In the end, it was hardly a case of “as Twitter goes, so goes the nation”. If the election had been decided by the number of followers each candidate had before the election, Ed Miliband would have been elected Prime Minister, albeit needing the help of the Liberal Democrats and, um, the Pirate Party. The FT Data blog (£) has some fascinating charts showing how the conversation on Twitter was dominated by supporters of Labour and the SNP. It is apparent how Twitter, and social media more generally, was not at all representative of the voting public at large.
Does this mean that social media doesn’t matter, and that politicians and aspiring politicians should delete their Twitter accounts, deactivate their Facebook profiles, and adopt an old skool approach to elections and campaigning? Not necessarily; although the signal is somewhat difficult to detect through the noise, there does appear to be a trend that a large social media following can have a positive effect on polling day, and can play a crucial role in tight contests.
The charts below show the average movement, up or down, from the 2010 General Election to the 2015 General Election, divided into four bands of Twitter followers. The lightest band shows candidates with no Twitter account at all, the next band shows candidates with between 1 and 1,000 followers, the next band those between 1,000 and 10,000 followers, and the final band shows candidates with 10,000 followers or more.
Firstly, this chart shows what happened for non-incumbent parties, i.e. where the party (not necessarily the candidate) didn’t win in 2010. By-elections between 2010 and 2015 have not been taken into account, so there are no UKIP incumbents.
For the “outsider” parties, UKIP and the Greens, it is striking how a larger following on Twitter resulted in a stronger performance in the election. Admittedly, there is a form of circular reasoning involved here. Non-incumbents tend to do better if they have name recognition, and the number of Twitter followers someone can be viewed as a proxy for how well somebody is known, and it therefore stands to reason that they would perform better at the polls.
For example, there were two Green Party candidates with 10,000 followers or more, and their vote increased by an average of 6,083, compared to a national average increase in the Green vote of 1,575. However, these candidates were their incumbent MP Caroline Lucas, and their party leader Natalie Bennett, both of whom will benefit from higher name recognition from media appearances. Despite this, there is a clear trend showing how a greater social media footprint led to a better performance in the election.
The trend was also conspicuous for UKIP candidates. On average, the UKIP vote increased by 4,752 in each constituency in the UK where they stood a candidate (their vote only decreased in one constituency, Orkney and Shetland). The 306 UKIP candidates with no Twitter account increased the UKIP vote by an average of 4,447, whilst the eight candidates with more than 10,000 followers increased their party haul by an average of 8,577. Whilst these candidates included leading lights such as Nigel Farage, Mark Reckless and Douglas Carswell, even the candidates in the next tier, those with between 1,000 and 10,000 followers, performed better on average than their party candidates who did not use Twitter at all.
It is possibly intriguing that the only cohort of Conservative candidates to increase their average vote, in seats where they did not win in 2010, is in seats where the candidates had more than 1,000 followers on Twitter. The Tories won three of the five races where they lost in 2010 and their candidate had more than 10,000 Twitter followers, including their long odds victory in Plymouth Moor View.
Whilst there is little evidence of social media helping the Lib Dems in seats where they were the challenger, the story is very different when you look at incumbents.
For Liberal Democrats, an active social media presence did appear to play a role in mitigating the disastrous national swing against them. In England and Wales, there were nine seats where the incumbent Lib Dem MP had 10,000 or more Twitter followers. They won five of these, and in two more they lost the seat, but their loss in vote terms was fewer than 2,000. In their remaining 37 seats in England and Wales, they lost all but two of the seats, and their vote declined by at least 3,225 (they lost 8,443 votes on average in these seats). Incumbent Labour MPs with more than 10,000 followers also managed to beat the national trend, although the impact was far less dramatic.
In case you were wondering, there was literally no pattern whatsoever between Twitter follower data and electoral performance for the SNP. They essentially went supernova everywhere in Scotland.
So, what are the lessons to be learnt from this election in terms of social media? Firstly, it is no magic bullet. Amassing hundreds of thousands of Twitter followers is by no means going to propel a candidate to victory in seemingly hopeless cases. George Aylett in South West Wiltshire may have amassed nearly a quarter of a million followers, but at the election only received a 2% higher share than the Labour candidate did in 2010.
On the other hand, the data does suggest that, on average, having a widely followed social media presence can help a candidate beat their party’s performance nationally. For incumbents facing a difficult climate in terms of their party performance nationwide, engaging with your constituents on social media can differentiate their personal brand from that of their party, and it fulfils much the same role that assiduous attendance at church or charity events may have done in days gone by.
It may not have been Twitter wot won it, but it probably didn’t hurt.
A qualified accountant and data analyst, interested in politics, economics and data. Twitter: @peterdonaghy