Early Edition with Kate Hawkesby

Polls are great. I love ‘em and I love the occasional surprise we get from them.  They're really useful in providing those snapshots of public opinion at any given moment.  And you'll be aware the political parties use them extensively, not just to see how they're tracking but for all sorts of nefarious stuff like testing out our reactions to policy ideas.  On top of that there's also a lot of research into the way leading figures are perceived, all part of the process of moulding and shaping them for maximum ballot-box efficiency.  But the thing about polls that we rarely take much notice of, and should, is the sampling process.  It's vital, if you want accurate answers to your questions, to be scrupulous about the sampling process.  Let me explain.  Say you're doing a political poll; you want to be sure your respondents match the demographics of the population as a whole.  So, you end up having to interview so many men, so many women, and a set number of people in each age group to match the voting populati