How good are current ratings as a pointer to the general election?

The fact that there’s been absolutely no change in the CON-LAB shares in today’s Guardian ICM brings to mind the success this series of polls has had in the run-ups to the last three general elections – even from this far out.

Just look at the table above showing all the ICM polls for the paper from October 2003 to May 2005 – about the same distance we are out now from the assumed May 2010 general election.

Examine the numbers for each of the parties in turn and observe how little movement there was in the October 2003 – May 2005 period. The Tories were locked in a range of 30%-35% with nine of the surveys putting them on 33% – what they actually achieved.

Labour are in a four point range of 36% – 40% while the Lib Dems, apart from one poll, were solidly in the 20% – 23% range.

So 2005 was a fluke I can hear you saying? Well it wasn’t. Just check out this link here for the same period ahead of the 2001 election. Apart from the freakish circumstances of the September 2000 fuel crisis the consistent picture looks very much the same. The ICM Guardian poll series performed almost the same ahead of 1997.

One of the drivers behind ICM’s consistency is their approach to the Lib Dems. The firm tends to produce the best ratings for the party and usually the most accurate. This is probably down, to the precise wording of ICM’s voting intention question and the operations of its past vote weighting formula and spiral of silence adjustment.

So if the ICM record over the past decade and a half is to be our guide then the current spread betting ranges suggesting a Tory majority 0f 40-50 are about right. That’s no Tory landslide but no hung parliament either.

Mike Smithson