Nate Silver on Panorama
I’m not a regular watcher of Panorama. It is a TV programme that has dumbed down considerably in my lifetime and I have probably only seen two programmes in the past decade. One was a couple of weeks ago on male suicide and the other was this Tuesday’s episode about Nate Silver attempting to predict the outcome of the forthcoming General Election here in the UK.
Nate Silver’s book The Signal and the Noise, on the uses of statistical methods for prediction, is one of my favourite books of recent years and is well worth reading if you haven’t read it. This programme is not so good.
Silver was successful in predicting, against a backdrop of pundits who said it was too close to call, that Obama would win the last US Presidential election. Hence, it is natural to see if he can do the same in a UK General Election. My recollection is that he discusses this in his book and points how it is much much harder it is. Given recent developments in the UK it is now even harder…
Anyhow, the programme itself is rather lightweight. The presenter Richard Bacon takes Silver round the country explaining the UK system, Silver goes back to US, and makes a prediction. There is no explanation of the statistical methods involved or how they relate to the election. How Silver arrives at the conclusion is overlooked. I’m not expecting an explanation of Bayesian inference or some such on prime time TV but the probabilities and ranges in his prediction have been removed and replaced with absolute numbers. An averagely intelligent viewer can grasp the concept of range surely.
Hence, no major insight is gained. The prediction is not that different to what many would guess. My prediction was that, due to soft UKIP support, the Conservatives would be largest party but no majority, and that Labour and SNP would have a combined majority. And, apart from the numbers in the centre of the ranges, that’s all that the programme gives us. That’s a bit disappointing.
The programme is available for 11 months on iPlayer but I guess that it is not available in all territories. Watch the programme here.
Silver’s report, made in conjunction with three British academics, is available on Silver’s site The Five Thirty Eight. The three academics, Chris Hanretty, Benjamin Lauderdale, and Nick Vivyan, maintain the site electionforecast.co.uk.
On the subject of election results, why do newspapers never give us the number of Don’t Knows when they publish a poll result before the election? I would say that group is important. Also, the number of Don’t Knows can be as large as 40% at the start of a campaign. It would be interesting to see how this changes during an election and to know why they are Don’t Knows. Just had to get that grumble off my chest…
UPDATE: Well, just about everyone got it wrong. The Conservatives achieved a majority government! David Spiegelhalter has quite a bit to say about the polling problems on his Understanding Uncertainty website.
Get the newsletter! Fancy a newsletter keeping you up-to-date with maths news, articles, videos and events you might otherwise miss? Then sign up below. (No spam and I will never share your details with anyone else.)