YouGov/Sunday Times - 36/41/10
Today's YouGov poll is up on the website here. Topline figures are CON 36%, LAB 41%, LDEM 10% - this is pretty typical of YouGov polls of late, but if the elections were to have a big impact on public opinion we wouldn't necessarily see it yet anyway - the fieldwork for the poll would have been conducted between Thursday afternoon and Friday, so not really in time to pick up any effect.
Looking through the discussion yesterday there were lots of questions about vote shares in the locals - here's a brief summary of the various different figures that will emerge, and what they mean
Firstly there is the BBC's Projected National Share - which I think ended up at CON 35%, LAB 37%, LDEM 15%. This is based on a selection of key wards across the country, not the sum of all votes, and are plugged into a formula to produce a projected national share of the vote if there were local elections across the whole of the country. It is important to note, therefore, that this is NOT skewed by there not being any local elections in Scotland, London or Wales - it is adjusted to take account of this. It's also important to note that it is a projection of local support across the country, not Westminster support across the country - people do vote differently in local elections (and indeed, if you poll people on their voting intention in Westminster and local elections they give significantly different answers - in YouGov's final pre-election poll they asked local voting intention and found the Lib Dems on 15%, compared to 11% in the standard Westminster question).
Secondly there is Rallings and Thrasher's National Equivalent Vote. This is calculated on a similar basis to the BBC's projection, and is intended to give a picture of what the vote shares would be if there were local elections across the whole country (i.e. it is adjusted to take account of there not being elections in London, Scotland, Wales etc). All the caveats about the BBC one also apply here - it is not intended to be a projection of Westminster vote and is not comparable to Westminster polls. Ralling & Thrasher's NEV does not come out on the night - it normally takes them a few weeks to collate the data. (Actually, they've produced it already! CON 38%, LAB 37%, LDEM 16%. It is normally slightly different to the BBC's due to different wards being used and different formulas being used for the projection).
Thirdly there are the actual shares of the vote, the sum of all the votes cast in all the local elections. This is again done by Rallings and Thrasher and takes a considerable amount of time to collate. In many ways this is the least useful data, since it is skewed by things like there not being local elections in London, Scotland and Wales and (depending on how the figures are calculated) people in councils with all out elections will often have three times as many votes as people in councils that elect by thirds.
UPDATE: More on the difference between the BBC's PNS and Ralling & Thrasher's NEV. The reasons seem to be the number of wards used (NEV apparently uses a larger number of wards from councils declaring on Friday, since R&T have an extra day to work out the numbers), different baselines (R&T base the figures on change from previous years election results, I think the BBC's PNS does much the same, but over time the two baselines will diverge).
As well as adjusting for the different parts of the country having elections each year, the BBC's PNS also attempts to adjust for the different pattern of candidates standing, so is a projection of what would happen if the whole country had elections that day, and if every ward had candidates from the three main parties and one other standing. I'm not sure if R&T's NEV tries to do this or not.
