You Can Ignore Allan Lichtman and His Presidential Predictions

He can’t hurt you

You might not know the name Allan Lichtman, but you probably know his handiwork.

If, while infinity scrolling through presidential election news, you scanned a headline that mentioned a “Historian who correctly predicted last 10 elections” or a “Professor who foresaw Trump 2016 victory”: congrats, you’ve heard of Allan Lichtman.

Lichtman, an American University history professor, is famous for developing a model to predict the outcome of US presidential elections. His model, “The 13 Keys to the White House,” ignores the usual polling data in favor of fundamentals. The 13 keys are true-false statements, which include:

• There was a serious primary contest for the incumbent party nomination.

• The incumbent party candidate is charismatic or a national hero.

• The incumbent administration is untainted by major scandal.

• There is no significant third party or independent campaign.

(For the full list of keys, see here.) To predict the next president, simply evaluate whether each of the 13 statements is true or false, then tally the number of “trues”. If there are 8 or more “trues”, then incumbent party will win; otherwise, the challenging party will win.

Lichtman’s clout comes from his track record. His model, as described by Newsweek, “has correctly predicted the winner of every presidential election since 1984.”

He especially touts his 2016 prediction that Donald Trump would win, as it ran counter to all polling. This victory, he might say, shows that horse-race polling can be misleading. The fundamentals are what shape the race.

This is another part of the keys appeal: their simplicity. In a world where Nate Silver creates models that gulp down hundreds of polls from dozens of hand-rated pollsters, calculate cross-state demographic correlations, and run thousands of simulations to spit out that Biden has a 31% chance of victory, the 13 keys can feel so much more straightforward. Don’t worry about the polls, Lichtman says. Just focus on the fundamentals. Has the incumbent party done a good enough job that people will support it on election day? You don’t need a degree in stats to know who will win; you just need to answer 13 straightforward yes-or-no questions. It’s simple, Lichtman says, and it’s right.

The problem, of course, is that it’s neither.

On the simplicity front, many keys are so open to interpretation that they can say whatever you want. Consider keys like:

• The incumbent party candidate is charismatic or a national hero.

• There is no sustained social unrest during the term.

• The incumbent administration is untainted by major scandal.

How does one draw the boundaries around “charismatic” or “social unrest” or “major scandal”? According to Lichtman, the Iran-Contra scandal was not a “major” scandal, nor is the current fiasco surrounding Biden stepping down. The race riots of 1992 did not constitute “sustained social unrest.” This is not to say that these individual calls are wrong, but rather that they’re subjective. The boundaries are wherever Lichtman thinks they are.

Or consider “charisma.” Lichtman deemed Obama charismatic in 2008, but uncharismatic in 2012. Lichtman's rationale, according to Wikipedia, was that Obama has “not regained the magic of his 2008 campaign.” And, sure, 2008's passionate fling with Obama had faded into a sexless marriage by 2012. But the swing shows that the criteria isn’t about Obama’s inherent charisma, it’s about whether he was popular at the moment of the election.

As such, “charisma” is really a way for Lichtman to sneak polling in the backdoor. Claiming 2012 Obama “has not regained the magic” is just using the polls without citing the polls. Later, Lichtman even directly cites polls to justify the charisma key for Trump, claiming that he can’t be charismatic because “he has the highest negative ratings of any presidential candidate in history.”

And I mean, I would want to use polls in my model too! But Lichtman then goes around in interviews proclaiming that he “eschews the polls”, that “the polls are not predictors”, and that “the biggest mistake political prognosticators make is relying on the polls.” Use polls if you want. But trumpeting the wisdom of fundamentals, while stealing glances at the polls to inform your prediction, is cheating.

Lichtman goes even further in twisting the definition of other keys.

In his 2016 prediction, Lichtman inexplicably leaves the “There is no serious contest for the incumbent-party nomination” key undetermined. This is to say, he’s claiming the 2016 Bernie-Hillary primary might not constitute a “serious contest.” He makes the excuse that, sure, Bernie winning 43% of primary votes does seem kinda serious and, yes, it would meet the usual criteria Lichtman uses, but Bernie didn’t stage a coup at the convention itself, so *shrug* who’s to say?

This is nonsense. Since 1972, the primary system was specifically designed to replace conventions as the venue for “serious contests.”

Moreover, it’s ludicrous to treat “was there a serious primary contest?” as something that will be determined later. By the time he made his prediction in October 2016, the primary, convention and nomination were long over. No information could arrive that would change whether there was a serious contest. Either it was, or it wasn’t. But Lichtman refused to choose.

There is only one reason I can see for his flakiness here: it gives him an out if Trump lost. Remember, Lichtman predicted Trump would win in 2016. But despite crowing about this prediction for years after, the actual prediction is very timid. It’s riddled with caveats about how Trump might “break the historical pattern” and how multiple keys might change after the election. Lichtman, in short, seemed very worried that his model might be wrong. So he states a few of keys might change after the election, in which case Clinton would win and his model would still be ‘right.’

In his rationale for leaving “was there a serious primary contest?” undetermined, he says:

It is also possible that the specter of a Trump victory has united Democrats regardless of the Sanders/Clinton contest.

This is the only hint he gives as to what would ‘determine’ the key. It’s hard not to read this as saying if Trump loses, the primary wasn’t really contested because look, all the Democrats united behind Clinton! Again, this is ridiculous. The keys are supposed to predict the election, but he’s using the outcome of the election to determine what the keys say.

Moreover, if Lichtman were to change other keys after the election (which he sometimes does) like the “there is no significant third party campaign” key (which he did), then the model would predict a Clinton victory. Thus, by giving himself wiggle room, Lichtman can claim that even if Trump lost, he simply misread the keys, and that his model still has a perfect track record.

Does this sounds unscrupulous? Well, Lichtman has already done this sort of thing.

In 2000, his model predicted Gore would win. Upon Gore losing — admittedly in freak circumstances involving the Supreme Court — Lichtman claimed that his model only predicted the popular vote. For the next 16 years, he continued to emphasize that this model only predicted the popular vote.

To be fair, this makes sense. His model is about broad national issues, so it should predict the broad national vote. It seems inconceivable, in contrast, that such broad statements could capture the byzantine nuances of the Electoral College. So while his retroactive claims that the keys predict the popular vote is a bit of a cope, I can go along with it.

As a result, though, when Lichtman made his famous prediction that Trump would win in 2016, he was specifically predicting Trump would win the popular vote. It’s right there at the top of his 13 keys paper for 2016:

As a national system, the Keys predict the popular vote, not the state-by-state tally of Electoral College votes.

He was obviously wrong, as Trump lost the popular vote. But fortunately for Lichtman, Trump won the electoral college.

So Lichtman just started claiming his model predicted the Electoral College, counter to everything he had said the prior 16 years, including in his own 2016 prediction. The online version of his 2016 prediction — which, unlike the official paper, can easily be edited — even contains a “correction”:

It has been corrected to read that Prof. Lichtman’s 13 Keys system predicts the winner of the presidential race, not the outcome of the popular vote.

Which, again, is a lie. That’s not what he was predicting before, and you can see it in his official prediction. With this little sleight of hand, he emerges every four years like some sort of presidential-prediction cicada to feed on media coverage of his ‘perfect record.’

There’s an element of astrology here. The 13 keys are flexible enough that they can be bent to match the contours of any situation. Is your party leader charismatic but polarizing? That’s not real charisma, right? Was the bitter, months-long, almost 50-50 split between Bernie and Hillary really a serious primary contest? Do the keys predict the popular vote or the Electoral College?

The answer to all of the above is: it doesn’t matter! The answers are whatever they need to be make the model right. (Also, feel free to change your answers after the election!)

That rant done, I admit the keys are not totally bogus. They’re all strongly correlated with election outcomes. A charismatic leader, a good economy, foreign policy successes, and no scandals, all make an incumbent party more likely to win. An incumbent party with a majority of the keys forming a wind at its back has good odds of winning. So it’s not to say that there’s nothing here.

But the mystical infallibility of the keys is bogus. They’re not simple, and they’re not always right.

Lichtman was recently doing the media rounds offering advice about whether Biden should drop out and, now that he has, whether the Dems should coalesce around Kamala. Unsurprisingly, his advice is based entirely around maximizing the keys.

To me, this is akin to hearing a stat like “the Green Bay Packers have never lost a home playoff game when the temperature is below 19 degrees F!” and blasting the air conditioning to ensure victory. Congratulations, you found a criteria correlated with victory. Meeting that criteria does not deliver victory. The Democrats would be far better off thinking through the specific situation, rather than contorting themselves to follow an open-ended model that has already proven it can be wrong.

Lastly, everyone should be supremely skeptical of Lichtman’s motivations.

As a person and professor of history, I dunno, maybe he’s great.

But as a public intellectual, he reeks of someone who values clout over intellectual honesty. A good intellectual keeps an accurate tally of when they were right and wrong. They note their mistakes and correct themselves. They calibrate. Lichtman does the opposite, touting a non-existent ‘perfect record’ and, worse, retroactively editing his predictions to maintain the illusion. Don’t make election decisions based on him.

---

If you like this post, you may also like Nate Silver’s Finest Hour, about the 2016 presidential election forecasting drama.

---