2016 Election Polls and Predictions: What Went Wrong?
During and in the aftermath of this year’s election night, one emotion was widespread among both major political parties: surprise. And no wonder; The New York Times had Clinton’s chances of winning at 85%, The Huffington Post had them at 98%, and several major news outlets (including Associated Press and NPR) also had her winning by an average of 76.17 electoral votes. And to top things off, Nate Silver, notable statician and editor-in-chief of ESPN’s highly influential FiveThirtyEight, had Clinton’s chances of winning at 71.4% right up until tellers began counting the ballots. In short, nearly all evidence-backed election forecasts had Clinton winning by a clean margin. As a result, the nation’s level of shock grew as it became increasingly clear that not only was Clinton not going to win the electoral college by a significant amount of votes, but Trump was also on his way to collecting most of the swing states’ electoral votes and eventually winning the presidency. To many, this was an indication of flaws in pre-election polling processes, but to others, this was yet another example of the liberal bias pervading today’s media.
One possible reason for the “misleading poll predictions” that has been publicized by Aradhna Krishna is that “societal pressures…kept [Trump supporters] from declaring their true intention[s]”, even in almost completely anonymous surveys (Krishna). According to Krishna, although embarrassment is normally thought of as “something that happens in public”, it in fact can still occur as the result of “actions made in private” (Krishna). This can be logically explained by what has been covered in lectures as well, and in particular the nuances of crowd behavior. To a single individual that may personally intend to vote for Trump, there is no direct benefit from voicing so; in fact, it is likely that the opposite is true instead. Partly due to the media’s liberal bias, a voter in favor of Trump is more susceptible to the criticism and judgment of others, even if the voter him- or herself is part of a demographic that Trump has attacked. As a result, the actual direct benefit to the individual comes from following–or, what may have happened in this case, pretending to follow–what is viewed as popular opinion. Krishna is one of many who believe that this contributed to the misleading skew of polls leading up to the night of the election.
Also exposing this influence of crowd behavior is Trent Lapinski in his very recent opinion piece on Medium. In it, Trent Lapinski laments what he coins as the “echo chamber” effect–that is, the consequences of “the media[] and social media ke[eping] everyone…isolated from differing opinions” (Lapinski). He argues that “getting the news from just your friends is a logical fallacy”, and that “you need to know your enemies” (Lapinski). Furthermore, he writes, doing otherwise (i.e. “default[ing] to what the media” says) has visible consequences: it can lead to extremely incorrect estimates of the chances that one candidate will win the presidency (Lapinski). Lapinski’s concerns, too, are supported by some of the concepts recently covered in this course. Information cascade theory, for one, can help explain the so-called “echo chamber” effect. It is not just that people tend to want to follow group behaviors in order to fit in, but also that when nearly all of the political news someone receives leans towards one side, that individual’s expected strategy is to also publicly lean towards that side, regardless of his or her own information. Such a phenomenon is dangerous to healthy political debate, and this election’s poll inaccuracies prove to be one of several indications that change is necessary.
Sources: