How did it happen? In an election characterized by outright lies, the biggest liars of all were the polls which predicted a reasonably comfortable victory for the less flawed candidate, Hillary Clinton. I thought that night, as the polls lay in shreds, of Benjamin Disraeli’s celebrated words, “There are three kinds of lies: lies, damned lies, and statistics.” Since political polls make powerful (if often incorrect) use of statistics, the saying applies forcefully to this election.
Why were the major predictive polls so far off? Exit polls, which try to record only what happened and offer insights into why it happened, as opposed to predicting what will happen, offer some clues. But even they are not fully revealing, as they are conducted in so-called bellwether voting precincts considered historically predictive of the outcome of an election, and not selected randomly. So for example, the early take on the lower Hispanic vote for Clinton has been challenged by other polling organizations, which show it was higher.
What the exit polls seem to show are the divisions that were predicted all along. First, the US population is divided sharply by class, education and income (one being the byproduct of the other two), by generational gaps, by urban vs. rural residence, and by gender although class, education, and race complicate gender differences. Overall, Trump won 53% of the male vote to Clinton’s 41%. Clinton won 54% of the female vote to Trump‘s 42%. But, a surprise, to some of the pollsters, was that more white women voted for Trump (53%) than Clinton, which meant that fewer women overall voted for Clinton than for President Obama in 2012. This probably reflects that some white women stayed home in the Republican Party, which was, I think a late-developing trend that worried some of the pollsters.
As a rule, younger voters favored Clinton while older voters liked Trump. Clinton took 55% of the millennial vote (age 18-29) to Trump’s 37%. This implies that 8% of the millennial vote went to third-party candidates as protest votes. Moreover, the real question is whether the young turned out in force as they did for Obama in 2012. (Another question is how many of those young people who are now protesting Trump’s election in a number of US cities could have brought an “I voted” sticker to the protest.) About 55% of the electorate voted. Turnout in 2012 was only a shade higher, just over 57%, But one wonders whether one percent of the voters, about 1.5 million more votes, would have made a difference in the outcome. Probably not if they broke as the actual votes cast did. But clearly, a larger turnout of millennials breaking 55% to 37% might have.
Education and income, more than any other of the exit poll results, underscore the deep divisions in American society. Trump earned 51% of the votes of those without a college degree, eight percentage points more than Clinton. College graduates voted 52% to 43% for Clinton. When parsed by race and income, however, the figures take on a different look: Trump won whites without college degrees 67% to 28%, while Clinton won non-whites with or without degrees by over 70%. Incomes also show disparity but with some wrinkles: Lower income voters (up to 50,000) gave Clinton large majorities, while middle and higher income voters (50,000 to 250,000) broke almost evenly but with Trump getting small majorities.
It is with the whites without college degrees that Trump made his major gain. His success with that group enabled him to win the White vote overall by a whopping 31 percentage points. In 2012, Romney had won the White vote by 20 percentage points over Obama, which was obviously not enough to win that election. Clearly portions of the white working class vote that had been traditionally Democratic went to Trump in this election. It seems that he either motivated voters who had dropped out of the political process some time ago to return with his vision of a return to a “golden age” (this should sound familiar to many Pakistanis), or he simply tore voters in this class away from their Democratic roots. Probably both were a factor, but there is no agreement yet among the various expert pollsters, still trying to figure out how they went so wrong. If it is the former, that would explain some of their error given the reappearance of voters who had slipped off the radar screens as likely voters years ago.
There was more to it, however, than just a surge of uneducated white voters looking to return to their golden age of the 1950s and 1960s. There were surges predicted which were supposed to offset this generally understood, if underestimated surge of white working class voters. Where was the surge of women voters to Clinton that was predicted, particularly after Trump’s vulgar talk about his sexual predations in the tape that was revealed a week before the third debate, and which Ms. Clinton made much of in that debate. She won the women’s vote by about 12 percentage points over Trump, but President Obama won women by 11 points in 2012. No surge there. Where was the Hispanic surge we all expected after all the pejorative comments Trump had made about Hispanics? While the exit polls are probably wrong in their conclusion that Hispanics supported Trump more than expected, clearly if there were an increase in support for Clinton, as I believe there was, it was not enough, possibly because turnout was not as great as expected. So Ms. Clinton’s problematic candidacy and/or campaign probably was a larger factor than it was thought to be at the time.
I think that there are two exit poll results that are most important in judging this election. First: the most important is candidate quality. Ms. Clinton won three out of four on this one very decidedly. This is about which candidate has the needed experience; which cares most about me; and which has good judgement. Trump won only one big time: which candidate can bring change. But in the election booth, change trumped experience, judgement and caring. Second, I find the answers to the last question on the exit poll list astounding—that a great number of voters say they had made up their minds in September or even earlier; thus it is possible that nothing in the rest of the campaign, the debates the revelations on Trump’s sexual predations, or the continuing email scandal made any difference to the result.
That leads us to demographics. The White vote, which was 84% of the total electorate in 1984, has shrunk to 70% this year (it was 72% in 2012). It will continue to shrink, under normal circumstances (more on this below). This shrinking majority of whites has made pollsters and forecasters believe that it will soon be out-voted by the growing minority communities. In my home state of California, the largest state in population and electoral votes, the population is now about evenly split. The white population is aging and decreasing, while minority communities are young and growing. They are also coalescing politically as well as gaining in income, education, and influence.
The writing is, or at least was, on the wall for the white majority. The question we have to ask is will the Trump administration try to stop this inevitable demographic wave? It would take an authoritarian government and extra-constitutional measures. But if change is what the voters want, authoritarianism and extra-constitutional measures may not be as far-fetched as they seem right now. The US is the world’s leading and bellwether liberal democracy. Our institutions and political resiliency may be sorely tested in the coming four years.
The author is a Senior Scholar at the Woodrow Wilson Center in Washington DC, and a former US diplomat who was Ambassador to Pakistan and Bangladesh
Why were the major predictive polls so far off? Exit polls, which try to record only what happened and offer insights into why it happened, as opposed to predicting what will happen, offer some clues. But even they are not fully revealing, as they are conducted in so-called bellwether voting precincts considered historically predictive of the outcome of an election, and not selected randomly. So for example, the early take on the lower Hispanic vote for Clinton has been challenged by other polling organizations, which show it was higher.
What the exit polls seem to show are the divisions that were predicted all along. First, the US population is divided sharply by class, education and income (one being the byproduct of the other two), by generational gaps, by urban vs. rural residence, and by gender although class, education, and race complicate gender differences. Overall, Trump won 53% of the male vote to Clinton’s 41%. Clinton won 54% of the female vote to Trump‘s 42%. But, a surprise, to some of the pollsters, was that more white women voted for Trump (53%) than Clinton, which meant that fewer women overall voted for Clinton than for President Obama in 2012. This probably reflects that some white women stayed home in the Republican Party, which was, I think a late-developing trend that worried some of the pollsters.
What the exit polls seem to show are the divisions that were predicted all along. Education and income, more than any other of the exit poll results, underscore the deep divisions in American society. It is with the whites without college degrees that Trump made his major gain. His success with that group enabled him to win the White vote overall by a whopping 31 percentage points
As a rule, younger voters favored Clinton while older voters liked Trump. Clinton took 55% of the millennial vote (age 18-29) to Trump’s 37%. This implies that 8% of the millennial vote went to third-party candidates as protest votes. Moreover, the real question is whether the young turned out in force as they did for Obama in 2012. (Another question is how many of those young people who are now protesting Trump’s election in a number of US cities could have brought an “I voted” sticker to the protest.) About 55% of the electorate voted. Turnout in 2012 was only a shade higher, just over 57%, But one wonders whether one percent of the voters, about 1.5 million more votes, would have made a difference in the outcome. Probably not if they broke as the actual votes cast did. But clearly, a larger turnout of millennials breaking 55% to 37% might have.
Education and income, more than any other of the exit poll results, underscore the deep divisions in American society. Trump earned 51% of the votes of those without a college degree, eight percentage points more than Clinton. College graduates voted 52% to 43% for Clinton. When parsed by race and income, however, the figures take on a different look: Trump won whites without college degrees 67% to 28%, while Clinton won non-whites with or without degrees by over 70%. Incomes also show disparity but with some wrinkles: Lower income voters (up to 50,000) gave Clinton large majorities, while middle and higher income voters (50,000 to 250,000) broke almost evenly but with Trump getting small majorities.
It is with the whites without college degrees that Trump made his major gain. His success with that group enabled him to win the White vote overall by a whopping 31 percentage points. In 2012, Romney had won the White vote by 20 percentage points over Obama, which was obviously not enough to win that election. Clearly portions of the white working class vote that had been traditionally Democratic went to Trump in this election. It seems that he either motivated voters who had dropped out of the political process some time ago to return with his vision of a return to a “golden age” (this should sound familiar to many Pakistanis), or he simply tore voters in this class away from their Democratic roots. Probably both were a factor, but there is no agreement yet among the various expert pollsters, still trying to figure out how they went so wrong. If it is the former, that would explain some of their error given the reappearance of voters who had slipped off the radar screens as likely voters years ago.
There was more to it, however, than just a surge of uneducated white voters looking to return to their golden age of the 1950s and 1960s. There were surges predicted which were supposed to offset this generally understood, if underestimated surge of white working class voters. Where was the surge of women voters to Clinton that was predicted, particularly after Trump’s vulgar talk about his sexual predations in the tape that was revealed a week before the third debate, and which Ms. Clinton made much of in that debate. She won the women’s vote by about 12 percentage points over Trump, but President Obama won women by 11 points in 2012. No surge there. Where was the Hispanic surge we all expected after all the pejorative comments Trump had made about Hispanics? While the exit polls are probably wrong in their conclusion that Hispanics supported Trump more than expected, clearly if there were an increase in support for Clinton, as I believe there was, it was not enough, possibly because turnout was not as great as expected. So Ms. Clinton’s problematic candidacy and/or campaign probably was a larger factor than it was thought to be at the time.
I think that there are two exit poll results that are most important in judging this election. First: the most important is candidate quality. Ms. Clinton won three out of four on this one very decidedly. This is about which candidate has the needed experience; which cares most about me; and which has good judgement. Trump won only one big time: which candidate can bring change. But in the election booth, change trumped experience, judgement and caring. Second, I find the answers to the last question on the exit poll list astounding—that a great number of voters say they had made up their minds in September or even earlier; thus it is possible that nothing in the rest of the campaign, the debates the revelations on Trump’s sexual predations, or the continuing email scandal made any difference to the result.
That leads us to demographics. The White vote, which was 84% of the total electorate in 1984, has shrunk to 70% this year (it was 72% in 2012). It will continue to shrink, under normal circumstances (more on this below). This shrinking majority of whites has made pollsters and forecasters believe that it will soon be out-voted by the growing minority communities. In my home state of California, the largest state in population and electoral votes, the population is now about evenly split. The white population is aging and decreasing, while minority communities are young and growing. They are also coalescing politically as well as gaining in income, education, and influence.
The writing is, or at least was, on the wall for the white majority. The question we have to ask is will the Trump administration try to stop this inevitable demographic wave? It would take an authoritarian government and extra-constitutional measures. But if change is what the voters want, authoritarianism and extra-constitutional measures may not be as far-fetched as they seem right now. The US is the world’s leading and bellwether liberal democracy. Our institutions and political resiliency may be sorely tested in the coming four years.
The author is a Senior Scholar at the Woodrow Wilson Center in Washington DC, and a former US diplomat who was Ambassador to Pakistan and Bangladesh