How did pollsters get the 2016 US election so wrong?

Zahedur Rahman Arman | Tuesday, 29 November 2016

The results of the 2016 presidential election came as a surprise to everyone who had been following election polling. The day before the election, most pollsters and statistical models had pegged that Democratic party candidate, Hillary Clinton, was destined for victory, her chances of winning ranging from 70 per cent to as high as 99 per cent (Mercer, Deane, & Mcgeeney, 2016). They also pegged her the heavy favourite candidate to win a number of democratic states like Michigan, Pennsylvania and Wisconsin that were ultimately won by Trump, the Republican candidate. The Republican candidate's victory dealt a devastating blow to the credibility of the leading pollsters, including Princeton Election Consortium, Huffington Post, Daily KOS, CNN, Predict Wise, New York Times, and FiveThirtyEight. Of 67 national polls tracking a 4-way race since the start of October, only four gave Trump the lead, according to the RealClearPolitics. Of 61 national polls tracking a 2-way race during that period, six gave Trump the lead (Bomey, 2016). Thus, there was a growing expectation that Hillary was winning the election, but the result was totally different. As Larry Sabato, director of the University of Virginia's Center for Politics, said, "We were wrong. The entire punditry industry, the entire polling industry, the entire analyst industry - and I want to use this opportunity to take my fair share of the blame - we were wrong," (http://insider. foxnews.com/ 2016/11/09/how-did-polls-experts-get-2016-election-so-wrong). 
The article will try to show how the media and pollsters got the election so wrong. There are several reasons behind the inaccurate data. These were: the methodological problems, non-response bias, undecided voters, the mentality and behaviours of Trump's supporters, and close polling margins. 
METHODOLOGICAL PROBLEMS: Acquiring accurate random samples of voters to poll has been creating continuous problems since cell phones and online surveys came into wide use. Prior to the use of these tools in the survey, pollsters used landline telephones to find out reasonably random and representative samples. Pollsters could just pick random names from the phone books, call voters, and take interviews, which helped them to analyse voters' alignment. This method also ensured high response rates and helped to control non-response bias (when the answers of respondents differ from the potential answers of those who do not answer). The problem with using mobile phones as survey tools is that they are not usually publicly listed, making it difficult to find representative samples. Some online survey methods also suffer from bias and are generally considered of lower quality than other polls. As an example, L.A. Times/USC polls established panel of online voters could gain traction. Declining rates of voter response to phone polls and increased cell phone usage are making it increasingly difficult and costly to poll voters through traditional methods.
It is also true that with the rise of the Internet, researchers and pollsters can find more reliable and detailed information about the public. They can define the public's mind based on the information they have collected from the respondents. At the same time, there is also the possibility of losing reliable sampling data. 
NON-RESPONSE BIAS: In these election polls, non-response bias is also a factor that occurs when some people do not respond to surveys. In this election, the less educated voters were a key demographic for Trump. Most of the pollsters did not reach these voters. It is possible that the frustration and anti-institutional feelings that drove the Trump campaign may also have aligned with an unwillingness to respond to polls (Mercer, Deane, & Mcgeeney, 2016). Thus, in most of the election polls, pollsters didn't predict a Trump victory in the 2016 U.S. presidential election. 
UNDECIDED VOTERS SWUNG BACK: In the election, a late surge of undecided voters secured Trump's victory over Clinton. According to HuffPost Pollster, Clinton was at 46 per cent in the Wisconsin polls, and won just less than 47 per cent of the vote. She was at 46 per cent in the Pennsylvania polls and won 47.6 per cent of the vote. In Michigan, she was at 47 per cent in the polls, and got 47 per cent on the election day (Shepard, 2016). On the other hand, Trump was at 40.8 per cent in the polls, and got 47.8 per cent of the vote. In Pennsylvania, he was 41 per cent in the polls, and got 48.79 per cent of vote. In Wisconsin, he was at 40 per cent in the polls, and got 47.87 per cent of votes on the election day. 
From this data, it is clear that Clinton got votes that the pollsters expected. The Republican candidate, Trump got more votes than the pollsters expected. Thus, it is clear that undecided voters swung back to the Republican candidate. 
TRUMP SUPPORTERS MAY THINK THAT IT'S A RIGGED POLL: In the election campaign, Trump was continuously saying that the election was rigged. Trump supporters thought that the polls were rigged as well. According to Timothy Johnson, professor and director of the Survey Research Laboratory, University of Illinois, said, "People who don't like the government often perceive the polls as being part of the government." As a result, some voters didn't participate in the election polls which is the most plausible explanation for the pollsters' miss.  
SHY TRUMP SUPPORTERS: In general, Clinton supporters are more likely to say they're a Clinton supporter than Trump supporters are to say they're. As the University of South California Dornsife/Los Angeles Times pollhttp:/www.latimes.com/ politics/la-na-pol-usc-latimes-poll-20161108-story.html corroborated, "Women who said they backed Trump were particularly less likely to say they would be comfortable talking to a pollster about their vote" (http://theconversation.com/voters-embarrassment-and-fear-of-social-stigma-messed-with-pollsters-predictions-68640). As a result, pollsters got wrong messages about the 2016 presidential election. 
NOT READY FOR FEMALE PRESIDENT: It may also be the case that America is not yet ready to accept a female president. Americans did not feel comfortable to the pollsters until they reached the voting booth. It's difficult for the pollsters to make a decision about such voters. There has never been a female presidential candidate from a major political party before. In general, people are reluctant to embrace change in the presidential election.  
THE CLOSE MARGIN: Trump's victory was decisive by electoral college, but it wasn't a landslide. He picked up Michigan, Pennsylvania, and Wisconsin, shocking pundits. However, Donald Trump won three rustbelt states by only a combined total of 112,000 votes. Pollsters did not find such close margin in their election polls. They thought that victory was ensured for Hillary Clinton as these States were democratic. 
There are several possible reasons such as methodological problems, non-response bias, undecided voters, shyness of Trump supporters, female candidacy, and close polling margins for misleading poll predictions. The next time around, researchers and pollsters need to consider these reasons to get accurate result. They may also need to try harder to account for human psychology to get accurate prediction in the presidential election. 
The writer is Graduate Assistant, Department of Communication Studies, Edinboro University of Pennsylvania, USA.
[email protected]