In my business, there’s a lot of fretting about the idea of representativeness. Pollsters put questions to, say, a thousand people – and take them as a sample of the country. How to be sure that you have the right sample? You need the right number of men, women, northerners, middle class, Lib Dem voters etc – but that’s the easy part. Your method of selection matters hugely, and it can skew the sample in other, less tangible ways. Every pollster is vulnerable to this so-called 'selection bias’ . If you survey people on high streets, for example, you'll get people tend to be out-and-about - rather than at home. If you survey people online (as YouGov does) you need to take steps to make sure that you’re not over sampling web enthusiasts.
