This is a table adapted from UK Polling Report showing all the published ICM polls during 2008.

Below is a table showing all the ComRes polls, excluding last night’s sensational CON 37%: LAB 36%: LD 17%.

We also had in the Observer in the Observer on Sunday the poll from Ipsos-MORI. I’ve not featured its 2008 table because it has gone through a big methodological change.

My main observation from the two tables is that ICM is the more stable with its shares for all three main parties. The pollster was reporting higher shares for Labour at the start of the year and during Gordon’s dreadful summer. In the past couple of months, though, it has been reporting significantly lower shares.

ComRes had some of the worst numbers for Labour in the first part of the year right into the summer but now has some of the best.

Both firms adopt a fairly similar approach. Polling is done by phone and quite often ComRes uses ICM’s telephone call centres. Both firms filter by certainty to vote and both firms use past vote weighting.

A possible difference which might have an impact is the form of the voting intention questions themselves. ComRes puts its question simply in one sentence “If there were a general election tomorrow, would you vote Conservative, Labour, Liberal Democrat or some other party?”

The ICM question is broken down into two sentences and reads: “The Conservatives, Labour, the Liberal Democrats and other parties would fight a new election in your area. If there were a general election tomorrow which party do you think you would vote for?”

In each case the party name prompts are adapted for Welsh and Scottish respondents.

My preference is for the ICM format which I think is clearer when it is communicated on the telephone and I think produces a more considered response. The one sentence approach is a bit of a mouthful – just trying saying it out loud. The two sentence approach sets the scene before putting the crucial question.

All we can say is roll on the next polling round.

Mike Smithson