“[T]he keys are phrased to reflect the basic theory that elections are primarily judgments on the performance of the party holding the White House,” he told the Washington Post in September 2016, while predicting a Trump win. “And if six or more of the 13 keys are false — that is, they go against the party in power — they lose. If fewer than six are false, the party in power gets four more years.”

His system, he says, is based on patterns observed since 1860, and he claims it let him accurately predict nine presidential elections correctly in a row since 1984. For what it’s worth, despite its apparent success, in my view the metric is too binary — it doesn’t give percentages of who is likely to win but simply an either/or determination. Nothing about presidential elections actually seems to work that way, and since we’ve seen close races in both 2000 and 2016, it’s plausible to believe that factors entirely outside Lichtman’s model — like extreme weather in key states — could have produced a different result. And indeed, even while he predicted Trump’s win in 2016, he also hedged at the time, saying, ” Donald Trump may well break patterns of history that have held since 1860.”

Nevertheless, the predictive success of his model is noteworthy, and he plausibly has latched on to key factors that contribute to or detract from a party’s ability to hold the White House. So his views are worth taking seriously.

And just as he went against the conventional wisdom that Democrats were primed to win in 2016, he’s going against the conventional wisdom now. In a new interview with CNN’s Chris Cillizza on Tuesday, he suggested that the idea — widely advocated by Democratic leadership, moderate pundits, as well as on the right wing — that impeachment would benefit Trump is completely off base.

In fact, he argued that foregoing impeachment could likely doom Democrats in 2020.

“Democrats are fundamentally wrong about the politics of impeachment and their prospects for victory in 2020,” Lichtman told Cillizza. “An impeachment and subsequent trial would cost the president a crucial fourth key — the scandal key — just as it cost Democrats that key in 2000. The indictment and trial would also expose him to dropping another key by encouraging a serious challenge to his re-nomination.”

He added: “Other potential negative keys include the emergence of a charismatic Democratic challenger, a significant third-party challenge, a foreign policy disaster, or an election-year recession. Without impeachment, however, Democratic prospects are grim.”

This is similar to an argument that I have been making, which is that the assessment of impeachment as a risky move by the Democrats underplays the extent to which not impeaching Trump is itself a risk. But Lichtman went even further and suggested that, at least as things look now, impeachment may be necessary to defeat Trump in 2020.

IMAGE: American University historian Allan Lichtman.