Comparing Swedish Elections on Twitter – first impressions

As part of my collaboration with Hallvard Moe, we are now looking at data on Twitter use from the recently held 2014 Swedish elections. We are thus in a position to compare various aspects of Twitter use from two elections – 2010 and 2014. The table below features some of the overarching characteristics of the major hashtags employed for each election – #val2010 and #val2014.

Type of tweet #val2010

N (%)


N (%)



Undirected 60 088 (60.2) 89 747 (36.5) + 29 659
Retweet 32 780 (32.8) 146 487 (59.6) + 113 707
@replies 6 964 (7) 9 411 (3.8) – 2 447
TOTAL 99 832 (100) 245 645 (100) + 145 813

Looking first at the row providing the “TOTAL” amounts of tweets for both elections, we can see that the traffic more than doubled in 2014 when compared to 2010 – from 99 832 tweets during the month leading up to the former election to 245 645 tweets during the same time period four years later. This growth was perhaps to be expected, given the overall increase of media coverage devoted to Twitter, as well as the more modest increase – but still an increase – in actual Twitter use. A change that was less expected was the shift regarding the most common form of tweets sent. In 2010, just over sixty per cent of tweets were undirected – tweets sent without any intended recipient.  In 2014, retweets – redistributed messages originally sent by some other user – make up for about the same percentage level, effectively becoming the most common type of tweet sent. For @replies, signalling discussions on the platform at hand, we see an increase in numbers of such messages sent from the 2010 to the 2014 period. Looking at percentages, however, these numbers indicate a distribution diminished by almost half, decreasing from seven per cent in 2010 to just under four per cent in 2014.

So what does this all mean? One initial, tentative interpretation goes as follows. The dominance of undirected messages during the 2010 election period would seem to indicate a ‘megaphone’-like application of the service at hand. In other words, Twitter was here used primarily in order to ‘get the message out’ rather than for interaction or dialogue. The prevalence of retweets in the 2014 sample suggests that such practices might have changed. Perhaps due to the increased use levels of social media by political actors – a tendency which is sometimes discussed in relation to the overarching idea of political professionalization processess – the results presented above could be an indicator that such actors (and those similar to them) are successful in assessing the viral qualities of the platform. Plenty of retweets means plenty of spread for your message. Indeed, the patterns discussed here do not reveal which users enjoyed the most retweets. Initial data analysis regarding is coming up in a couple of days.

Leave a Reply

Your email address will not be published. Required fields are marked *