Winners And Sinners

No new national polls today.

WINNER: The report of a reasonably well conducted and analysed focus group done by Live Strategy Ltd for the Telegraph. Reasonably reported. It doesn't turn a half dozen or so people (seven in this exercise, in Borehamwood, Hertsmere, Cecil Parkinson's old seat, which Labour would take from the Tories if present national polls were reflected on the day) into any kind of referendum on the standing of the parties. Instead it does what focus groups should do, gather the views of a group of people in an informal but informative way, about the election.

Focus groups are tricky. They are subject to manipulation in a way that polls are not. A moderator can powerfully influence the way people talk, and can be selective in the recruitment of the panel, the questions that are asked, the way they are asked (eyebrows are powerful communicators), and what is reported. People have been known to commission focus groups to get the answer they wanted, rather than what people really are thinking.

The late Percy Clarke, former director of communications at the Labour Party, several decades ago dismissed focus groups as a waste of time, saying he might as well go to the local labour club and listen to the chaps at the bar. He underestimated their usefulness, as I believe Peter Mandelson overestimates them. I used to use them, when doing Labour's private polls, for two purposes: testing concepts, e.g. the design of the 'red rose' or of party literature, and for 'semantics', listening for the way people are talking about the issues and images of the day. They can be useful for the parties in this way, but their value for the press is less useful.

SINNER: Certainly the disgraceful way in which the Financial Times reported their focus groups (May 10). In one of the worst analyses I've ever seen, Rosemary Bennett reported "Forty per cent (sic) of the sample (sic), interviewed last week by Banks Hoggins O'Shea FCB, the advertising agency, said they had already made up their minds how to vote in the June 7 election, with two-thirds (sic) backing Labour." Later they said the panel was drawn from 76 floating voters...

So, we now know that, maybe, 40% of an unknown proportion of 76 people, defined as 'floating voters' (some of whom might not vote at all), 'have drifted', and, later, that 32 per cent (sic) said Hague would be forced out of office and another 40% said he would still be in charge after the election. Not what they thought should happen, but what they thought would happen.

Nonsense

SINNER: More nonsense from the Daily Record yesterday, who reported on 'six battleground seats' from 744 people interviewed in six Scottish seats, so reporting on each an average of 124 people in each seat with a sampling tolerance of +/- 8.1%. Thus almost all of their detailed reporting of their poll, by Scottish Opinion Ltd was useless. It's for this reason that the Association of Professional Opinion Polling Organisations has a Code that precludes reporting voting intention on fewer than 1,000 interviews. Members include Gallup, MORI, NOP, Systems 3 in Scotland and Beaufort Research, in Wales. ICM, although not a member, subscribes to the convention of not reporting on fewer than 1,000 samples.

SINNER: Finally, 'sinners' medal to Catherine MacLeod, Political Editor of the Herald, who quotes (May 10) the 'ICM (sic) poll for the Times'. It's MORI Catherine, who polls for the Times!