YouGov/Sun - CON 28, LAB 42, LD 12, UKIP 11

This morning's YouGov poll for the Sun has topline figures of CON 28%, LAB 42%, LDEM 12%, UKIP 11%. The Conservative figure is the lowest YouGov have shown since the election... though normal caveats apply, almost by definition record breaking polls are the ones that turn out to be rogues and outliers!

While there are sensible responses in the comments here (I would expect nothing less!), it has naturally released a glut of excitement elsewhere about the effect of Thatcher's death on the polls. So far there have only been two opinion polls asking voting intention since the death of Margaret Thatcher (ICM also did a quick poll for the Guardian, but it did not include voting intention). TNS-BMRB published a poll earlier in the week, but it was conducted wholly before Thatcher's death so doesn't tell us anything.

YouGov's poll on Wednesday morning had figures of CON 33%, LAB 41%, LDEM 10% YouGov's poll on Thursday morning had figures of CON 28%, LAB 42%, LDEM 12%

Now, as I hope everyone reading this knows, opinion polls have a margin of error of around about plus/minus 3 points (it is actually both more complicated than that, and a bit artificial since no polls use a pure random sample, but plus/minus 3 points is a good rule of thumb). Therefore both these polls are within the normal margin of error of CON 30.5%, LAB 41%, LDEM 11%: the average of YouGov's polls over the last month.

It may be that in coming days we get lots more polls showing Labour leads up in low teens, there is a genuine drop in Tory support and Wednesday morning's poll turns out to have been a rogue. It may equally be that we get lots more polls showing Labour's lead dropping into single figures, the Tories recover and this morning's poll turns out to be a rogue. It may also end up that the average continues to bump along at around 10-11 points and that both of these are just normal sample variation. Any of the three are possible (though my expectation is the last one!).

Anyway, if any change is a Thatcher effect (as opposed to a welfare cuts effect or something else), it wouldn't actually be that important anyway. Almost by definition, it would be a short term boost from positive coverage of Thatcher, negative comparisons with Thatcher or reminding people of their negative opinions of Thatcher - nothing of long term importance.

As ever, in writing and interpreting polls I point people back to what I wrote last year about how not to do it. Don't cherry pick polls you like and assume the others must be wrong, don't forget there is a margin of error that - more often than not - ends up explaining away the polls that people on Twitter wet their pants over.