Scotland, Social Data, And McRussians
As voting for the Scottish independence referendum begins, many are left thinking how we got to this point so suddenly. Who would've thought that against all odds, the pro-independence camp would rapidly organize and gain so much support, such that polls would effectively show even chances for a win?
In his very visible critique of media coverage around the referendum, George Monbiot wrote that out of all major local, regional or national newspapers, only the Sunday Herald supported independence. Those Scots who will vote "yes" today have been effectively without representation in the media throughout their campaign. How did they manage to gather so much momentum?
We're witnessing, once again, a great reshuffling of power as online networked spaces impact every single function in our society. Attention is a scarce resource and everyone, including traditional power players, have to work to gain it. The better one understands the networked nature of media spaces, the better one can actively engage with potential constituents, supporters and fans. With social data, we can start to understand how people choose to focus their attention on certain topics, and how information spreads.
We pulled together a number of interesting data points from social networks over the past couple of weeks, showing networks and communities of users, local trends, and popular Facebook pages. We also pulled some of the funniest things we saw, because why not?
Networked Constituency
Thinking about our audience (in this case "constituents") in terms of groups — where one either belongs or doesn't — is a thing of the past. Rather, we need to have a networked view of our supporters, our fans, those engaging in conversations about us. When taking a networked approach, we tend to see certain trends, groupings and clusters that many times reinforce our intuition.
For this graph we looked at over 160k tweets that included either the #VoteYes or #VoteNo hashtags over the past week. Then identified who of these users follow each other and constructed a graph that represents these relationships. Tight regions of interconnectivity (clusters) represent users that follow each other much more than they do other users in the graph.
– There's a clear separation between the two sides, while each is further split into multiple communities.
– The purple region of the #VoteYes camp on the left is much more interconnected compared to any other region in the graph. The average degree of each user in that group is 27. This means that every user, on average, is connected (followed or is followed) to 27 other users from the group. A tightly knit group reflects some form of organized community, and in this case it is clear that these users such as @blissapp, @Scotsfox and @mikebell48 are optimizing for information to spread.
– In contrast, the #VoteNo groups are clearly more dispersed and not as organized for optimal information flow. That said, the Glasgow Rangers fans, as expected, are very much on the #No side, yet they have their own structure of connectivity.
– There's a separate group of users who support Catalan independence on the top left, who are also very much in line with the #VoteYes camp.
– Guess where many of the undecideds are situated?
What we don't see on this graph is a group of Russian users who have been fully supporting Scotland's separation from the UK, perhaps as a way to justify the Crimea annexation, or as an argument to liberate eastern Ukraine. Regardless, they're very much into the #YesScotland hashtag (ex1, ex2, ex3), and some of them, like Konstantin Rykov, even changed their name and profile image to show support. (Never trust a Russian politician sporting "Mc" in his Twitter handle!)
If we dive into the different regions of this graph we can start to identify trends in the type of content that tends to circulate within each side. Using an internal betaworks tool, we can see a visualization of this content — the larger the image, the more central it is within the observed community at that point in time.
#VoteNo
When we look at the type of content shared by this group of users, some common themes, such as military pride, the economy, and some David Cameron bashing, arise. When not posting about the referendum, there's a good bit of provocative content around topics such as Islam and ISIS.
#VoteYes
Here we tend to see images of happy, smiley white Scots holding signs and billboards at rallies, messages from famous artists (Hi Bjork), assurances about the Scottish economy and the continuation of the NHS government funding.
The content flowing through these networks constantly changes. As I'm writing this post, I'm seeing clear targeting of content toward undecided voters from both sides. At this moment, I'm also seeing some odd poll figures, such as this one, claiming there's a 26-point lead for the "Yes" camp based on a Sky News poll. Hard to know what to trust.
Distribution Of Attention
The following plot highlights the top performing content about the Scottish Referendum published over the past week. On the y-axis, we're plotting an internal score we use that's a proxy for the number of unique users sharing the link, while on the x-axis we're taking a weighted score of the number of social timelines the link appeared in. The third dimension, size, represents the potential audience, effectively aggregate followers, likes and so on.
Note the stark difference between the one Guardian story (blue dot at the top right) compared to the rest of the distribution of scores. This is the Monbiot piece, which was heavily shared and consumed around the web. The article did exceptionally well at capturing people's attention since being published on September 16th. This BBC summary from the last day before the referendum received a sizable audience, and this Economist piece saw a good bit of attention. The most surprising piece identified in the top list of links came from Tatler, on "Old Scotland" and how Scottish aristocrats are fiercely opposed to independence.
A Tale Of Two Facebook Pages
There have been various public Facebook pages used throughout the campaign on both ends, yet two seem to stand out.
On the one side, there's the "Vote NO to Scottish independence and protect the union" community page, sporting 69k likes and the "Naw" symbol. And on the other, the "Scottish Independence – Vote YES" community page, has 64k likes. While the raw number of likes doesn't tell us much, we can start to identify other data that will help us measure engagement. For example, we can calculate the average number of shares per post for each page. We see a surprising difference: On average, an item on the YES page sees 1423.5 shares, while an item on the NO page gets just 229.9. This is a major difference that highlights just how much more engaged the Scottish independence proponents are on the platform.
Additionally, when we look at the page statistics, it is clear that the YES page is seeing much more growth and momentum compared to the NO page.
Trending Topics
Another way for us to get a sense for where people are focusing their attention is to look at Trending topics. Twitter gives us the ability to identify locations in the world where certain terms, phrases or topics are trending. By creating a dispersion plot mapping the word 'Scotland' trending in different locations over the past couple of weeks, we could see how people in different parts of the world start to pay attention to the topic.
Unsurprisingly, "Scotland" has been trending around the UK for most of the past 6 weeks. As the referendum comes closer, the trend spreads to other locations, starting with previous British colonies such as Australia, New Zealand and Malaysia. Two surprising trends that we see in this data are denoted by the yellow ovals. For some reason "Scotland" started to trend across the UAE on Sept 5th, and in Hyderabad on Sept 11th, and has been consistently trending through today. This means that there are enough users in that location tweeting the word "Scotland," making it trend locally.
Perhaps the global popularity of "Scotland" has to do with close ties to immigrant populations in Scotland, wealthy expats owning estates, conversations comparing UAE and North Sea oil resources or Hyderabadi residents reminiscing on their 1948 attempt for independence from India. We can see that both UAE and Hyderabad are locations where users are attentive to the Scottish referendum, much more than users in nearby geographic locations.
And Finally, Some LOLs
We can't have a social media post about an important topic without showing off some of the weird and funny content passing through the network, now can we?
Some of the most liked Instagram posts include:
And from Twitter:
And then there's this:
If you have any thoughts, ideas or questions about our data, feel free to ping me on Twitter – @gilgul