First, I assumed, as in other years, that the traditional measures of candidate quality mattered, and I relied on a handful of poll questions to show me not only what voters were thinking but also how they would behave.
Over and over again, post-convention polls found Trump with higher “unfavorable” ratings than Hillary Clinton. Both candidates were upside down (having higher negatives than positives), but Clinton’s numbers on that key question were normally a few points better. In addition, when pollsters asked whether Clinton and Trump were qualified to be president and had the right temperament for the job, the differences were stunning — Clinton had huge advantages on both qualities.
Just as important, on Clinton’s most glaring political liability, her trustworthiness and honesty, polls repeatedly showed voters saw little difference between the two candidates. Indeed, in exit polling, 36 percent of respondents said that Clinton was honest and trustworthy, compared with only 33 percent who said the same about Trump.
Together, these facts convinced me that Trump’s ceiling of support was considerably lower than Clinton’s. After all, if only 38 percent thought Trump was qualified to be president, how could he win when the eventual winner would probably need at least 45 percent of the vote?
I mistakenly concluded that voters could not and would not vote for a candidate they deemed “unqualified” for the presidency.
I also read too much into the various ballot tests in the weeks and months before the election.
I’ve been relying on public opinion surveys for decades, comparing public polls — such as those conducted by The Washington Post-ABC News and NBC News-Wall Street Journal — with poll numbers gathered, but not released to the public, by Republican and Democratic pollsters I hold in high regard. I did that again this year.
What I found was that both the public and private polls showed Clinton had a clear advantage in the presidential contest. In the final week, the consistency of those surveys was remarkable.
While those polls ultimately called the winner of the popular vote correctly (Clinton), they never conveyed the sense that the race would be a squeaker. Yes, each ballot test was within the margin of error of the final outcome, but the fact that they all missed in a single direction — toward Clinton — gave me greater confidence in the eventual margin than I should have had. The outcome never appeared so close that one candidate could win the popular vote while the other candidate would win a majority of the electoral vote. Yet, that is exactly what happened.
In most elections, the responses to survey questions about candidate qualities and vote preference predict the outcome. But this year was different.
What I didn’t understand, of course, was that enough voters in just the right combination of states were so frustrated and angry with the status quo that they simply were ready to burn down the building rather than try to rehabilitate it.
Voting for Trump, the candidate mocked by the media and disdained by the establishment, was the way to do that. Even if he was unqualified. Even if he had the wrong temperament. Even if he wasn’t more honest and trustworthy. Even if — maybe because — he was a vulgar, intemperate bully.
Second, I was completely wrong in my assessment that there weren’t enough white voters to elect Trump. While I was right that whites as a percentage of the electorate would drop to 70 percent, the lowest number in history, that fact turned out to be irrelevant, not predictive. Those whites who turned out gave Trump a larger margin than ever before. That was surprising, since most surveys during the fall showed Trump winning whites by 12 or 15 points, far less than Romney’s margin in 2012.
I had no reason to expect Clinton would underperform compared with President Obama’s 2012 showing among Latinos and Asian Americans — or that Trump would slightly outperform Romney among those two minority groups. All the national data indicated Trump might draw 1 in 5 Latinos, certainly nothing near 29 percent, as the exit polling found.
In a fundamental error, I also could not imagine there would be many Obama-Trump voters. But as a recent Washington Post piece showed, there were some of those voters in key states that decided the election.
Finally, you might ask, why didn’t I know that voters were so angry that they might upset the apple cart on Election Day?
In fact, I wrote repeatedly about the public’s dissatisfaction, anger and desire for change (citing the “right direction/wrong track” question). I knew there was anger, especially on the political right. The outcome of the GOP nomination fight made that quite clear.
But I also saw a Republican Party that nominated Romney, an establishment candidate, just four years ago, and I saw few signs of political revolution during the 2016 House and Senate primaries. And even with the public’s anger, I thought Trump was a deeply flawed — and ultimately unacceptable — messenger.
Having watched elections for decades and learned a few lessons about the American people, I didn’t think that the country would elect a nominee who said that he knew more than the generals about defeating the Islamic State, was endorsed by the Ku Klux Klan, would not release his taxes, belittled and bullied political adversaries, journalists and other public figures, and was churlish, undignified and narcissistic.
My view of the basic nature of the American electorate, which had been confirmed by decades of past election results and seemed to be confirmed by the polling data, proved to be incorrect this time.
The question is whether the indicators I used in the past are outdated and now lack any predictive value or whether the 2016 election was an aberration — a rare temper tantrum by part of the electorate — after which our politics returns to normal.
In my mind, that is now an open question.