Let’s talk about one of my least favourite innovations in recent politics: the Twitter poll. I understand the thinking behind Twitter introducing it – ‘interact with your friends and get their opinions’ – but like all seemingly benign social media inventions, its taken on a new and more twisted life of its own complete with the annoying accompanying phrase ‘please RT for a larger sample’. While as a social scientist I’m glad that people have understood that sample size is an important aspect of polling, it’s only been taken on in the way of a cargo cult, where the more click offerings one makes to the gods of polling, the more accurate the poll will magically become.
The problem is that it doesn’t work like that – your sample size can be as large as you like, but if you’re not sampling the right thing, you’ll not get the right result. The classic example of this is the 1936 US Presidential election, where the Literary Digest predicted Roosevelt would lose based on asking millions of its readers to respond to a poll. Their results – from a sample of over 2 million people! said there’d be a clear victory for Alf Landon over Franklin Roosevelt by 57% to 43%. The actual voters went decisively the other way, giving Roosevelt a 60%-37% victory. The Literary Digest sampled lots of people, but it didn’t cover a wide enough range of the population to get a genuine picture of how people would vote. Actual opinion polling is more concerned with getting a representative sample of the population rather than the size of it because if you properly strike the balance between a random sample and a representative one you’ll get a broadly accurate response. There’s still a chance they’ll get it wrong – that’s the problem with taking any sample rather than the full amount – but they’ll usually be in the right range. Any poll with a self-selecting sample is only going to be right by accident, not by design and trumpeting that your Twitter poll shows the result in a constituency or that the Bus Pass Elvis Party are about to sweep the country isn’t far away from demanding to know just why Wikipedia’s page on Alf Landon doesn’t mention his Presidency.
and no, I’m not going to discuss the more accurate polling right now because the numbers are just far too depressing and we spent far too much time in 2015 obsessing over variations that were firmly in the margin of error than was ever healthy.
In more important news, the retreat into Potemkin campaigning continues with Theresa May now apparently only taking pre-approved questions from journalists while Buzzfeed are now barred from Corbyn events for the the crime of having reported something he said to them. However, we will get the spectacle in a couple of weeks time of seeing Paul Nuttalls of the Ukips being interviewed in primetime by Andrew Neil. But not Caroline Lucas, either because the party with an MP that won more than one council seat last week isn’t important, or because the death stare she’d give Andrew Neil after half an hour of his patronising nonsense pretending to be questions might harm the audience.
And finally, it’s time for Election Leaflet Of The Day, and we have out first independent General Election candidate to be featured. Ajmal Masroor is standing in Bethnal Green and Bow, and his leaflet makes much out of the fact that he came second to Labour there in the 2010 election while standing as a Liberal Democrat candidate. He also, in a rare event for someone standing who’s not previously been elected to Parliament, has his own Wikipedia entry which indicates he’s quite well known and unlike many other candidates we’ll likely see over the next few weeks, understands enough to get his leaflets properly designed and written. That may make him a candidate worth watching for a surprise result, or it may just indicate an independent who’ll achieve the rare feat of keeping his deposit.
Thirty days till he, and many hundreds of others, discover just what’s happening to that £500.