What Happened during the 2016 Presidential Election?

Like mentioned in the book, the 2016 Presidential Election is a great example on why polling is not always perfect predicting an outcome of an election. However, a lot of the news organizations and scholars use polling information as a way to gain information and try to determine elections before they even occur. However, are pollsters and other research groups now have too much information to be effective (predicting the right outcomes)?

Last New York Times poll going into the Election

In this previous Presidential election, candidate Hillary Clinton had a lead of 1 – 7 percentage points in the popular vote based on polls. However, as we know now, the information the polls gathered predicted the outcome of the election wrong especially in terms of the electoral vote. According to the Pew Research Center, they identified three ways why the polls varied in results during the 2016 election. However, it is very tough to determine which of the three played the biggest factor in the results of the pre election polling.

The first way the Pew Research Center identified inaccurate polling is through a common sampling error called nonresponse bias. This occurs when individuals who are invited to express their opinion chose not to respond despite equal opportunity. The Pew Research Center suggests that pollsters had trouble getting responses from Trump supporters because his campaign targeted those who were less educated. Furthermore, according to NPR, they claim that due to his supporters being less educated, they are less likely to answer or stay on the phone when pollsters call.

The second explanation the Pew Research Center identified is that some people were “shy Trumpers.” This term describes individuals who lied to pollsters who were asking them their candidate preferences because it was deemed socially unacceptable to vote for Donald Trump. In states which are mainly Democratic, someone could definitely be a shy Trumper because of the fear of being rejected by the public for his or her views.

The last explanation the Pew Research Center identified has to do with the system that pollsters used to determine the sampling of individuals. Like most polls, the pollsters try to determine a sample which is representative of the group of people who are going to show up on election day. However, in the 2016 Presidential Election, the individuals who represented the Midwestern states did not show up to the polls on election day like most of the pollsters predicted. As a result, a large segment of voters which many thought were going to be the most important did not actually come to the polls.

Recently, polling has been turned into such a science with all the different types of forms of measuring public opinion. People can receive polls in the form of calls, texts, emails, websites, social media, face-to-face, mail, etc… As a result, does having too much information actually make poll results more inaccurate?

This video is CBS talking about why their polls and other pollsters were off on the 2016 Presidential Election:

Sources:

http://www.pewresearch.org/fact-tank/2016/11/09/why-2016-election-polls-missed-their-mark/

https://www.npr.org/2016/11/14/502014643/4-possible-reasons-the-polls-got-it-so-wrong-this-year

https://www.nytimes.com/interactive/2016/us/elections/polls.html

https://www.youtube.com/watch?v=RCgbeBE42JI

4 Responses to "What Happened during the 2016 Presidential Election?"

Jennifer Munnings says:

February 24, 2018 at 6:36 pm

Thank you for your insightful post! From reading this, I am left wondering about different ways in which pollsters could ensure that they have an accurate sample of the population in order to reliably and accurately determine outcomes. To what extent does personal bias influence who is being polled, and then how does that affect the information broadcasted? I am still a little bit confused however about what you mean when you say too much information. Do you mean the idea of cognitive dissidence having a negative affect on polling?
Cassandra Martin says:

February 25, 2018 at 9:53 pm

Holden,

I really appreciate your analysis of how the 2016 Presidential Election was poorly predicted in polling, and the ways in which polling failed specifically in this election. I think the part about the Midwest’s effect on altering the electoral college vote for example by certain populations not showing up as they normally do was especially insightful.

However, my primary criticism of this post is your lack of development upon the idea of having “too much information”. This doesn’t seem very related to the rest of your post, and I am confused on what you mean. If you meant that pollsters collect too much information, or make too many predictions, I would have to disagree. Making such a claim would require more statistical evidence than was provided, as well as examining other cases of unpredicted political results. As you discuss in the three aspects that had the most influence on the deviance of predictions from the actual results, it can’t be determined which of the three was the biggest issue. Furthermore, I would actually argue the opposite of whether enough information was gathered. If the typical political involvement and likeliness to participate of either candidate’s supporters in each type of polling had been considered, perhaps it could have been predicted that the actual distribution of voters was not accurately representative of the population. Therefore, I would say that in the next election, it would be beneficial for pollsters to more thoroughly examine the meaning behind the polling results, and consider based on additional information, if it is actually representative of the entire voting population.
Brett Tivnan says:

February 25, 2018 at 10:58 pm

Hi Holden,
I felt that some of your points about the issues in collecting poll data were spot on, especially on how you pointed out how surprising it was that Hilary Clinton did not gain any percentage of the women’s vote over President Obama. Poll data relies heavily on previous voter tendencies and unreliable statistics to forecast outcomes, but as the 2016 election pointed out people are individuals and every decision they make is unique from all others. Though one person may have held certain beliefs in a past election, a lot can change in four years and many people don’t know what they will do until the choice is put in front of them.
Though you articulated some points well, I never really felt like there was a central argument to all of the anecdotes you brought up. Your post consisted mainly of restated information from the Pew Research Center and I didn’t really feel like the information was used to discuss anything. Your post ended by asking the question of whether too much polling information can be a bad thing, but I feel like you should have been the one to answer that in your post.
Akshit Sharma says:

March 1, 2018 at 12:09 am

Holden, I enjoyed reading your post about the last presidential election. To answer the question that you pose at the end of the post, I would argue that having too much information is not the reason for poor accuracy of polls, however, having too much conflicting and/or poor quality information seems to me to be the real issue. As an example, there have been multiple claims of Russian backed pro-Trump and anti-Clinton information floating around before the election that led to the polls proving to be inaccurate since Trump ended up winning instead of Clinton, who was predicted to win. When people’s views are shaped by flawed foreign influences, they try and make the best decision that they can by drawing their own conclusions. It seems to me that the general feeling among voters (not the population) was that Trump seemed to be a better bad option than Clinton, based on the flawed information available to them, and hence Trump became President.

Comments are closed.