How Social Media Can Predict Poll Results

trump on social media

Recent elections have shown that conventional poll predictions are giving out erroneous forecasts, with actual results coming out as the opposite of what pollsters had indicated. On June 8, 2017, in the UK snap election, the Conservatives of Prime Minister Theresa May narrowly won but lost its majority, resulting in a hung Parliament. Opinion polls had wrongly predicted a strong lead for the Conservative Party.

Seven months earlier, on November 8, 2016, Republican nominee Donald Trump won the United States presidency. No one, not even the candidate himself, saw it coming. How could a xenophobic and misogynistic narcissist be chief executive of the United States, the country that all other countries in the world look up to? Almost all major media outlets, political pundits and forecasting models had predicted a Clinton landslide victory or, at the very least, a win by a narrow margin. Only a few, like FiveThirtyEight, were fairly accurate. Even Trump’s own partymates in the GOP didn’t expect him to win the Republican nomination, much less the presidency.

Pollsters and political analysts are always dumbfounded when an election outcome goes against popular predictions. In the good old days, ballot forecasts done with traditional methods were usually right. Winston Churchill’s loss to the oppositionist Labour Party was confounding, as was Al Gore’s defeat to George Bush were baffling exceptions.

But the effectiveness of conventional public surveys for poll forecasting has greatly diminished, especially among the young citizens who make up a sizable chunk of the voting population. Technological innovations, shifts in attitudes, and other social and political factors are impediments to the tried and tested data-gathering methods. People now use cellphones and are wary of answering calls with unknown caller IDs. Or they are simply too busy to bother with requests for surveys. They are also vigilant about protecting their privacy and will not share information about themselves. These have helped to widen the discrepancy between election forecasts and actual results.

How the internet has changed the political landscape

But information technology has provided election prognosticators with new ways of predicting poll outcomes. Opinion mining data firm Brandseye correctly guessed the Brexit referendum outcome and Trump’s victory in 2016. From the conversations of over four million authors in social media, specifically Twitter, the company, using crowdsourcing methods instead of machines, was able to understand the sentiments of the participants. As election day neared, pro-Trump and anti-Clinton tweets surged. In particular, the company’s people search tool located tweets coming from the 11 key states, which was the battleground for the election.

Social media giant Facebook as a forecasting model has proven to be more effective than conventional polling prediction. In the 2012 and 2014 US elections, a Facebook Model tested by Matthew MacWilliams, a Political Science PhD graduate at the University of Massachusetts, had forecasts that hewed closer to the actual results than the major forecasting models, including the New York Times and Washington Post. For the 2016 election, Trump dominated Facebook conversations and had garnered 12 million likes in the few days leading to election day. Clinton’s likes did not reach 8 million in the same period.

With 66 percent of Americans using Mark Zuckerberg’s creation for their source of news, Trump’s win becomes not so surprising anymore. The Democrats just didn’t see it coming.

Granted, social media is a powerful predictor of polls, it was also misused by Russian entities to spread fake news, costing Clinton the presidency. Facebook became the unwitting abettor of Moscow’s troll farms working to get Trump elected. Zuckerberg has set up protection tools on the social media’s platform to guard against disinformation. Facebook also has tips for users on how to spot fake information. Fact checking, which is much like a background check employers routinely conduct, involves scrutinizing the URL and verifying the news with other trusted sources or media.

Social media’s role in politics has not yet reached full maturity. Politicians, their advisers and connivers will find more creative ways to use this popular networking technology to their advantage. Advocates for truth and decency should be just as zealous in upholding ethical standards and remain vigilant to avoid being outsmarted.


Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.