this was predictable, actually.

the survation poll - the one where the tories are only up by 1% - is the only one with an undecided option. whether that is good or not is a tricky question.

so, the polls where you can't pick "undecided" have the tories up by as much as 13 points, whereas the polls that allow you to pick 'undecided', and track 'refused', only have the tories up by a few points. one even has labour winning by 2.

polling companies tend to dislike the 'undecided' response because it decreases confidence in the results. but, this is a business consideration. they are scientifically superior.

the survation poll, then, and it's only one poll, but what it actually says is:

tories: 33.9%
labour: 32.6%
undecided: 12.1%
lib dem 6.4%
refused: 6.3 %

what that means is that 18.4% of the sample did not answer.

"but it distributes!"

well, it might seem a reasonable suggestion that it might, but then why is this poll so different? let's at least distribute the refuseds:

tories: 36%
labour: 34.6%
undecided: 12.9%
lib dem: 6.8%

the biggest difference in the other polls was 13%.

so, there is a bigger consistency here than initially apparent.

but, these swings usually exist on the left. certainly, they always do in canada. so, this is a bit of a reversal of the expectations and may confuse some models.

see, the "shy tory" effect was identified in a period where the media and broad society was clearly on the soft left. tories were afraid to speak in public, lest they be outed as such. but, i think the situation has reversed. the same conditions would instead apply to a "shy socialist" effect, as it is the less socially acceptable position in the uk as of today. and, this is consistent with the results - when you force people to answer, the undecided voters claim they will vote tory in disproportionately higher numbers, no doubt because that is what the media is implying they ought to do.

i've identified this problem elsewhere: the election is in the hands of the least capable voters. what this 13% is telling pollsters is that they don't know who they should vote for, but are leaning towards doing what they're told to do by establishment media. they could, of course, be lying. but, we have to assume that they aren't.

i think that this election may actually work the opposite way in turnout, as well. if a lot of these very soft tory voters stay home, it could end up closer than expected.

i think there is reason to expect a hung parliament, or something close to it. i do not expect a landslide, at least.