Worcester's Weblog

MORI chairman Sir Robert Worcester analyses the latest opinion poll data.

There's a great deal of misinformation about (as usual) in the 'phoney war' so far. Tim Hames wrote in the Times that MORI's surveys are face-to-face while the rest are by telephone, ignoring that the first MORI poll of the campaign, in the FT, was 1,001 electors by telephone.

The bookies are all saying -- as are some pundits -- that the best predictor of the 'real' state of the parties is where the money is being bet with them, not noting that many of the bets follow the findings of the polls.

The Racing Post has an intriguing feature that they are publishing daily, giving the bookies' betting odds on the election, seats, majority, turnout and as it goes along, no doubt many other quirky things on the election. Speaking of betting, if you want a punt I understand a fellow named Cardinal Angelo Scola is a good bet to be the next pope. If you do, and he wins, I'll have 10 per cent; if not, don't blame me!

At this time, we begin to get some quirky polls indeed. I recall doughnut polls, chocolate replicas of party leader polls, tie polls, bumper strip polls and other phoney polls that are a bit of fun and give the perpetrators of these nonsenses a bit of publicity, and what's the harm of that? I'll try to report them as I see them.

The real phone(y) polls are those that are phone-in polls which are done by newspapers, radio stations, television and even the New Statesman, which ask biased questions of an unrepresentative sample, are not 'representative' in that they are self-selecting and not able to be projected to any known universe, like electors in Great Britain. Watch out for them.

Tell tale signs are questions that are loaded, frequently of the yes/no variety, and even sometimes ungrammatical, such as last week's (typical) New Statesman's "Does (sic) any of the UK political parties understand women?", to which an unknown number of readers went to the NS website and are therefore NS readers, and are nerdy enough to bother to answer, and who are unrepresentative of anybody other than those nerds who are readers of the NS who bothered to take the trouble to go on to the NS website to vote in a meaningless poll.

That 45 per cent of those undisclosed number who took part voted 'yes', is neither here nor there, and is in effect meaningless. I'm only surprised they didn't carry the results out to one decimal place, a sure sign of an amateur poll.