The BBC Trending team is all over the relationship between social media and Scotland’s independence referendum, which is being voted on today.
They point out that Facebook interactions around the Yes campaign are higher than the No Thanks interactions — but does that tell us anything useful about the actual vote?
In today’s data-saturated climate, I find it doubtful that Facebook will be able to predict the vote with any semblance of accuracy. There are too many factors.
- Facebook’s demographic skews younger — over 70 percent of Facebook users are under the age of 34 — and it’s known that the youth vote (including 16 and 17 year olds, who can now vote) support independence.
- Historically, Facebook’s ability to identify trends accurately has been imprecise and sometimes downright wrong.
- Facebook has skin in the game. They published the data on Tuesday to promote a new “I’m a Voter” button on the website.
- Facebook tracked the data by identifying a few phrases they believed to represent each side. However, it is possible that a significant percentage of people discussing the issue declined to use those specific phrases in their posts.
- Most importantly, Facebook chose “interactions” as the metric to track, and they falsely equated an impression to support for whatever cause mentioned. This is seriously flawed. A post proclaiming support for the Yes campaign, for example, could have gotten 100 comments filled with rationale supporting the No vote — yet Facebook would be counting 101 interactions for the “Yes” campaign.
Although Facebook itself doesn’t claim that this data accurately reflects the voting public, it is hinting that the “trend” could reflect the larger social conversation.
If you want to follow the Scottish referendum vote, @BBCtrending will be discussing social media reactions, patterns, and stories throughout the evening and night.
Photo by Marcus Quigmire, uploaded by Princess Mérida (cc 2.0 via Wikimedia Commons)