Analysing Elections through Social Media Data: Why we Should Care


Over the past 60 years, national elections have become a topic of keen interest for political scientists all over the world. This isn’t surprising, of course: voting, after all, represents the central act of the democratic process. And central to understanding elections is understanding public opinion. Why do voters make the choices that they do? What influences them, and what causes them to change their mind?

Until recently, there were two main ways to understand public preferences and what shifts them. Researchers used surveys to understand more about the effect of socio-demographic characteristics (age, sex, education, job, religion) and political preferences (left-right political positioning, issue-positioning) on voting behaviour. On the other hand, political communication scholars have focused on the “supply side” – primarily, investigating the framing and influence of political parties’ communication strategies.

However, over more recent years, as the rise of the digital world has profoundly changed where and how elections are fought, so too has social media become an increasingly prominent field through which to analyse them. This represents a new and exciting opportunity for researchers, simply because we have never before had access to so much data.

What could all this data teach us about how politics works? First, on a tactical level, it teaches us about how political parties and politicians fight elections. Politicians use social media to boost their profile, to share their ideas, to increase their popularity – and each of them goes about this in a different way, so that – for the first time, we can really build a sense of their individual approaches and how effective they are.

Secondly, social media allow us to monitor citizens’ reactions to what politicians say in real-time: just one hour of a televised electoral debate generates millions of tweets. But it’s not all just about big numbers. The beauty of social media is that it is a two-way street: people can respond to politicians’ messages. For this reason, the universe of interactions within social media represents is an entirely new object of study, one in which the relationship between citizens and elites can be directly investigated. For example, are leaders actually “listening” – and answering – to their online public, or do social media represent only another opportunity for them to scream their own opinion? How can we rethink agenda-setting in an online world?

But perhaps the most exciting opportunity social media creates doesn’t relate to politicians themselves at all. It has to do with how it enables us to study everyone else. It is in this area that the impact of social media becomes revolutionary. Before now, we have never had the chance to listen to citizens’ voices outside of traditional polling mechanisms. We’ve been able to ask the public questions, of course – but listening to social media allows you listen to the things the public freely and spontaneously decide to talk about. With all its limits, this is the closest thing we have ever been able to access, to understand what people think, independently from the assumptions of the researcher. More than this, we have the unprecedented opportunity not only to grasp what people attitudes are, but also to monitoring how they change and respond over both long and short periods of time.

Am I arguing that from now on we need only to focus our political science research on social media? Not at all. What am I saying is that there are incredible, new opportunities out there, and that we have an intellectual duty to grasp them, if we want our understanding of politics and society in present times to be as comprehensive as possible. Surveys and social media data, as much as traditional and new methods, qualitative and quantitative approaches, theoretical and empirical understandings, need to go hand by hand – and to not be pitched against one another in the fear of loosing ground – if we want public opinion research to truly progress.

Yes, there’s a long way to go. We will need new methods, new standards, new practices. Demos is working on developing some of the tools for this new frontier through our ongoing academic partnership with Ipsos MORI and others. It is clear that researchers will need to see a much wider sub-set of the population embracing social media, in order to form a more representative online sample. We will also need to see many more comparative studies between online and offline research, results, publics. In ten years’ time, current social media research, I suspect, will look hopelessly crude. But this is what happens whenever a major disruption occurs. Now is not the time to be afraid of mistakes being made: these are times for curiosity and enthusiasm.