In August, I mentioned that I had chosen the title “Political Reality” for my memo in part because of my liking for oxymorons. I classed that title with other internally contradictory statements, such as “jumbo shrimp” and “common sense.” Now I’m going to discuss one more: “expert opinion.”
This memo was inspired by a thought that popped into my head when the outcome of the election settled in. You may point out that at the end of my November 14 memo “Go Figure!,” I said I wouldn’t write any more about politics. True, but I didn’t say I wouldn’t think about politics. Anyway, this memo isn’t about politics, it’s about opinions.
Last spring I attended a dinner where one of Hillary Clinton’s senior advisers was soliciting input, as she and her campaign were struggling to come up with an effective counter to Bernie Sanders’s populist message. Most of those present expressed frustration on the subject, until an experienced, connected Democrat assured everyone, “Don’t worry. She’ll win. The math is irresistible.” The Hillary supporters were relieved, and he turned out to be right: she won the nomination going away.
In late October, with the issue of Clinton’s private email server and the FBI’s new investigation further dogging her, that same seasoned Democrat was asked whether the election was in jeopardy. “Don’t worry,” he said. “She’ll win. The math is irresistible.” We all know the result.
The opinions of experts concerning the future are accorded great weight . . . but they’re still just opinions. Experts may be right more often than the rest of us, but they’re unlikely to be right all the time, or anything close to it. This year’s election season gave us plenty of opportunities to see expert opinion in action. I’ll start this memo by reflecting on them.
The Year Polls Stopped Working
Pollsters got off to a tough start last year with the June referendum concerning Britain’s membership in the European Union. Right up to the end, both pollsters and bookmakers considered U.K. citizens 70% likely to vote to remain a member. But, in the end, “Leave” won by a few percent.
The reaction was shock. Voters on both sides of the issue were unprepared for the outcome. Within a day or two, the leaders of Britain’s main political parties had stepped down. People began to seriously discuss what that outcome meant and how “Brexit” would be accomplished.
The explanations for the pollsters’ error centered around Britain’s lower level of experience with, and expertise in, polling. It couldn’t happen in the U.S. In fact, in the 2008 and 2012 presidential elections, Nate Silver, the proprietor of website FiveThirtyEight, correctly predicted the outcome in all 50 states once and in 49 the other time.
In 2016, FiveThirtyEight estimated the odds of Hillary Clinton winning as slightly better than 50/50 as of the end of the Republican convention in July. Then it had her as an 8-to-1 favorite in August, when the Democrats concluded their convention and Donald Trump’s perceived missteps peaked. And then it again said she was slightly ahead just before the first presidential debate on September 26. It never made her out to be an underdog. And on election day, it estimated that she was 71.4% likely to win.* Most other pollsters put her chances of winning at between 80% and 99%, and only one considered Trump the favorite.
In the end, of course, Trump won in the Electoral College by a final count of 304 to 227, despite losing the popular vote by almost 2.9 million votes, or about 2%. In particular, he won in a number of “swing states,” such as Pennsylvania, Michigan and Wisconsin, where the polls had him well behind. So much for experts’ forecasts.
Finally, rounding out the pollsters’ failures in 2016, the reform referendum that Italy’s Prime Minister Matteo Renzi bet his career on – which had been considered 3% behind – lost by 20%. The outcome wasn’t a surprise, but the margin certainly was.
No one really knows why polling failed so miserably last year. Clearly there was a groundswell of populist, anti-establishment and anti-insider sentiment, but shouldn’t it have been detected? In particular, Trump did much better than predicted (or much less badly) with a number of important groups, such as Hispanics and college-educated women.
For some reason, in 2016 pollsters in all three countries either failed to talk to a representative sample of voters, failed to elicit honest responses, or failed to accurately interpret the data. Thus their opinions may be accorded less weight in the future.
So Much for the Experts
I’m struck by how dramatically opinion can flip-flop:
During the run-up to the election, Clinton’s campaign organization and “ground game” were considered sophisticated, efficient and unstoppable, and Trump’s were thought of as rag-tag, underfunded and uncoordinated. Now Trump’s machine is described as having been highly effective, and Clinton’s as having missed important signs and opportunities.
Clinton’s message was thought likely to carry a lot of weight with a broad swath of the electorate, while Trump’s was viewed as appealing deeply to a few fervent but narrow fringe constituencies without enough voters for him to win. After the fact, Trump is described as having had “perfect pitch” and Clinton as having a “tin ear.”
In particular, now it’s considered to have been a big mistake for Clinton to fail to address the concerns of white men and set out a solution for those who lost jobs and were omitted from economic progress. But during the campaign, no one pointed to this error.
* It should be noted – to his credit – that Silver insisted repeatedly that Trump could win. In fact, he often reminded his followers that the Clinton landslide most people expected was no more likely than a modest Trump victory. Silver also entered Election Day citing a 10.5% probability that Trump would lose the popular vote but win the presidency. We can’t say he predicted that outcome, but (a) he was more explicit about it than most and (b) he assigned a fairly material probability to an event that in the past has been quite rare (so it can’t be said that he was just extrapolating).
Finally, up until Election Day, most observers (including me) talked about the likelihood that the Republican Party would emerge from the election torn between its traditional faction, the Tea Party conservatives, and Trump’s economically disgruntled, anti-establishment supporters. That may turn out to be the case, but now the Democratic Party is described as being at risk as well because of the schism between the Clinton-type moderates and the Sanders/Warren progressives.
Here’s some of what I wrote in “Go Figure!,” six days after the election:
Think back to just before last week’s election. What did we know?
The polls were almost unanimous in saying Hillary Clinton would win . . .
There was a near-universal belief that a Trump victory – as unlikely as it was – would be bad for the markets.
So what happened? First Clinton didn’t win. . . . And second, the U.S. stock market had its best week since 2014! . . . Thus two key observations can be made based on last week’s developments:
First, no one really knows what events are