The Polls Got It Wrong In 2016. What Now?

Noa Yolkut, Grade 10, Entertainment/Features Editor

I count my blessings every single day that I am an American,” said Hillary Clinton, several hours after she lost the 2016 presidential election. “And I still believe as deeply as I ever have that if we stand together and work together with respect for our differences, strength in our convictions, and love for this nation, our best days are still ahead of us.” America was bewildered. What just happened?


The defects of the 2016 polling data have become infamous in our country’s political history. In 2020, one of the only things the Trump and Biden campaigns did similarly was use the suggestion of inaccurate polling to motivate their supporters to vote. But were the polls really accurate this year?


Let’s start with the basics: what is a poll? According to FiveThirtyEight’s Database Journalist, Dhrumil Mehta, “Polls are a survey taken of a sample of voters that helps us guess how a population as a whole might vote.” The 2016 Wisconsin polls help put this into perspective. Before election day, the polls showed that 49.6% of Wisconsin’s voters would vote for Secretary Clinton and 44.3% would vote for Trump. In the end, only 46.5% of Wisconsinites voted for Clinton, while 47.2% voted for President Trump. How did the polls get the numbers so wrong?


When pollsters select a random number of people to poll, they may get more Democrats than Republicans, women than men, 20-year-olds than 50-year-olds, and people of color than white people. When this occurs, pollsters weight their polls. They adjust the poll data in an attempt to ensure that the sample more accurately reflects the characteristics of the whole population. The Wisconsin upset forced pollsters to reexamine their methods, and that’s when they realized their mistake: they never weighted for education.


The impact of their mistake was huge. A person’s level of education often correlates to what party they affiliate with and for whom they will vote. Pollsters in 2016 unwittingly used more data from college-educated people than non-college-educated people, and, therefore, did not accurately measure the results. Since the polls disproportionately recorded the responses of college graduates, it seemed as though Clinton would win. This, paired with the fact that most people who polled as undecided ended up voting for Trump, was why there were huge upsets in states like Wisconsin, Michigan, and Pennsylvania.


In 2020, pollsters weighted by education and fewer people polled as undecided, so the margin of error was smaller.


This made it appear as though polls should have been in favor of President Trump, but the 2020 Democratic candidate was notably different. In 2016, polls found that Secretary Clinton was -11 points among non-educated voters, whereas Vice President Biden was +3 among that same group. The polls that said Clinton was ahead nationally by 4.8 points, said that Biden is up by 10.8 points on October 17th.


So, were the polls accurate in 2020? Yes and no. If we look at each state only as voting blue or red, then the polls were accurate. They correctly predicted Wisconsin, Michigan, Pennsylvania, and Georgia going to Biden, but were wrong about North Carolina and Florida. The reason the polls were not wholly correct was because of the margins of victory. Let’s look at Wisconsin. On November 1st, CNN gave Biden a 10 point lead in Wisconsin, but by November 8th, Biden won the state by 0.7%.


Were the polls right? No. Were the polls wrong? Also no. What is definite is that it will take a long time for pollsters to get back on their feet, and it will take America a long time to trust them again.