I’m a bit of a poll geek. I quote them a lot. People are often sceptical.
So here are some questions answered and some myths busted.
- “They didn’t ask me, or any of the people I know.” There are 60 million of us in the UK, so the chances of being surveyed are pretty low. I’ve been interviewed twice by polling companies in 30 years. Neither time was I asked my voting intention. Most polls are of around one to two thousand people, with as much of a representative sample across age, income group, gender, region and so on as possible.
- “That’s a ridiculously small number, no wonder they are wrong.” Polls have a margin of error of plus or minus 3%. Polling more people does not change that margin of error. Some polls do cover up to 20,000 people (a very costly exercise) but are no more representative. Anthony Wells of UK Polling Report explains it better, using a soup analogy from George Gallup, here: http://ukpollingreport.co.uk/faq-sampling
- “But the polls have been wrong. Look at some recent elections.” Actually some pollsters have been bang on in recent General Elections, some have been out by around 3% – the margin of error. Polls were out in 1992 and 2015, and each time pollsters have adjusted for specific factors that contributed to that error. For example in 1992 it was found that many people were not keen to admit they voted Tory, so that was subsequently taken into account: http://ukpollingreport.co.uk/faq-dont-knows
- “Oh come on, we all know the pollsters are biased, and owned/founded by/run by Tories.” No. Political opinion polling is the most high profile thing pollsters do, but it is only a relatively small part of their business. Most of their work is market research and opinion polling on consumer habits for major companies. If pollsters deliberately got political polling wrong, their main customers would not trust them or pay them to find out where and how to advertise their products. Reputation is the main reason the pollsters have to try their utmost to get it right.
- “But that poll last week said something completely different.” Remember the margin of error, plus or minus 3% either way. So if Party X is at 37%, they could be as high as 40% or as low as 34%. And different pollsters have different methods of surveying people, either online, on the phone or sometimes in person, which can produce slightly different results. What you need to do is look at trends over time, and across the nine or ten main pollsters. And look behind the headline voting figures, at how the Leaders and their policies on the economy and other key issues are viewed.
- “I did a poll on Twitter/Facebook, and fifteen thousand people voted, and that proves the pollsters are wrong.” The people you are connected to on social media, much like your friends down the pub, are far more likely to share your views than the general population, so it’s not a representative sample. And if anyone can pile in and vote, the results can be skewed massively by a campaign group or political party getting their supporters to vote.
In summary, polls are not perfect, but they are scientifically-conducted representative surveys of the views of the population as a whole. They are the best guide we have. Any one poll is a snapshot in time, a series of polls will show a trend. Events happen and views change.
There’s a similar article from The Independent on polling here, if you don’t believe me: https://www.independent.co.uk/news/uk/politics/opinion-polls-should-we-believe-them-trust-truth-real-fake-news-a7704116.html