First up tonight is Communicate Research in the Indpendent on Sunday which has LAB 39 (-1): CON 31 (-4): LD 23 (+5).

Next Thursday will be the first time CR has been tested against real results. It remains to be seen whether it can perform better than the pollster the Indy on Sunday used four years ago, Rasmussen, which with its automated phone methodology was the only firm to get the Tories right and got the Labour margin within 2%. All the others underestimated William Hague’s party. CR do not weight by past vote recall and until they introduced party prompts three weeks ago reported the lowest LD shares.

Second up is YouGov with LAB 36 (nc) : CON 33 (+1): LD 23 (-1) . So the only change since Friday has been a slight shift from the LDs to the Tories.

Although the embryonic YouGov did carry out polls for the 2001 General Election this will be the first major UK test for the internet-based firm which is so often criticised by other pollsters. Until now, however, whenever it has been tested in UK elections it has come out pretty well. This is not its final polls and perhaps two more surveys are expected before polling day.

Third up is ICM poll for the Sunday Telegraph which has LAB 39 (-1): CON 31 (-2): LD 22 (+2) so unlike YouGov a move from the Tories to the Lib Dems.

ICM’s final poll in 2001 had an 11% Labour lead compared with the 9.3% that actually happened. Another ICM poll started at the same time with a larger sample had a 17% Labour lead. The pollster has not carried out any surveys since 2001 when it could be compared with real results. Further ICM polls are expected before polling day.

My big bet – several hundred a percentage point on a spread selling Labour at the equivalent of 38.6% – looks pretty good. Two polls showing Labour down a notch to 39% and the third showing no change.

Mike Smithson