Worcester's Weblog - Mischief Makers

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

About this time in every general election of the ten elections I've been involved in over these 35 years of my residence here, two things happen. First "Wobbly Tuesday" (or Wednesday or Thursday, or Sunday or Monday or Friday or Saturday for that matter), when a surprising finding from an opinion poll out of line with all the others gets all the play in the media, because it is different from all the rest (as sampling theory has it, one poll in every twenty may in fact be out of line: there have been over 30 so far reported in 2005).

This is in keeping with Crewe's propositions[note1] of the way polls are reported in British newspapers and on the television and radio:

1. "However static public opinion actually is, the polls provide the media with a basis for giving the impression of flux, change and excitement. The more polls there are... the more true this is."
2. "However improbable a poll finding is, the media will publish or broadcast it... the more improbable a poll's finding is, the more likely the media will give it prominence."
3. "However clear the election outcome and the election trend, polls allow the media to hedge their bets."
4. "The duller an election is, the greater use of and prominence given to the polls."

So it is becoming in this election. In 1991 I propounded a corollary: "However close the polls get to the actual results, the academic psephologists will pick at their performance. The more polls there are, the more likely there will be something to pick at." (Ibid)

And so it goes. Now it's not just the newspapers and the academics. The mantra of the Conservative leader, repeated ad nauseam and parroted by Tory candidates destined to doom on May 5 is that "the only poll that counts is the poll on election day", and most are also repeating those tired words "that's not what I'm finding on the ground". I recall hearing that from Labour in 1979, 1983, 1987 and from Liberal Democrats in every general election since my arrival on these shores.

The trackers compound this. Every day, tiny changes of a single percentage point are reported as if they are both significant (they aren't) and are fresh information (they aren't), and carried into the polls of polls, sometimes overweighting their importance, instead of being taken into the calculation every three or four days, when there is entirely fresh data, i.e. when the 350 or so interviews done four days earlier have been dropped in favour of the latest 350.

The betting fraternity have entered the lists as well, repeating endlessly that they are more accurate than the opinion polls, neglecting to mention that the smart money follows the poll figures with the sharp eye and dedication they do with trainers and jockeys, form and fortune of the nags themselves. Some punters must bet with their hearts, not their heads, otherwise who'd put money on the Tories even on odds of 20 to 1, much less the Liberal Democrats, even at c. 350 to 1?

Second, there are the made up quotes and findings and the voodoo polls. Nearly every day I see the political parties and others making up quotes from me and other pollsters saying that they are sure to win in their constituency, reported to us by local people checking (thanks folks) to see if such a daft quote could possibly be from us (they're usually not), but then we know, don't we, that a "lie travels round the world before the truth gets its boots on." Today I had a call from a guy wanting me to stand up his internet poll which has the Tories ahead. When I asked how was the sample chosen, he asked me to explain what a sample is. Lord deliver me...

Today's polls

Two new highs for this election so far in two of the three polls out today. The Populus tracker of c. 1,400 people over four days has Labour at 41%, up just a point but a new high for Labour, and YouGov's internet panel tracker of c. 1,800 people over three days has the Lib Dems at 24%, The ICM update at 39% / 33% / 20% is just about what they've had all along, every one of their polls so far in this election having Labour at either 39% or 40% with any change in the gap accounted for by churning between the Tories and the Lib Dems.

Notes

1. "Improving, but could do better", in Worcester & Harrop, Eds., Political Communications: The General Election Campaign of 1979 (Macmillan, 1982), quoted in Worcester, British Public Opinion: A Guide to the History and Methodology of Political Opinion Polling, (Blackwell, 1991)