Alex Massie

Nate Silver on Scottish independence: Alex Salmond has “no chance”

Nate Silver on Scottish independence: Alex Salmond has "no chance"
Text settings

Nate Silver, in Edinburgh to punt his new book, appears to have annoyed some Scottish nationalists today. The "polling guru" (according to all newspapers everywhere) has told the Scotsman that he thinks Alex Salmond's merry bunch of nationalists have 'no chance' of prevailing in next September's independence referendum.

It is true that Scottish politics is not Mr Silver's area of special expertise. It is also true that I am not sure his views are necessarily all that important. They do not carry top-weight in this handicap. I am not sure they merit being treated as some sort of Oracular revelation.

Nevertheless the man can read a poll and since there's been no shortage of polls in recent months and since the numbers have, generally speaking, hardly shifted at all there is, correspondingly, no reason to suppose that the polls are wrong and that there are hordes of secret nationalist voters withholding their true preferences from the pollsters. (It is true that Panelbase polls offer more encouragement to the SNP; it is possible that Panelbase are measuring opinion more accurately than their peers.)

Instead too many nationalists cling to findings from bogus surveys of people on Facebook or Twitter or to meaningless questions such as 'Would you favour an Act of Union if it were proposed today?' (Only 18% said they would). These things, you see, show that the Yes side is winning. Apparently. I think.

The rest of us suspect that the balance of probability lies with the No campaign. It is possible, as some SNP types wish, that this will lead to a measure of complacency amongst Unionists. One cannot rule out human stupidity. But, as today's Scotsman editorial reminds us, there's no upside in under-estimating Alex Salmond.

The Yes campaign, for all its blundering on currency and pensions and borders and all the rest of it, retains some advantages. They have a story to tell, for instance, that is more easily understood than anything the No campaign is likely to peddle. So, yes, if I had to wager on the outcome I would argue that the result is liable to be rather closer than the present opinion polls suggest.

In any case, Nate Silver is one of those people simultaneously under and over-rated. His record of forecasting in Britain is not especially impressive. His model forecast that the Liberal Democrats would win 120 seats (they actually won 57).  Perhaps Silver was just a Clegg Maniac.

Well, everyone makes mistakes. No shame in that. Still, Silver's great strength is less the substance of his predictions or the robustness of his various models but, rather, the manner in which he has presented his data. Plenty of other people predicted the results of both the 2008 and 2012 presidential election. And Silver's model was really very simple: he looked at the opinion polls, averaged them, and used them to predict the actual results. And guess what? The polls were right!

However, Silver was - is - very good at presenting his information. He begins with a model based on past election results, demographic trends, socio-economic data and so on. This allows him to make long-range forecasts. However the closer to election day the more his models align themselves with the traditional opinion polls. This makes sense. After all, this information is more up to date - and more pertinent - than the results of previous elections which are, in any case, more a guide to what to expect than a robust prediction of what will happen.

Still, Silver really made his name during the 2008 Democratic primaries. That seemed a harder contest to predict than either of Obama's general election victories. Here Silver's attention to demographic detail paid off. Many pundits persisted in deeming the Obama vs Clinton contest 'too close to call'. Silver called it long before it was fashionable to do so. By looking at the demographic make-up of states and combining this with poll-analysis, it was possible to see that there were just more states Obama was likely to win than states in which Clinton was favoured. So Obama was going to win. This was good but it was the clarity with which Silver wrote - and presented, visually - his analysis that really separated him from the pack.

Of course journalists had an interest in deeming the Clinton/Obama race and, more recently, the Romney/Obama tussle too close to call. It makes everything more interesting! And, conveniently, it also ensures you are not too closely tied to any single prediction. If it is too close to call then there's no reasonable penalty for being reasonably wrong about the actual result.

The same might be said about the Scottish independence referendum. Declaring it too close to call is the risk-free option. It allows for a measure of hedging that lessens the risks of eventually being seen to be hopelessly, ruinously, wrong about everything. (Then again: punditry is a place where always being wrong can, perversely, be a good career move.)

So the value, if any, of Silver's contribution to the discussion lies precisely in the fact that he's an outsider unencumbered by too much knowledge of the particular details of Scottish politics. He has, really, no skin in this game. Even if he is wrong no-one is going to care about that, least of all his new paymasters at ABC and ESPN.

And there are times when it is useful to be reminded of the bigger, outside, picture. Silver may be wrong to say the nats are done for, but it is still a useful data point to be reminded that this is how our stramash looks to the outside world. It helps to clarify the picture.

Perhaps only momentarily for, of course, that picture may change. From the perspective of the Yes voter it needs to. So, as I say, Silver's contribution is useful even if it is subsequently contradicted by events (a possibility to which, it should be said, he is open). We are where we are and, right now, a little over a year from the referendum the Yes side are losing.