In the blurry-eyed morning after an election, many Americans are wondering how Donald Trump won. Even for Republicans who have been involved in campaigns for years, there was an expectation that today would be the end of Trump, not the beginning of his presidency. Trump was seen as too different from politics to win. The polls appeared to back up this assertion. Yet, this morning, it is Trump, not Hillary Clinton, who won.
Political science influences journalism more today than ever before. Major national newspapers and new organizations hire political scientists to assess politics. Polls and complex statistical models developed by political scientists are used by journalists and pundits. But something happened last night that was so unusual that political scientists have egg on their face this morning.
Not so fast.
The biggest mistake made by pundits and prognosticators was to assume that this election was unusual. It wasn't. It confirmed what political science has learned by studying elections over the past 60 years or more.
Here are five things political science got right.
1. Election results are driven by fundamentals that predicted a tight race
There is a debate among political scientists over whether campaigns and candidates matter in elections. Our ability to predict vote outcomes months, even a year prior to an election, without using any data on the candidates suggests that campaigns don't matter.
Instead, they are based on certain election fundamentals. Elections are largely predicted by the national economy (which is currently positive but not booming), how many years the White House has been held by the same party (which hurt Clinton), how popular the current president is (who is popular but less so than presidents whose party eventually won), and the polls (which were close).
A recent symposium brought together political scientists who have developed models based on these fundamentals. These statistical models predicted a tight race. The models that excluded any polling data predicted that Trump would win the popular vote; the ones that included polls predicted a tie or Clinton victory. Regardless of the final prediction, each of the models predicted a race so tight that the election could reasonably go either way.
Drew Linzer is Chief Scientist at Civiqs, an independent opinion research company. His graph of presidential approval rating, GDP growth, and vote share demonstrate how "normal" 2016 was: Clinton's vote share was exactly where we would expect given economic growth and President Obama's approval rating.
The mistake by pundits was to assume that because Trump's campaign was so different in style and substance that this must mean a change in these fundamentals. Instead, Trump's victory confirmed what political science has consistently found: elections are driven by forces that are often out of the hands of the candidates.
2. Voting is based on partisanship, not issues
One of these fundamentals of elections is the dominance of partisanship in how individuals vote. Voters rarely vote on issues, values, or beliefs. Instead, they adopt positions advanced by their party.
I think we can all agree that Trump took some audacious policy positions. Some voters---including many Republicans---would be repelled by these stances, right?
Nope. People rarely vote on issues. Instead, they vote based on partisanship. When Trump takes a position that Republicans have rejected for decades (such as rejecting free trade deals), Republicans did what political science would expect them to do: they changed their minds on the issue to fit their partisanship.
One of the starkest examples of how partisanship drives positions was the change in how white evangelicals linked the private immorality to his or her public ethics.
The Public Religion Research Institute asked Americans the following question:
Do you think an elected official who commits an immoral act in their private life can still behave ethically and fulfill their duties in their public and professional life?
In 2011, only 30 percent of white evangelicals said that an official with immorality in his or her private life could be ethical in their duties. This year, 72 percent took this position. Such a shift was driven by partisans coming to grips with a Trump candidacy.
3. Race and religion also drive votes. Gender? Not directly
It's impossible to talk about 2016 without addressing gender. We had the first woman to lead a major party ticket. Her opponent was a candidate whose behavior and talk was at worst misogynistic and at best crass.
Did this expand the gender gap in the election?
This is a trick question because there isn't a gender gap in presidential elections. Men and women may vote differently, but this difference is explained by partisanship. Men are more likely to be Republican and women are more likely to be Democrats. There is little to no added effect of gender on voting. The final survey results aren't in yet for a full analysis, but it appears to be more of the same.
There are two factors that do have an effect above and beyond partisanship. One is race. The other is religion. We can see these two intertwine by looking at white evangelicals. One result is that this year had highest support by white evangelicals for a Republican since they voted for President Bush.
4. Campaigns reinforce the fundamentals
Candidates campaign, which suggests that campaigns matter. But campaigns don't change the dynamics of politics. Instead, they reinforce the fundamentals.
George Washington University's John Sides, who writes for the Washington Post's Monkey Cage blog has literally written the book on campaigning.
“Instead, campaigns matter more profoundly by making the economy, party loyalties, and other fundamentals more important to voters,” Sides said. “Very many studies have found that, as the campaign goes on, voters are increasingly led toward the fundamentals.”
In 2016, this meant that the campaigns did not change the election outcome. Instead, they made the fundamentals important to voters.
5. Polling works
Don't blame the polls. Instead, understand them better.
Polls cannot predict any outcome perfectly. There are many reasons for polls to be off, but even at their best, a poll during an election will be off by around three percentage points. Because they're only a sample and not the actual vote, they could be off by a few percentage points.
The most recent polls put Clinton as winning the national vote by about two points. This means that a tie or a loss could reasonably occur. Still, the popular vote appears to be going to Clinton. A smaller victory in the election than (most of) the latest polls, but the result is one that is comfortably within what pollsters would expect.
There was less precision at the state polls, but most of these polls are poorly run and do not follow the same rigorous practices as national polls. These can and should be improved. Survey researchers know how to do these polls better. The problem is a lack of resources, not knowledge of how to poll.
Bottomline: 2016 wasn't something new or unusual. It appears to support what political science has found in its decades of studying American elections.