The technology behind the opinion polls

Whether you believe they're gospel or it's a case of lies, damn lies and statistics, you have to appreciate the power of the opinion poll. They have politicians quaking with fear and can even be the very fuel to make their own predictions come true. So, with so much riding on these fast, hard hitting, public surveys we spoke to Ipsos MORI to find out just what technology the pollsters use to put them together and how it crunches those numbers.

"We'd love to be able to use online polling but sadly it would just mean that we only get a biased sample demographic", said a MORI spokesperson. "Not everyone's online or comfortable enough with it yet for us to get a fair representation of people's opinions, so it's largely done by traditional hard graft over the telephone".

The company has call centres in Harrow, Edinburgh and Lowestoft full of 1000 employees working morning, daytime and evening shifts in order to capture enough of each age bracket of society for a statistically significant sample. In the monthly political polls that MORI carries out, the aim is to interview between 1200-1500 people all round the country. Their directory data isn't limited either with a special clause so that pollsters can get access to even ex-directory numbers. Naturally, that doesn't always make for happy interviewees.

"Most people are quite happy to go through the questionnaire but some get upset with the invasion of privacy and we actually also get calls from people angry that they haven't been phoned up for interview as well. The truth is that with sample sizes of 1200 and 60 million people living in the UK, the odds of getting a call from us are quite slim".



Each questionnaire is carefully put together and placed on call centre's computers in tailor-made applications which then guide the employees through the interview one screen at a time. After each call, the raw data is sent off to the head office where it is weighted according the demographic of the person who was interviewed and the importance of each question before the findings are sent off to the clients.

Older groups get more of a bias because the 18-25 section are the least likely to turn out to vote. By the same token Labour supporters are also less likely than Conservatives to make it to the ballot box, so the results from those interviews have less of a significance on the results too but, of course, it's not just the elections that pollsters like MORI cover.

"Politics is only a small part of what we do. It's a large part of our public sector work but the majority of our polls are for the private sector working in advertising, marketing, customer loyalty, brand reputations and areas like that. We're also not allowed to take on any new public sector work within 6 weeks of an election being announced".

The results of any public sector polls, whoever they're commissioned by - News of the World, Reuters etc - are sent over as PDF bundles of largely pie charts and graphs and must legally be published on the MORI site, and put into the public domain within 48 hours. Normally, though they aim to get this done within at least 24 giving the media agency clients enough time to break the results first.

In the short term, there are no plans to carry out the surveys online - at least not until the internet generation are silver surfers themselves - so, if it's heavy technology, swing charts and voting graphics you're after, you'll have to wait for the election night specials.



>