Can Social Media Predict Political Elections?

It’s that time again: the biannual meeting of thoughts, words, and (usually) “he said/he said” performance reviews leading to that one important day. Yes, it is the season where people get to exercise their civil rights. The season, where most people are simply looking forward to one important thing: the day after. I speak, of course, of the election season – specifically, the 2014 midterm elections. We’re just a few days away from the end of a long political process that has been full of negative ads, false promises, and general confusion that extends beyond the clearly defined lines that separate elephants from donkeys.

While this process is an evil necessity, it is one that seems to come earlier and earlier every two years. It is necessary because of the way our government works and the issues that exist within the government that might be fixed with new blood. It is evil because of the experience everyone goes through during the election process. This experience not only includes the negative ad campaign and the false promises, but also the vast amounts of political commentary that gets thrown up everywhere. This commentary, from both sides of the aisle, can be found on any 24/7 hour news station. From CNN to Fox News, political pundits have to fill every space and crevice of the day with their own political opinions, assumptions, and flat out questionable thought processes.

It is this commentary that doesn’t know the appropriate beginning of the election process. For example, the 2016 political gossip regarding who will be running for the presidential elections started well before the main apex of these 2014 midterm elections. The gossip surrounding Hillary Clinton, in particular, started almost immediately after the 2012 presidential elections.

In essence, there’s a lot of unnecessary pundit pillow talk that happens way too early for anyone’s comfort.

What if, however, we had a way of eliminating the need for this pundit talk? What if there was a method for accurately predicting political elections before they happen? Sure, we have political surveys that try to measure the thoughts and opinions of people a few months, weeks, or days before election day, but they only provide a measurement of the political pulse at a given time. Instead, what if we could turn to one of the largest databases of human commentary that exists today? I’m speaking, of course, of social media. Is there a way that the content people express through social media could be harnessed to scientifically and accurately predict the outcome of political elections? Well, science suggests that this may actually be possible.

Predicting Electoral Outcomes

Social media tends to be a mixed blessing. On one hand, it has the ability to express the opinions of countless individuals. On the other hand, some parts of social media tend to be a cesspool of questionable human expression. Sometimes, you never know the rhyme and reason for particular comments, especially during the election season.

Despite this one side of social media, it is important to think about the potential capabilities this medium of communication offers. We live in a world driven by “big data”. The Internet has given us the ability to track a lot of information about people. Basic demographics, preferences, and opinions are just the start of the types of information that can be mined, stored, and analyze from the Internet. Social media, in particular, is unique because people use these websites as a means of expressing their ideas. As a result, many political experts and scientists view social media as the logical choice for analyzing political opinions that are expressed during an election season.

Right now, one of the primary ways of measuring these opinions is through survey polling. Research agencies, media outlets, and individual political campaigns use various methods of asking questions and measuring responses to statistically get an idea of politically related viewpoints. Many scientists view social media as a giant database that already contains a vast amount of this information. Think about it, an online database that invites people to express their ideas about any given topic and that isn’t constrained by particular time requirements. In other words, social media is a continuous and never ending political poll. Unlike other traditional polls, however, this one contains the viewpoints of millions of individuals.

The potential social media offers is obvious, but it still remains to be seen how it will live up to this potential. Using the 2014 midterm elections, we will have an other case that will indicate how effective social media can be in predicting electoral outcomes. How does this work exactly?

Using Social Media To Predict The Future

The idea that social media can be used to predict electoral outcomes rests on two important assumptions. First, it is assumed that a vast majority of politically conscious or involved individuals take to social media to express their viewpoints. The sample size we collect from social media needs to represent a holistic slice of the larger voting population. If the opinions people express through social media aren’t from a variety of individuals, any prediction might then be skewed.

The second important assumption is that political viewpoints can be accurately extracted using digital means. In other words, as I have discussed in a previous article, automatic sentiment analysis is still a fledgling idea. The accuracy of this method – using a computer to get positive and negative opinions about the topic automatically – is still being tested.

Given these assumptions, and the fact that many researchers are still testing the accuracy of these assumptions, it’s no surprise that this idea has gathered differing viewpoints. On one side, many researchers have found many compelling correlations between political sentiments expressed via social media and electoral outcomes. For example, in 2012, Politico analyzed the connection between candidates engaging voters through Facebook and their ultimate electoral fate. The website found that the more candidates engaged with voters on Facebook, the more of an increase in voter support they had on election day. Research in Italy also found a strong correlation between political views expressed through Twitter and the electoral outcomes of three separate European electoral seasons.

The accuracy of this method is still being tested. For the 2014 midterm elections, researchers are still pouring through vast amount of social sediments that have been expressed through social media this election season. Facebook, in particular, is being used to predict the results of four separate Senate races in Kentucky, Alaska, North Carolina, and Michigan. Since this election season has yet to wrap up, the accuracy of these predictions remain to be seen.

Even with these promising findings, there are many individuals that doubt the ability for computers and people to make predictions from social media. These doubters often express a number of concerns. For one thing, social media isn’t necessarily an accurate representation of the big picture since individuals using social media often fall into a few specific demographics. Without a greater representation of the voting population, it is difficult to ensure the accuracy of specific predictions.

Another issue doubters express about this method is the lack of complexity. Traditional polling methods take an in-depth look at the political views and leanings. Not only do these traditional survey methods get a bigger slice of the big picture, they also go into greater depth with the views being expressed.

Finally, as I’ve discussed previously, automated sentiment analysis is still tricky business. The technology – particularly the algorithms that gather the sentiments from social media – is still being developed. The accuracy of the information that is gained from digital content may still be questionable. Researchers are still continuing to develop new methods of testing the accuracy of this information. As a result, many doubters believe that it is impossible to get the true complexity individuals express in their views. More so, many doubt that social media itself can be used to communicate this complexity.

As Alexander Furnas described in 2012: social media provides a public opinion snapshot – a small window into a greater group of people and ideas.

The Future

Can social media predict future election results? Well, the jury is still out. We need a few more elections to figure out whether this information can truly make accurate electoral predictions. However, the implications of this technology are great. For politicians, researchers, analyzers, and the general public, the ability to predict the future from information on social media could open up many doors.

In the political arena, the ability to make such predictions would be helpful to candidates as well as the general public. For political campaign organizers, this information can used in the same way as traditional survey information. This information provides feedback on the messages and strategies politicians use when running for office. This benefit could give politicians and campaign organizers the ability to adjust their overall campaign strategy on-the-fly.

Even outside of the political arena, this technology has a lot of promise. Any number of subjects could be analyzed, which could benefit companies, nonprofit organizations, and anyone who needs the ability to rapidly gather, analyze, and use information to predict future outcomes.

So while the verdict is still out, one thing is for sure: social media is a source that holds a wealth of important information. This is a fact we simply can’t ignore as digital content makers. If we learn how to use this information, and create content from this information, our efforts will be more effective overall.

Can Social Media Predict Political Elections?