Under a headline "Polls predict big Labour win in Oldham election" today's Guardian reports that "(a)n ICM poll in the Mail on Sunday put Labour on 44%, Lib Dems on 27% and Conservatives on 18%"
However an Email from Liberal Democrat party president Tim Farron cites the very same poll as predicting a narrow Liberal Democrat lead!
Mr Farron quotes the raw data on which the poll is based as follows:
The raw, unweighted responses were as follows for the 549 calls
Refused to state intention 256 46.6%
Elwyn Watkins 89 16.2%
Debbie Abrahams 85 15.5%
Undecided 68 12.4%
Kashif Ali 27 4.9%
Another Party 24 4.4%
The Mail on Sunday must have done same pretty imaginative "seasonal adjustment" to convert that into a Labour lead of 17 percentage points.
One of the most significant criticisms of the press in Nick Davies's "Flat Earth News" is that the media magnates have so pared down their reporting staffs that there is little time for anyone to check stories. It looks as though the Guardian is not exempt.
One must also question the reliability of conclusions based on such a tiny survey in which over half either refused to answer or hadn't yet decided.
I'm aware of the practical difficulties, but it seems to me that a ban on opinion polls during elections, and particularly by-elections (and also those dubious graphics in Liberal Democrat literature purporting to show that only we can win) would go some way to curb the herd instinct and improve the quality of our democracy.