Why even this 4 percent lead on votes projection might not be enough for the Tories to win most seatsJanuary 31st, 2014
— PolPics (@PolPics) January 31, 2014
My 4/1 bet that LAB wins most seats but CON gets most votes
Yesterday I placed a bet at 4/1 with PaddyPower that LAB will win most seats at GE2015 but will trail the Tories on national vote share.
What makes this a value bet is because the vote share window when this could happen is much larger than the commons seat calculators suggest.
Look at the uniform national projection above from Electoral Calculus in which the Tories retain the 37% of 2010 but LAB only edges up 3.3% from the 29.7%. On the face of it the projection looks fine but there is a huge problem that is not factored in.
This is, of course, CON-LD battlegrounds which are the only source of blue gains, to offset losses to LAB in the model above. The projected CON total includes 22 many of them in seats they would have won on a universal swing in 2010 but didn’t
The evidence is that strong yellow incumbency makes these difficult nuts to crack and the blues will do worse here than the seat calculators point to. This was supported by the Ashroft marginals polling which found a swing of just 0.5% from the LDs to the Tories.
Interestingly, as well, a University of Nottingham study last year suggested that GE2010 would have produced an additional 13 CON seats but for Lib Dem incumbency.
The underlying issue, of course, is that the LAB seats:votes ratio is much lower than for the Tories and other parties. When the ComRes 1% LAB lead poll came out on Monday I ran the party shares through the Electoral Calculus seat predictor which pointed to LAB getting 327 seats just above the 325 seat threshold for a majority.
The main reason for the electoral bias, of course, is that LAB voters are generally much more reluctant to turn out to vote where it doesn’t matter – notably CON and LAB heartlands.
GE2015 will be decided in 100 seats with the other 531 mainland being almost foregone conclusions. In the latter I’d expect even poorer turnout levels by LAB voters than we’ve seen before.