Social media a good indicator of the Brexit outcome

In the aftermath of the dramatic decision by the British people to leave the European Union, many are seeking to understand just how the polls and pundits could have got it so wrong.

Sentiment leading up to the vote suggested that the Remain camp would win, given the conservative nature of the British people and the potential financial fall-out. However, we now know that despite well-reasoned arguments and general media opinion, this was not to be.

There was, however, one indicator that did point to Brexit – social media. While social media analysis has been criticised in the past as being unrepresentative of the general population, you only have to look at the rapid rise of both Bernie Sanders and Donald Trump to realise that the voice and influence of social media should not be underestimated.

According to the CEO of media analytics provider BrandsEye, JP Kloppers, analysis shows that social media has come of age as an overall representation of the population. “It’s no longer a refuge for the angrily opinionated. It seems to talk for the ‘silent majority’ as well. In addition, there’s no larger source of public opinion. With accurate analysis, social media performs better than polls in understanding the true sentiment in a population, which is not surprising given the much larger samples involved.”

Social media is becoming such an accurate indicator that in the days preceding the Brexit vote last week, BrandsEye – by using its human-powered sentiment analysis on a large sample of online referendum commentary – revealed that most Brits were in favour of leaving the EU, reports Kloppers. He explains that in the days running up to the vote, the company analysed a sample of 10,000 mentions representing half a million opinions across a range of social media platforms, and found that 56.9% of the opinions expressed were in favour of Brexit, contrary to the results of most of the polls conducted at the time.

While the traditional approach to social media analysis has been algorithmically driven, machines fail to understand sarcasm and nuance, particularly in the social media context, Kloppers says. BrandsEye uses a crowd-sourcing approach to sentiment analysis. Every online comment regarding the referendum was independently analysed by several trained contributors to create a 95% confidence level with a 2.5% margin of error, an unheard-of degree of precision in media analysis.

Kloppers maintains that in the run up to the US presidential election, social media analysis will become more important than ever. “While one clearly cannot predict the outcome based solely on the polls and the pundits, if we can accurately understand the general opinion across the population, we’ll be less likely to be surprised again as we were with Brexit,” he concludes.

The big take-out: Social media is proving to be an accurate indicator of public sentiment. Taking a crowdsourced approach, it’s possible to predict how people will react based on their comments on social media.