If you have read any election forecasting models, such as The New York Times, FiveThirtyEight, Huffington Post, Real Clear Politics, etc., you would assume the election is over. As of Wednesday, the Times predicted with 23 percent certainty that Clinton will win the presidency. According to the same election forecaster, that gives Clinton a better chance of winning the election than Trump’s chance of wining Mississippi. Let me remind you that Mitt Romney won Mississippi by 12 points last election. It would seem logical to conclude that Clinton will win the election handily, but with one caveat: the election has not even occurred yet.
The respectable major newspapers and news sources, or as Trump refers to them as “the corrupt mainstream media,” have already assumed a Clinton victory. NPR’s Domenico Montanaro wrote an article titled “NPR Battleground Map: Hillary Clinton Is Winning — And It’s Not Close.” If any Democrat looks at that beautifully illustrated map, with its plethora of big blue squares, he or she would find comfort in the lack of red.
The Washington Post featured an article titled “Donald Trump’s chances of winning are approaching zero.” This American map featured even more blue states, and far less red ones. However, these maps and many of the election forecasting models rely on polls.
Polls are those annoying statistical methods to gauge which people favor which thing; in this case, who is voting for Clinton, Trump or another candidate. The problem with polls is that there is a significant margin of error that applies to each candidate, typically around 3 percent. For example, if a poll with a margin of error of 4 percent puts Clinton at 48 percent and Trump at 40 percent, then a tie would be within the margin of error. In other words, an eight-point lead in a poll could in reality be a tie. Granted, that is one extreme case, but polling itself is flawed.
“Unfortunately, polling in general is facing a crisis of sorts: The methodology that has worked in the past is not holding up in today’s digital era, and it’s hard to distinguish a quality poll from a dud,” Dan Centinello, the executive vice president of the global political consulting firm Lincoln Strategy Group, wrote in a U.S. News op-ed. “As has been true throughout history, polling mechanisms are worthy of a healthy dose of skepticism if we want them to evolve and improve.”
“Rasmussen Reports and the historically most accurate Investors Business Daily Poll both say it’s just a one point race,” Jake Novak of CNBC wrote. “Two other polls, Reuters/Ipsos and Economist/YouGov have it as a four-point lead for Clinton. This is hardly landslide or even comfortably certain territory.”
Both quotes demonstrate the importance of taking polls and forecasting models with a grain of salt. This election has certainly not been a typical one. It is difficult to predict which demographics will go to the polls in big numbers. Moreover, nobody knows how many “closeted” Trump supporters, or even “closeted” Clinton supporters, there are.
The purpose of this article is primarily not to dismiss the predictions that Clinton will win. Rather, it is to criticize many news sources for not explaining the problems of polling, and for not following the issues. We hear polls like “Clinton up 4 points in Florida” and “Trump only up 4 points in Texas,” but for what purpose? It is for the media to abandon horse-race politics, however tempting it is to read and write about. While an article about the rising premiums caused, at least in part by, the Affordable Care Act will get less “clicks” than one about a new poll, the former article serves the reader more.
