If we are even close to 97 style tactical voting (and that's far from the current reality) it's always best to play with regonal breakdowns and 'class' breakdown (in a similar style to how the Fabian Society suggested that the 'E' voter shifted significantly to the Lib Dems in 2005 IIRC) We need to look at the 97 stats, then see what happened in 2005 and project or guess what could happen next. It's all rather complicated. Predicting tactical voting is a fools game
That makes me a fool too!
If we take the last populus poll (36:32:20) and put it into the 'electoral calculus' we get
LAB 292
CON 278
LIB 47
If we reverse what happened in 1997 in order to help the Conservatives in the same way Labour were 'helped' in 1997 we get
CON 316
LAB 252
LIB 49
If we add into the mix, Conservative voters voting Liberal Democrat in seats where the Lib Dems are second to Labour we get
CON 316
LAB 247
LIB 54
If we do it all over again, but for the ICM poll (40:31:22) we get:
CON 330 (MAJ 10)
LAB 233
LIB 55
and with tactical voting
CON 364 (MAJ 78)
LAB 192
LIB 62
All 'just a bit of fun' of course