The Failure of Public Opinion Polling in 2016

As we discussed in class, public opinion polling has recently become one of the most important aspects of modern politics. Its ability to gauge public opinion on issues, policies, and candidates has allowed politicians and government officials to tailor their policy choices to the electorate. This is done both in order to pass the most favored policy and, more cynically, to maintain their offices in an increasingly divisive political environment. However, polling faces a wide variety of issues. These issues came to public prominence during and immediately following the 2016 Presidential Election, during which all polling seemed to fall flat. The issues present in the 2016 election seemed to be deeply exacerbated versions of the issues presented in the textbook.

One of the most prominent issues in public opinion polling is sampling. This was one of the biggest problems pollsters faced in the 2016 election for a wide variety of reasons. As discussed in the textbook, fewer and fewer people use landline phones and polling via cell phone is much easier to avoid. This results in the reduced randomness of the sample, as those with landlines can be overrepresented. Similarly, as technology becomes more advanced, internet polling has become more common. Internet polling brings with it a myriad of problems, especially due to high levels of non-response and only the most passionate taking the time to actually take the poll. These systemic issues in the field of opinion polling have shown a massive under-representation of the Trump voter base. As many in Trump’s base of support in swing states were lower income and less educated, they would inherently mistrust the polling agencies and refuse to participate in the poll (Mercer). Other issues in the 2016 election were what Mercer of the Pew Research Center called “shy-Trumpers” who voted for the President, yet did not express this support publicly, and the inability of pollsters to predict which demographics will vote and in what quantity. In this way, many people who opposed President Trump may have decided to stay home due to their disdain for Secretary Clinton or other deciding factors.

The biggest issue, however, was the failure of many pollsters to consider the possibility of a popular vote victory, but an electoral college defeat. Newkirk of The Atlantic writes that the discrepancy between electoral results and the popular vote has grown much wider over the course of the past 50 years, an issue which is abundantly clear in the elections of 1972, 1984, 2000, and 2016. Many pollsters, like Nate Silver, took this issue and many others into account in determining their polling model. Silver, the founder of FiveThirtyEight, aggregated all polling his system deemed credible and used the electoral college and polling trend lines to determine the candidate’s chance of winning. While his model was still massively inaccurate in 2016, his deeply scientific method was still the most successful of any major media outlet and any failure can be contributed to these glaring, systemic issues in public opinion polling.

Sources:

Mercer, Andrew, et al. “Why 2016 Election Polls Missed Their Mark.” Pew Research Center, Pew Research Center, 9 Nov. 2016, www.pewresearch.org/fact-tank/2016/11/09/why-2016-election-polls-missed-their-mark/.

Morris, David Z. “What the Latest Nate Silver Controversy Teaches Us About Big Data.” Fortune, Time Inc. , 6 Nov. 2016, fortune.com/2016/11/06/nate-silver-controversy-big-data/.

Newkirk, Vann R. “What Went Wrong With the 2016 Polls?” The Atlantic, Atlantic Media Company, 9 Nov. 2016, www.theatlantic.com/politics/archive/2016/11/what-went-wrong-polling-clinton-trump/507188/.

Lotte Elfering says:

October 12, 2017 at 12:10 pm

This is a sound post, analyzing all that was wrong with the polling during the 2016 presidential election. I believe it is good to analyze the problems, but I am a bit skeptical about what can be done to combat them. If a populist candidate like Donald Trump is involved, who seemed to be widely disliked in the media and whose voters were painted as deplorable people, many people will not want to voice their opinion openly, even if it is anonymous. This problem can thus not be solved.

Furthermore, there seems to be no method to properly poll people if one tries to avoid landlines, the internet, and cellphones, and if one were to go door to door it will still be hard in several clearly important rural areas. Every method of polling thus seems to have its downsides, for which a remedy does not exist. My question all throughout the class and while I was reading the chapter at home was whether there was any way in which pollsters could develop a representative polling method, and I realized there was none. For now, it will remain necessary to combine several methods to get as representative a sample as possible, as this is the only way possible.

Finally, of course there is no way to counter the idea that people do not trust the polling agencies, as it is difficult for people who do not trust an institution to turn around and trust them. I would say this is the biggest challenge for pollsters in this time of new political developments. Polling will always be slightly skewed, as we talked about in class today, and there is no one perfect way to get a perfect sample, but if we combine multiple methods of polling, there might be a way to get as close as possible to the most representative sample.

Stephanie Hughes says:

October 12, 2017 at 12:59 pm

During the 2016 election, I saw a poll that showed Donald Trump having a 1% chance of winning the Presidency. How can a poll, meant for public consumption, be so far off? This was an extremely irresponsible poll to make public, as it may have influenced Clinton supporters not to come out and vote. However, based on the media’s portrayal of the election, it really did seem like Donald Trump had a little to no chance of being elected. 99% of the readily available content on all forms of social media was Anti-Trump. I think this is an example of a population being over-represented. The people with the largest voices – i.e. celebrities – were largely Clinton supporters. In fact, I think being a Trump supporter would actually make a social pariah in many circles. As you mention in your post, this led to the phenomenon of “shy Trumpers”, who did not publicly express their opinions, but voted for the Republican candidate. It seems that misrepresentative polling and pop culture together truly led to the failure of public opinion polling in the 2016 election.