Few prognosticators predicted a Donald Trump victory ahead of Tuesday night. Polls showed Hillary Clinton comfortably ahead, and much of America (chiefly the media) failed to anticipate the wave of pro-Trump support that propelled him to victory. But a Washington, D.C.-based professor insisted that Trump was lined up for a win - based on the idea that elections are "primarily a reflection on the performance of the party in power."
Professor Allan Lichtman uses a historically-based system of what he calls "keys" to predict election results ahead of time. In our conversations in September and October, he outlined how President Obama's second term set the Democrats up for a tight race, and his keys tipped the balance in Trump's favor, even if just barely.
At the end of our September conversation, Lichtman made another call: That if elected, Trump would eventually be impeached by a Republican Congress that would prefer a President Mike Pence - someone who establishment Republicans know and trust.
"I'm going to make another prediction," he said. "This one is not based on a system, it's just my gut. They don't want Trump as president, because they can't control him. He's unpredictable. They'd love to have Pence - an absolutely down the line, conservative, controllable Republican. And I'm quite certain Trump will give someone grounds for impeachment, either by doing something that endangers national security or because it helps his pocketbook."
So while Republican voters clearly came home before Nov. 8 - network exit polls show 90 percent of GOP voters cast ballots for Trump - it's less clear that the party leadership is onboard. (Lichtman actually isn't the only person to predict a Trump impeachment; Friday morning, The New York Times' David Brooks suggested a Trump impeachment or resignation was "probably" in the cards sometime within the next year.)
It's worth noting that Lichtman's predictions use very different methods than pollsters and data-based prognosticators. Some statisticians take issue with the structure of his system, a set of 13 true/false questions, saying that the binary nature of his keys leads to what's called "overfitting," which is basically creating a system that fits the data but has little statistical significance. But Lichtman counters by saying that system has correctly predicted every election since 1984 (specifically, his predictions have picked the next president correctly in all of those elections but 2000, when he picked Al Gore, who won the popular vote). And Lichtman has his own criticism for data-based predictions.
"Polls are not predictors," he said Friday in an email. "They are snapshots that simulate an election. They are abused and misused as predictors. Even the analysis of polls by Nate Silver and others which claimed a probable Clinton victory with from more than 70 percent to 99 percent certainty are mere compilations that are no better than the underlying polls."
And he has particular disdain for prediction systems that assign a likelihood of winning.
"For all his acclaim, Nate Silver is only a clerk, not a scientific analyst," Lichtman said.
As for the real reason for Trump's win, Lichtman says the blame can't be put on Hillary Clinton or her campaign - rather, he says, it was decided by the larger forces that shape American politics.