If the “ICM August polls before general elections” rule works again then Ed Miliband is home and dry

August 12th, 2014

Look at the record for when LAB is in the lead

Five years ago I made a comment on PB that August polls should not be trusted because of the holiday effect and got into an email exchange with Nick Sparrow – then head of polling for ICM.

He pointed out the following from his firm’s record over several general elections which on the face of it looks convincing.

August 1996 ICM poll had LAB ahead by 12%.
The result – LAB won by 13%

August 2000 ICM poll had LAB ahead by 10%
The result – LAB won by 9%

August 2004 ICM poll had LAB ahead by 3%
The result – LAB won by 3%

The only problem with this “rule” is that it doesn’t seem to apply with the Tories. The August 2009 ICM poll had Cameron’s party leading by a massive 16% ahead looking all set for a big majority. Their vote margin in the election itself was 7%.

There is a big point about this which Oxford’s Dr Stephen Fisher should note. Following the polling debacle at GE1992 ICM’s Nick Sparrow led the way in developing a different approach to polling which didn’t grossly overstate LAB. He was proved right at GE1997 and GE2001 and by the time GE2005 came along other firms had followed the Sparrow approach.

Much of Fisher’s polling trend data embraces the big LAB poll over-staters of the past.

Last night, as reported on the last thread, ICM’s phone poll for the Guardian had CON 31, LAB 38, LD 12 and UKIP 10.

Mike Smithson

Ranked in top 33 most influential over 50s on Twitter