Comments on: ICM show 14 point Tory lead

By: Mark Senior

Mark Senior — Thu, 22 May 2008 18:17:46 +0000

Joe , the average LibDem figure in March 1992 was 16% with a couple of outliers at 12 and 14 . The final result was 18.3% . True some polls had 20% but the polls that GE were none to accurate as we know . The mayoral contest is a completely different ball game and will have as much relevance to the next GE as the last one did to the 2005 GE .

By: Joe James Broughton

Joe James Broughton — Thu, 22 May 2008 11:00:32 +0000

1992 – You reached 20 and I think 21.
The result was 18.

Also, you had a derisory 10 in the Mayoral election,
whereas polls predicted it reaching the teens (+ the inevitable Lib Dem tidbits from the “count” saying 15 on 2 May).

By: Mark Senior

Mark Senior — Thu, 22 May 2008 08:41:26 +0000

Sorry Joe , the evidence of the polls at the 1992 GE is against you and LibDem support increased through the campaign as it usually does . I am sure that in retrospect many people regret there was no hung parliament and that the LibDems were not able to resue the country from the incompetent Major years .

By: Joe James Broughton

Joe James Broughton — Wed, 21 May 2008 23:59:37 +0000

David D is wrong – the Lib Dems were squeezed at the end of the 1992 campaign.

They spent the campaign posing as the pious high principled do gooders who were going to come to the rescue demanding this and that in a hung Parliament, and the voters rightly punished them.

Nearly all the evidence – in totality – shows something similar is happening in the run up to next time.

By: Mark Senior

Mark Senior — Wed, 21 May 2008 12:30:57 +0000

This is a technical comparison of the LibDem figures in the ComRes and ICM polls and an attempt to try and give the reasons for the 3% gap between Comres at 19% and ICM at 22% despite very similar raw data figures 128 LibDems in Comres sample and 133 in the ICM sample .
The first point to note is the gap is actually 2 1/2% as the final Comres figure was a fraction under 19.5% and therefore rounded down to 19% – indeed just 1 more LibDem voter and the figure would have been rounded up to 20% .
Comres found that in their sample 128 voters said they had voted LibDem in 2005 and adjusted the 128 now saying they would vote LibDem up slightly to 134 . This appears to be a little more generous weighting than they have used in previous months . The ICM sample by contrast had only 93 people saying they voted LibDem in 2005 much lower than expected circa 130 and those saying they would vote LibDem now 133 . They adjusted their weighted figure up to 155 giving a weighted figure with a LibDem figure 21 higher than Comres which is equivalent to around 3% of vote share .
The final published figures in both cases are also affected by adjustments for likelihood to vote and treatment of don’t knows but the effect here is overall 1% or less so not significant .
A good check between the 2 pollsters can be made by looking at the voting intention changes of those who voted in 2005 .
Comres
Lab to LibDem 20 LibDem to Lab 6 net 14 to LibDem
LibDem to Con 17 Con to LibDem 4 net 13 to Con
ICM
Lab to LibDem 24 LibDem to Lab 6 net 18 to LibDem
LibDem to Con 21 Con to LibDem 3 net 18 to Con
These figures are pretty consistent and on their own would indicate movement since the 2005 GE from LibDem to Conservative matched by an almost equal movement from Labour to LibDem leaving LibDems at around the same level of support as in 2005 . It would be nice to include the movement between LibDem and Others but annoyingly ICM do not give the movements from the minor parties .