November 23, 2016

On November 17, six prominent political scientists gathered in front of a packed room at Chez Chicago to make sense of an election year that seemed to defy conventional wisdom at every turn.

Ethan Bueno de Mesquita, deputy dean for research & strategic initiatives at Harris Public Policy, moderated a discussion among University of Chicago political science professor John Patty; Mary McGrath, a political science professor at Northwestern University; and Harris faculty members Chris Berry, Alexander Fouirnaiesand Anthony Fowler.

Throughout the 90-minute discussion, the panelists touched on key points that ran contrary to much post-election analysis. Here are a few of the popular conclusions they called into question:

1. Data-driven forecasting doesn't work.

One expert kicked off the discussion with a critical question: “How did the forecasts go so wrong?"

Hillary Clinton had been widely expected to win against Donald Trump in the presidential race (the more optimistic polls had put Trump’s odds of victory around 28 percent). The shock of Trump’s surprise victory led many media commentators to proclaim the death of data-driven forecasting.

Not so fast, said the political scientists. Rather, this disconnect between reality and polls betrays imprecise inferences drawn from shoddy data—and reveals a need for more careful data and data scientists going forward.

First, the polls had major blind spots that were well known to those who study elections. State polls in particular have been found to systematically under-represent those with a high school diploma and those who are more difficult to contact, both of which are groups that skewed decidedly toward Trump.

Furthermore, election polls have a notoriously low response rate—around nine percent—forcing pollsters to use complex statistical techniques in order to extrapolate to the general voting public.

“And this enterprise is part science, and part art,” said one political scientist of the polling industry. “Really, it’s incredible that that’s the worst they’ve ever done.”

Another panelist suggested that pollsters might have skewed their interpretations of the data in response to a candidate as unconventional as Trump. “We were suffering from the idea that ‘This isn’t going to happen, because it can’t happen,’” said one expert. “So we saw people second-guessing the results they were getting from models.”

2. Bernie Sanders would have defeated Donald Trump.

Several audience members reiterated a point that had been floating around social media in the aftermath of the election: Had Bernie Sanders received the Democratic Party’s nomination instead of Hillary Clinton, he would have defeated Donald Trump.

One political scientist was quick to refute this claim: “There’s no compelling evidence that Bernie Sanders would have done better than Hillary Clinton.”

The panelist referred to a body of election research showing that, all else being equal, more centrist candidates tend to do better than radicals. Extreme candidates not only represent a smaller slice of the general public, the research shows, but also tend to mobilize support for their opponents, who see the extreme candidate as a radical threat.

It’s a finding that might not have boded well for the far-left senator from Vermont.

“We can’t say for sure. But there’s a lot of good evidence that moderation is good for candidates.”

3. We're seeing a major realignment of voters.

In the aftermath of the election, pundits have been fiercely debating whether the results indicate a dramatic shift in voting habits among particular demographics. Particularly, did Trump’s brand of conservatism appeal to typically Democratic working-class voters?

“One of the earliest studies of political voting behavior shows that people, time and time again, go back to their party ID,” said one political scientist.

In other words, the party with which a voter identifies is an excellent predictor of how they’ll vote.

Only rarely do candidates systematically mobilize a group from across the aisle—and 2016 doesn’t appear to be one of those occasions.

This year, about 90% of Democrats supported Clinton, while some 80% of Republicans supported Trump, a panelist pointed out.

While the voters haven’t necessarily realigned, one panelist speculated that the success of Trump’s unorthodox Republican platform—which combined traditionally conservative ideals of small government with protectionist trade and anti-globalization policies—may have repercussions in future elections.

“You’re going to see the new candidates taking seriously the idea that maybe they need to forge a new platform based on these outcomes."

4. Trump's victory guarantees success for Democrats in 2018.

Downtrodden Clinton supporters may have taken solace in the idea that, following Trump’s victory, Democratic congressional candidates will be sure to fare better in 2018. They would be referring to a phenomenon known as the mid-term slump.

“One of the big causes of the mid-term slump is regression toward the mean,” summarized one political scientist.

Essentially, an unusually popular presidential candidate is able to generate extra support for down-ballot candidates of the same party. But in the mid-term election two years later, that party’s candidates, no longer able to ride the coattails of the popular presidential nominee, don’t fare as well.

Which is why a razor-thin margin of victory is bad news for the opposing party. “The fact that Donald Trump did not win the popular vote means there may not be a lot of regression to the mean,” the panelist concluded.

However, other panelists emphasized that a number of critical factors could affect what happens in 2018. Trump’s performance as president, for instance, could affect which candidates choose to seek reelection or enter the ring.

“And midterms are often referenda on the performance of the president,” added one panelist. “So we’ll have to see how the economy is doing then."

5. President Trump will enact a slate of unprecedented policy measures.

While much has been made of Trump’s drastic policy proposals on issues like immigration, free trade, and foreign policy, one political scientist pointed out that not everything on his policy agenda was so unexpected.

“What he’s proposing on the regulatory front are exactly what I would have expected from Jeb Bush, Marco Rubio, or any other Republican," said the panelist.

Some pundits believe that Trump might use the current Supreme Court vacancy to appoint a jurist determined to overturn Roe v. Wade. The panel, though, saw little chance of that happening. “Scalia is the one being replaced,” they reminded the audience. “He was a die-hard pro-life vote. So the ninth justice that Trump will get to appoint won’t change the balance on that particular issue.”

In general, the panelists agreed, President Trump would need congressional support to legally enact meaningful changes on almost any front—although, there are exceptions.

“Immigration is a very, very gray area,” noted one panelist. “Courts don’t like to get involved there. It’s not clear who would stop him from saying, ‘Start building a wall.’”

--Jake Smith