I’m a pollster and I’ve spent the last 20 years or so finding out what people think of companies, brands, and political parties.

In the early days with Labour, the battle was to persuade sceptical politicians that the techniques we used, especially the less well-known qualitative methods, were useful. The findings from focus groups are often so stark and vivid that this was quite an easy sell. The far bigger hurdle was to persuade the same sceptics we were reliable: that the insights were truthful and free from manipulation. Frankly, any pollster who uses their work to push their own ideas will have a very short career: any politician who believes them will lose elections. Transparency and methodological rigour are vital, both in terms of data collection and analysis.

The media has a different agenda though. It wants to reflect current political thinking and practice. It wants to be accurate. But it also wants to get a good story. At Opinion Leader Research we’ve worked with lots of journalists and understand the tension. But the imperative for good practice is just as important – maybe even more so given the influence that the media can have, especially at sensitive and volatile times.

My concern is that, increasingly, editorial opinion polling is being used to create headlines rather than unearth facts. This has been common practice in commercial research (9 out of 10 cats’ owners prefer …) but has been refreshingly rare in political published polling. But consider two recent examples.

ICM’s eve-of-conference poll set delegates to Labour’s Manchester conference buzzing. It compared voters’ views of David Cameron with Gordon Brown under the heading ‘Brown feels the Cameron effect’. However, as any seasoned pollster will tell you, before comparing attributes its important to know how important each attribute is. In other words, ask first what the salient characteristics are for a party leader (or PM) and then evaluate according to those that matter.

Yet the poll was reported (and then reported on) with the greatest prominence given to ‘having a pleasant personality’. If that mattered in a PM how come voters chose Margaret Thatcher over the infinitely more simpatico Neil Kinnock? By contrast the attribute that does count (and you don’t need to be a psephologist to know this) is ‘making the right decisions when the going gets tough’. And here Cameron trails Brown by seven per cent. Not such a story then … but, the buzz around conference was that Cameron was ‘beating Brown’ thanks to his pleasant personality, when all other research suggests Brown is ahead on the issues that really matter.

Against this backdrop Newsnight’s ‘people meter’ session run by Republican pollster Frank Lunzt appeared to confirm the bad news this time comparing Brown with other potential Labour leader candidates. On this occasion, it’s not just the analysis but the method itself that raises questions.

Luntz’s ‘focus group’ was not a focus group at all. A focus group is a recognised technique whereby a small (six to eight) homogenous group of people is moderated by a research professional. The aim is to enable all voices to be heard, and encourage all people to speak out honestly. Bringing together a heterogeneous group of more than 30 does not meet these recognised criteria. Watch the film on Newsnight’s website and witness how only the loud speak out and how ‘group think’ is encouraged rather than challenged by Luntz – lack of consensus makes less good telly I guess.

More problematic still is the choice of material used for the ‘dial test’ whereby people are shown clips and asked to turn a dial to indicate approval or disapproval. It is vital with this method to know what you are approving: the material or the man. In the Newsnight piece these were confused as we see John Reid giving a powerful anti-terror speech while both Gordon Brown and David Miliband are shown answering tricky questions about Labour’s internal problems. Clearly the end vote (show of hands not secret ballot) reflects the information that the audience has been provided with – especially given the low public profiles of all but Brown.

I still believe passionately in the value of market research techniques as a strategic aid to political decision making, but only when the standards are the highest possible and the analysis is conducted in the spirit of truth seeking not headline grabbing (although for media research commissioners it’s obviously great when the two coincide) In the US, especially in broadcast media, there are rather blurry lines between aggressive push polling and accurate political reporting. If we allow the same to happen here the sceptics will have been proved right.