#GE2015: What Twitter’s API revealed about the 2015 UK election debates


Televised political debates are a relatively new addition to election campaigns in the United Kingdom. The first debate was not broadcast until the 2010 Election, and then only for a second time during last year’s General Election. With viewers unaccustomed to watching their candidates battle it out on the telly, unlike other jurisdictions where televised debates are a campaign staple, the United Kingdom presents an interesting opportunity to gauge the impact of debates on fresh audiences. 

Using Twitter engagement as a yardstick, a study conducted by researchers at the University of Reading used data analytics to gather insights about the electorates’ engagement during the 2015 General Election’s four debates.

As well as using the official Election and TV events' hashtags - #GE2015, #battlefornumber10, #leadersdebate, #challengersdebate, #bbcqt - the team collected tweets that contained terms related to politics in the United Kingdom, like labour, Greens, and Cameron. Since these terms could also be referencing unrelated topics, they enlisted the help of domain experts to ensure that unambiguous references would not inhibit the integrity of their results.

Following this process, the team collected over 28 million tweets associated with United Kingdom politics or the General Election. Tweets were retrieved starting from three months before the election, to three months after it, using a java-based application. In addition, the Twitter streaming API was used to gather content in real-time.

Simply looking at the volume of tweets reveals substantial social engagement during the debates:


Image: Di Fatta et al, University of Reading.

But what about audience sentiments?

The team coupled their data collection with political sentiment indexing to assess the public’s attitude towards what was being said throughout the debate. Using the Penn Treebank parser, which automatically tags words with their part-of-speech, the team extracted adjectives and manually scored them for positivity, negativity or neutrality.

Applying this sentiment analysis to the 600,000 tweets gathered during the Leader's debate allowed the researchers to derive insights into the audiences' mood towards each of the debate's four major topics: the deficit, NHS, immigration and future for young people.

"40 minutes into the debate, Ed Miliband outlines his plans on how to finance the NHS and following this statement, Labour reaches the peak of positive evaluation," explained Sylvia Jaworska, one of the project's researchers and Lecturer in Applied Linguistics.

"Conversely, UKIP should seriously re-think its NHS policy; stigmatising HIV patients is not going to win public support, though UKIP’s views on immigration seemed to do the trick."


Image: University of Reading General Election Blog.

Although the study had certain limitations – the Twitter streaming API caps traffic to 1% of global traffic, for instance – it provides a solid taster of how new audiences respond to televised election debates.

The researchers behind this project also adopted a data driven approach to look at other aspects of the 2015 election. Read more on their blog here or read the full research paper here.

Photo: David Dixon