The following is a guest post from political scientists Philip Habel and Sarah Birch, along with computing scientists Iadh Ounis, Craig Macdonald, Anjie Fang and Richard McCreadie, all from the University of Glasgow.
On Thursday, voters went to the polls for the 2015 British general election, which some have dubbed the Hashtag Election due to the widespread usage of social media by both politicians and citizens alike. Social media has been perceived as an important tool for reaching an audience – particularly in the aftermath of the Scottish Independence Referendum of September 2014, where Twitter and Facebook played a prominent role. Indeed the ability for users to voice their preferences and to connect with others with similar views likely played a role in magnifying enthusiasm and even birthing new political activism in the #IndyRef, particularly among younger voters with 16- and 17-year-olds enfranchised to vote.
Here at the University of Glasgow, a team of political and computing scientists with expertise in information retrieval have been gathering and analyzing tweets pertaining to the election, crawling the Twitter API for tweets containing keywords and features such as the commonly used election hashtags (e.g. #ge2015, #ge15) event hashtags (e.g. #leadersdebate), parties (e.g. Labour, UKIP), politicians (the names and Twitter handles of party leaders). For the period of April 3-30 – immediately following the first debate featuring all seven parties’ leaders – we have collected and analyzed over 19.6 million tweets contributed by over 1.6 million unique users. Here we present descriptive analyses of these tweets as part of a larger project using computational methods to tackle research questions surrounding the quality of deliberation over social media and the dynamics of influence among citizens and elites.
To begin, we examine the most widely used hashtags and most widely referenced Twitter handles in our dataset. We represent these data using a word cloud, where larger fonts represent greater usage, although placement in two-dimensional space is inconsequential. We color the parties, leaders, and related items by their traditional colors, with a U.S. readership to be reminded that here the Conservatives use blue and the Labour party, red.
Looking first at hashtag usage in figure at the top of this post, we see that beyond #ge2015, the most used hashtags belong to #UKIP and #SNP respectively, the UK Independence Party that has built its reputation around anti-E.U. politics and immigration reform, and Scottish Nationalist Party has risen to prominence particularly in the wake of the Independence Referendum, with an expectation of a massive net gain in seats in Westminster.
To offer some perspective on what the figure represents numerically, we have seen over 430,000 uses of #UKIP and 335,000 uses of #SNP. Of course not all of these have been positive. Indeed one can see from the figure that for every four reference to #VoteSNP (~266K), there was at least one reference to #snpout (~72K). In fact many negative uses of #WhyImVotingUKIP have attracted attention by the mass media, both in a past 2014 election and here again near Election Day.
Turning to Twitter handles, @UKLabour was the most referenced overall, with over 287,000 inclusions in Tweets – 120,000 more than the handle for the Conservative party, @Conservatives – this despite the fact that the Conservatives currently hold power. The higher total for Labour could in part be due to the demographics of Twitter users, or perhaps more activism among Labour, or generally more interest due to the possibility of Labour forming a coalition with SNP. And of course one should keep in mind that these frequencies are indicative of usage, with many tweets discussing “Labour” or “Conservatives” (or “Tories”) without using their explicit handle, which are excluded from this particular figure.
Among the party leaders, @NicolaSturgeon is referenced nearly twice as many times (~238K) as the next leader, @Ed_Miliband (~122K). But in fairness to the leader of the Labour party, Miliband has managed to generate his own unique following through hashtags such as #Milifandom and #EdStones that have appeared over the course of the election.
As one of our motivations for collecting the data surrounds concerns over the quality of deliberation and civility (and incivility), we examined which tweets were most widely disseminated. In the next figure, we present the accounts of those whose single tweets traveled furthest. We boxed those accounts in brown who were not members (explicitly) of the mass media, nor political parties or leaders.
As one can see by the number of brown boxes, the most retweeted messages do not always stem from elites (politicians or journalists) – although most of these users have mass followings. That being said, and perhaps as a lesson for social media novices and youngsters out there, as one can see from an example, one does not need a mass following to generate a viral Tweet. @DeKroonCat with a following of less than 1,700 users posted a flier from a UKIP candidate that was riddled with grammatical errors and subsequently corrected by an English teacher. This tweet generated more than 15,000 direct tweets and many more indirect ones, placing it among the top 30 most retweeted items related to the election. Indeed among the most retweeted messages we typically found humor/satire or controversy. And continuing a theme that was also relatively new to the 2015 general election – major media outlets picking up stories as they emerged on social media – several reported the tweet as its own unique news story.
We can also observe how the Twitterverse responds to real world events. Given the rise of the SNP and the presumed absence of a single party with a majority, one discussion that continued over the course of the campaign was the extent to which the Labour party and the SNP would work together, whether they would form a coalition.
In the figure below, we harvested Tweets that included the mention of one or more parties (here not their specific account names – merely mentions) and the term “coalition” from 3 April to 2 May. In this figure we see spikes in the data that correspond to real-world moments, particularly comments by either Sturgeon or Miliband related to the possibility of a coalition with their two respective parties. Also interesting is that we see that despite the Liberal Democrats being part of the extant coalition with the Conservative party, they are mentioned even less often than UKIP, who are unlikely to play any substantial role in a coalition due to a limited number of projected seats.
We are also interested in the ways in which Twitter can set the public or policy agenda, or where users follow the lead of elites. One of the ways we have focused on this question is to use descriptive information that users provide in their Twitter account profiles to categorize users into groups such as politicians, the business community, journalists, and others, and then to observe the topics of discussion in tweets among these communities, either by first identifying terms of interest, or second by using automated approaches to understand topics as they emerge organically. In our final figure – available online in a dynamic form for further exploration – we do not distinguish among such categories of users, but we do show interactions among pre-selected topics. Here we chose differing issues or policies that were prominent during the election (reflected in gray), as well as including references to party leaders and parties in their respective colors. Topics are represented on the wheel, with the size of the bands indicative of the number of tweets including the term. The size of the bands across topics are also indicative of the number of tweets, with the color selected of no real consequence (e.g. tweets including “Labour” and “Conservative” are here in red, although blue would have been equally suitable.)
Several interesting findings emerge. Tweets related to the parties and leaders dominate the conversation. Even important issues such as the economy or the future of the NHS (the national health-care system) are not nearly as prominent. Despite much rhetoric surrounding immigration and relations with Europe, the number of Tweets is relatively small compared with Labour, SNP, or Tories/Conservatives.