Next week Scotland goes to the polls and for the very first time this week the BBC quoted a YouGov poll saying that the Yes campaign may just about be in the lead. Until now all the polls that I saw them reporting showed the “NO” campaign to be ahead. Everyone is saying that the vote will be close.
Majestic is a big data source which can extrapolate deep insights from the way in which pages and entities on the web connect to each other. We thought it would be useful to turn our insights onto the Scottish independence referendum.
From my analysis, it looks like the Yes campaign will beat the No campaign on polling day. As I write this I am personally very disappointed as I think Scotland would be stronger in the union. We predicted it correctly for the Mayor of London, and Obama vs Romney – so we do have a track of calling these things.
On the surface it Looks close…
Pundits may not have seen this chart before – and it is the first time it has been available for an election like this, but if you know what you are looking at, this chart tells me that the Politicians think Scotland may be “Better Together” (staying within the union) but society significantly disagrees.
These numbers are logarithmic numbers between 0 and 100. The headline metric is “Trust Flow” which is the circle on the left. It tells me that overall, the YesScotland campaign is slightly less influential (28/100) than the BetterTogether campaign )29/100 As the numbers are logarithmic, 29 is quite a bit more than 28… but what is hugely interesting is that we live in a democracy and for once society gets to vote on an ISSUE rather than a politician. We can now look at where that Trust Flow is emanating from, rather than just the headline influence. What we see when we look at the TOPICAL TRUST FLOW is a very different story. We see that all of the influence in the Better Together camp is being pushed by politics (28/100) but the Yes Scotland campaign also has a load more traction from Society. In particular, from People in society (23/100) compared to the more general society category from the no camp at 20/100.
It would make some sense intuitively to suggest that Westminster – with it’s 800 year old political infrastructure (if you start at the Magna Carta) – has more persuasive tools than the Yes camp, but the story becomes more murky when you see that society and people of influence seem to be keeping pace with the establishment. The establishment doesn’t get a vote… so I think the Yes camp is further ahead than the polls suggest.
My prediction is only tempered by the observation that the Citation Flow (the right hand circle) shows a larger value for the Yes camp. This suggests that the Yes camp have been pushing harder on smaller – lower quality – websites. If this number was immensely high, I would suggest that they had overcooked the picture (like Mitt Romney’s camp did in the US) but in this case I am prepared to say that this is a consequence of the Yes camp debating in more town halls and smaller groups and niches than the no camp trying to manage the campaign from Whitehall.
Is it already Game Over?
When we look at the number of referring domains going to each site over the last 90 days, we can definitely see more momentum for the independence camp. They are winning, it seems.
Breaking the vote down by sector
For the first time, Majestic can now also tell you what sectors of society are likely to vote in which direction. This is because Topical Trust Flow cuts through verticals. Here’s some interesting observations:
The Health Industry wants to keep the Union (12:0)
You won’t find many influential health sites suggesting a Yes! It seems to me that Scotland will lose the National Health Service and that’s bad news for people in the NHS and drug companies looking for a central distribution channel for their drugs. The Health industry overwhelmingly says “no”.
The Newspapers want devolution! (22:16)
They may wish to appear independent, but devolution is going to sell more papers than a United Kingdom. Imagine all the stories as the political status quo starts to unravel. Chaos ensuing as Tax systems, Welfare systems, monetary systems and everything else goes up for grabs.
Business wants a Union (16:15)
Its not totally cut and dried – but a union is generally easier for business. One VAT system, one market, a closely aligned legal system all makes for cheaper business within a United Kingdom. Scottish businesses, though, may still find benefits in branding themselves “different”… just as long as we all get along after the vote! In the shopping vertical the message is even more marked, but that could just as easily be a short term signal as people clamber for “Independence” T-shirts and merchandise. [Edit: For clarity, the stats are saying business category is linking to the Yes camp more than the No camp]
The Historians think this is decisive for Scotland (16:13)
The reference category is where history will be written for years to come. The Wikipedias and the minutes of record. Here it seems history is being pre-written in favour of the no campaign. [Edit for Clarity: the Reference category has been more inclined to link to the No camp (to keep the union).
The Last Word
I wish I could be sure… but I think Scotland will vote for Independence by much more than the polls suggest. I can’t say for sure because I had to work this out on a morning’s analysis. With more time I would start drilling down into the location and demographics of the users on the sites passing Trust Flow into the two campaigns. I am not paid to do that, but any political commentator ignoring Topical Trust Flow as a signal in any election is missing a trick now.
Of course, Stephen J. Dubner – co-author of “Freakonomics”, reminded me recently that me sticking my neck out has little consequence if I get it wrong… but I think this is different. This time we are using Big Data to make the predictions. Being able to derive deeper insights into what the world is thinking, just by analyzing how the world connects on a grand scale is potentially a huge advance in the world of predictive analytics. We will make mistakes, for sure, but to not try would be the real omission.