The Facebook Parliament

Social media has become increasingly important in politics. Posting something badly phrased can cause a shitstorm and have major implications as far as stepping down as a party member. But at the same time, social media can also give a party completly new possibilities to campaign and stay in touch with voters. For this project I wanted to look at how parties in the UK differ in terms of what kind of reactions they cause on Facebook.

More specifically: What would happen, if we did not give parties seats based on the constituencies they win but based on the number of comments they receive per post? Or the number of reactions such as likes, sad, happy, angry and love emojis? Well, the House of Commons would be a pretty different parliament.

For all parties that are currently in the British parliament, I collected on the 11th of April 2017 their last 201 Facebook posts and the last 5000 user comments on their page. This makes a total sum of 2,412 posts and 60,000 comments. Then I analysed how users interacted and visualised how a parliament based on Facebook engagement would look like. For smaller parties like the Democratic Unionist Party, Plaid Cymru, Social Democratic and Labour Party and the Ulster Unionist Party the shift would be marginal. This is why they are summed up under "Others". But the changes for all other parties would be definitely bigger and can be explored below.

If the "Facebook parliament" would be based on the number of comments, the Conservatives would probably still be in power - but likely in a coalition together with UKIP. UKIP gets 21 percent of all comments that were left under the 2.412 posts that parties had made. So, what kind of posts can make people comment so much? Here is an example:

If you look at reactions in general, the race is much tighter with all parties getting roughly the same amount of likes or other reactions per post. However, the types of reactions can be very different. In fact, if you look at who gets the most angry reactions per post, the picture changes. The conservatives of Theresa May do not ignite a lot of rage. But Labour, Greens and UKIP do, with especially the latter two exceeding by far their actual importance in the House of Commons. But is this anger against the party or against a topic that both parties want their users to rage against?

These two examples from the most "angry reactions" causing posts suggest that it is indeed the well-calculated igniting of anger rather than the reaction to a badly phrased statement. This makes sense as people tend to like a party's Facebook page, when they can relate to the party. In contrast, it is rather rare that someone sees a post of a party that he does not like. Hence, Facebook pages and their posts act as a kind of "preach to the choir"-scheme. Causing angry reactions is a well-calculated action rather than an accident.

But Facebook reactions as a measure for rants can be problematic. For one, a lot of people are still reluctant to use the different reactions. Even in the case of the three parties that caused the most angry reactions on Facebook, the angry reaction only makes up about a tenth of all the reactions or even less. (Green party: 11%, Labour: 9%, UKIP: 5%) Also there is the chance that people rather express their anger by commenting.

Hence, I did an analysis of the comments on parties' Facebook pages and analysed for the comments the ratio of spelling mistakes and exclamation marks per words. Furthermore, I looked at how these ratios correlate with the position of a party in the political spectrum. To have a somewhat reliable measure for that, I used a study from the London School of Economics that asked party members of Greens, Labour, SNP, Lib Dems, UKIP, and Conservatives where they would see their party in the left-right spectrum.

Interestingly enough, after matching my Facebook data with these political positions, statistical analysis shows that each party has significantly less spelling mistakes than the average of the parties on its right of the political spectrum. So commentators on Facebook pages of the right-wing parties tend to make more spelling mistakes which could be a hint for anger.

However, this correlation was not present for exclamation marks. I have visualised this relationship in the graph below. The dot in the middle of the box shows the average ratio, the box around it is the 95% confidence interval.