would have helped portray what was really happening — a very close election tipping toward the Republican nominee. But no one had a solution for the problem of responses. After all, neither robo-calling or live interviews, on their own, had nailed the numbers. The Dispatch's mail poll, which had been so accurate in the past, nailed the results of the state's Senate race — a walkover for Sen. Rob Portman (R-Ohio). But Trump supporters simply didn't respond to the poll in representative numbers.

“While the 1,100-plus we got back this time is far better than the samples of 500 used in some surveys, our return rate has gotten smaller over time,” said Rowland. “It's not just the 'younger generation' that is turning away from snail mail these days. All of this — not the results of a single poll — will go into our consideration of whether to continue conducting the Dispatch Poll.”

For other pollsters, 2018 represented a fresh start, without a Democratic candidate whose post-State Department unpopularity and negative story lines scrambled the electorate.

“Were it not for hacking, WikiLeaks, and the Comey letter, she would have won,” said EPIC-MRA's Bernie Porn. “If she had stuck to the economic message instead the of all-attack-ad approach, based on Trump's comments, she might have won. And if the Democrats had nominated anyone else, they probably would have won. Joe Biden would have won. Bernie would have won.”