I see Kezia has picked up on Jeff's post, and is taking comfort from the fact that Labour's twenty-point deficit in the latest Ipsos Mori poll is based only on those respondents who declared themselves certain to vote at the next election. I have to say I think they're both missing the crucial point. A quick look at Mori's figures reveals that, of the 1001 people questioned, 510 said they were certain to vote - that's roughly 51% of the total. The logic of basing the headline figures only on the responses of these people is an assumption that, broadly speaking, they are the only ones who will actually turn out to vote when it really comes down to it. The additional 22% who say they are more likely to vote than not are for the most part not, as Kezia imagines, 'undecideds' just waiting to be wooed by Labour, but are in fact 'will not votes'. In other words, they're just telling the pollsters what they feel they ought to say, rather than giving an accurate assessment of their likelihood to vote.
Is this a plausible assumption? Well, 51% is certainly much lower than turnout has ever been in a modern election, but then turnout fell an astonishing 12% between 1997 and 2001, so we certainly shouldn't completely rule it out. But if on the other hand Mori had included all those who said they were likely to vote, that would take you to 73%, which is improbably high compared to the 59% and 61% turnout rates recorded at the last two elections. Perhaps Mori should be looking at a compromise of including all those who rate themselves 8 or higher on the standard 1-10 likelihood to vote scale. But what they certainly shouldn't be doing is what Jeff and Kezia seem to want them to do, ie. include all respondents in the headline figure. Whatever else you might say about a projected turnout of 51%, it's certainly a lot more plausible than a default assumption of 100% participation in an election.