Newcomers Survation show tighter race in OE&S
David Wooding of the News of the World is reporting that the long-awaited Survation poll of Oldham East and Saddleworth has voting intention figures of CON 6%, LAB 31%, LDEM 30% (these figures imply a huge share for "others", but my understanding is that this is because it was not repercentaged to exclude don't knows.
If don't knows were around about 25%, then these figures are the equivalent of something like Labour 41%, Lib Dems 40%. Obviously these figures have a much harder squeeze on Conservative support and, consequently, have the Lib Dems much closer to Labour. Until we see all the detailed tables of the polls tomorrow or Monday, it's hard to speculate on what might be reponsible for the contrasts (Survation are putting up their tables at midnight, Populus on Monday, ICM presumably Monday).
Meanwhile, YouGov's weekly poll for the Sunday Times has topline voting intention figures of CON 38%, LAB 41%, LDEM 10% (suggesting that 7% for the Lib Dems in the week was indeed just normal variation within the margin of error).
UPDATE: There are more details of the Survation poll on their site here. There are not full tables, but enough to judge that ICM and Populus are more likely to give us an accurate impression. Survation seem to have suffered from an exceptionally high refusal rate of 47% (perhaps it was their questionnaire, or introduction to it - ICM typically get about a fifth of that) meaning the final voting intention figures were based on just 225 people, giving a very high margin of error.
They haven't provided their weighting targets, but the weighting regime itself was very simple - age, gender, past vote. Compare that to the more complex regimes used by established phone pollsters, which include social class, tenure, work status and so on. I'd stick with the experienced hands on this one (though as Robin Hood has mentioned in the comments, we won't really be able to compare these polls against the actual result because the fact these three polls have been published could itself change people's votes).
There are also some more details of the Populus poll here
