ICM reports an 11% Tory lead
After a week in which the Conservative lead in the opinion polls has ranged from 11% down to 3%, it bounced back to 11% in the latest ICM opinion poll.
This ICM poll gives the party shares as:
CONSERVATIVES 42% (-1)
LABOUR 31% (+1)
LIB DEMS 19% (+1)
Practically every opinion poll for a year has estimated the Conservative share of the vote at 40% or above, including all the polls in the past week, but Labour has been between about 31% and 37% and the Lib/Dems between 12% and 19%.
So the range in Conservative opinion poll leads in the past week - 11%, 5%, 3%, and back to 11% - has largely been due to different estimates of how support divides between Labour and the Lib/Dems.
It is also noticeable that the media has given much less attention to the opinion polls showing the Conservatives still more than 10% ahead than to those with a narrower lead.
It is worth remembering that if a poll has a sample size of 1000 or so, as most of them do, the standard margin of error is about three percentage points for each party's estimated share of the vote. As the poll lead is the difference between the top two shares, the standard margin of error on a poll lead is rather more than that. For that reason anyone who is really interested in what opinion polls are saying is wise to pay more attention to the shares of the individual parties than to the lead.
The opinion poll with the latest 11% Conservative lead was ICM, while another poll which had a much lower Lib/Dem share, higher Labour share, and lower lead was Ipsos MORI.
The main difference between ICM and MORI is that the former takes measures to ensure a politically balanced sample through past vote weighting. MORI doesn’t and as Mike Smithson of the "Political Betting" website was arguing on Wednesday the latest survey simply had far too few 2005 Lib Dem voters in it to make it representative.
As Mike put it
"What’s happening at the moment is that it’s the way the different firms are polling the Lib Dems that is driving the top-line numbers."
and he added
"My view, as I have argued here before, is that the ICM methodology is sounder than MORI’s."
This ICM poll gives the party shares as:
CONSERVATIVES 42% (-1)
LABOUR 31% (+1)
LIB DEMS 19% (+1)
Practically every opinion poll for a year has estimated the Conservative share of the vote at 40% or above, including all the polls in the past week, but Labour has been between about 31% and 37% and the Lib/Dems between 12% and 19%.
So the range in Conservative opinion poll leads in the past week - 11%, 5%, 3%, and back to 11% - has largely been due to different estimates of how support divides between Labour and the Lib/Dems.
It is also noticeable that the media has given much less attention to the opinion polls showing the Conservatives still more than 10% ahead than to those with a narrower lead.
It is worth remembering that if a poll has a sample size of 1000 or so, as most of them do, the standard margin of error is about three percentage points for each party's estimated share of the vote. As the poll lead is the difference between the top two shares, the standard margin of error on a poll lead is rather more than that. For that reason anyone who is really interested in what opinion polls are saying is wise to pay more attention to the shares of the individual parties than to the lead.
The opinion poll with the latest 11% Conservative lead was ICM, while another poll which had a much lower Lib/Dem share, higher Labour share, and lower lead was Ipsos MORI.
The main difference between ICM and MORI is that the former takes measures to ensure a politically balanced sample through past vote weighting. MORI doesn’t and as Mike Smithson of the "Political Betting" website was arguing on Wednesday the latest survey simply had far too few 2005 Lib Dem voters in it to make it representative.
As Mike put it
"What’s happening at the moment is that it’s the way the different firms are polling the Lib Dems that is driving the top-line numbers."
and he added
"My view, as I have argued here before, is that the ICM methodology is sounder than MORI’s."
Comments