Election Polling: What It Is and Why It Failed
To anyone who has been paying attention to the developments of the 2016 presidential campaign, it is apparent that polling entirely underestimated Donald Trump’s performance on November 8th. Criminal investigations, data leaks, and lopsided media coverage have all contributed to this historic upset, which defied almost every poll and expert prediction. To analyze the potential causes of this upset, it is important to have a basic understanding of the history and intricacies of election polling.
The first known instance of election polling dates back to 1824, in a local poll conducted by The Harrisburg Pennsylvanian. It showed that Andrew Jackson was leading his opponent, John Quincy Adams, with 335 votes over 169. While Jackson did not go on to win the presidency until the following election year of 1828, he did win the popular vote in the state of Delaware, where the poll was held, as well as the entire country that same year. After that instance, polling became increasingly popular as a method of election prediction. Polls remained localized, however, until The Literary Digest conducted a national survey wherein they predicted that Woodrow Wilson would be re-elected in 1916. The Literary Digest went on to correctly predict the elections of Warren Harding, Calvin Coolidge, Herbert Hoover, and Franklin D. Roosevelt.
Despite all of their accolades, The Literary Digest failed to realize the bias in their sample group of around 2.3 million voters. This led them to claim that Alf Landon was the more favored candidate compared to Franklin D. Roosevelt just a week before the election. Iowa native George Gallup, however, conducted a much smaller poll, wherein he sampled a demographically representative group of voters. Gallup went on to correctly predict FDR’s landslide victory.
Fast-forward to November 8, 2016. People all around the world were shaken to their core by a historic political upset accomplished by Republican presidential nominee Donald Trump. Social media outlets were taken by storm, and many people began comparing it to the so-called “Brexit” vote of June 2016, where the residents of the United Kingdom also defied pollsters and voted to leave the European Union. The consensus of national and state polls showed Hillary Clinton at a close, yet stable average distance of around 3+ points from her opponent just days before the election. Clinton was expected to win a majority of the battleground states, which would have pushed her towards her goal of 270 electoral votes, but an unexpectedly strong performance rewarded her opponent with 6 crucial states (WI, MI, OH, PA, NC, and FL).
Demographics played a large part in Trump’s unprecedented victory. Clinton overwhelmingly won the Latino vote, with 65% of that demographic voting in her favor. However, she underperformed compared to the performance of President Barack Obama in 2012 (71%) and 2008 (67%). Trump performed slightly better with this demographic than 2012 Republican nominee Mitt Romney (Trump received 29% of the Latino vote, Romney received 27%), but fell short of 2008 nominee John McCain, who received 31%. Although Hillary Clinton performed very well with the Hispanic demographic, it only makes up about 12% of the electorate in 2016, compared to 4% for Asians, another 12% for Blacks, and a significant 69% of Whites, which is what boosted Trump throughout the country, according to the Pew Research Center.
Many voters, however, are blaming the election of Donald Trump on other causes, such as mainstream media coverage. Starting with the primary season earlier in the year, the idea of media bias, shown through disproportionate and favored coverage of a certain candidate, extended into the presidential race. The second presidential debate drove this argument further, as voters felt the questions were unfairly skewed in favor of Clinton. Some say the letter from FBI Director James Comey to Congress, in which he informed them of “the existence of emails that appear to be pertinent to the investigation”, may have hurt Clinton’s chances.
Another theory floating around suggests that media sources overhyped the Clinton camp’s polling numbers and their chances of winning. For most of the election, reputable polling analysis site FiveThirtyEight gave Trump the best odds at winning compared to The New York Times or The Wall Street Journal. Nate Silver, founder and editor in chief of FiveThirtyEight says “our models are based on how accurate polls have or haven’t been historically, instead of making idealized assumptions about them.” In all likelihood, late deciders are what helped put Trump over the top. In many of the battleground states, Trump gained an overwhelming amount of voters who remained undecided until the last minute, which could well be a reason why the results told a different story than the polling. There was an estimated 7 percent of likely voters who were undecided.
Sen. Bernie Sanders of Vermont, who lost the Democratic Party nomination to Hillary Clinton, issued a statement after the election saying “Donald Trump tapped into the anger of a declining middle class that is sick and tired of establishment economics, establishment politics, and the establishment media.” — a statement that many Americans can relate to. Protests of President-Elect Trump show another side of how people feel, but young Americans know there is still hope. The 2018 midterm elections will give voters a chance to elect new candidates all over the country. While polling can never always accurately predict the results of an election, it is nonetheless a useful method of election forecasting. In this case, it just so happened to be that the underdog came out on top.
