Getting Brown into focus

People are talking about Frank Luntz’s focus group for Newsnight yesterday. A Labour-leaning group of voters was shown pictures, biographies, speeches and interviews to gauge their thoughts on possible Labour leaders.

A year ago, Luntz ran a similar Conservative group, which placed the little-known David Cameron well in front. This year, he found little support for Gordon Brown; John Reid was clearly ahead. Cue panic/schadenfreude at Brown’s unelectability. Is this justified?

No. I used to work in qualitative market research, and I know better than to trust generalisations from one small focus group. So, are there grounds for thinking that this group was unrepresentative?

Yes. We know from a very recent Sunday Times/YouGov poll of 1500 people that voters favour Brown over Reid by 23% to 10%; Labour supporters prefer Brown by 51% to 9%. But when Luntz, at the start of his group, asked for initial attitudes, just seven of the 30 participants went for Brown while 17 were for Reid. So the group was disproportionately biased from the start.

A focus group can be good for telling you why the people who think X do so; but it’s far too small to reliably tell you how many people generally think X. To illustrate this, a similar group run by ICM for the Guardian – their equivalent last year also favoured Cameron – produced many of the same attitudes to the Labour contenders, but numerically preferred Brown to Reid.