From a conference I’m attending today organised by Brunel University’s Magna Carta Institute. First speaker, Peter Kellner, from polling agency YouGov. Hastily-typed notes follow:

“I’ve never been so uncertain about an election that’s only nine days away than I am today”

Last time, more people voted for a party other than Labour/Conservative than Labour and Conservatives combined

Lib Dems seem to do worse in Proportional Representation elections than First Past The Post

The expenses scandal’s effect on the polls seems to have stabilised, but could be more twists

Current projections (as of the morning of 26th May, so likely to shift)

Tories – 26-28% (roughly the same as 2004) – a month ago would have been 35% at least

Labour – c.20-22%, though wouldn’t be surprised if they ended up on 18-19%
(If turnout = 35-40%, that means only c.8% of the electorate voting for the governing party)

Lib Dems – 15% last time, and likely to be there or there abouts this time – maybe 16-17%

UKIP – 16% last time; at the moment looks like they’re slightly up (17-19%) – but 3 weeks ago = 7% (Telegraph expenses stories boosted them to 19% in a week); “the key thing we find is that UKIP voters are the most determined of all the supporters to say they are going to vote”; BUT: “past evidence shows ‘certain’ voters aren’t certain at all”

Greens – 9% – last time = 6% and got two seats, but largely due to London and the South East having most MEPs, so a lower threshold – wouldn’t have got seats in any other region. 1-2 more seats maximum, depending on how close the fight is for the last seat in other regions – benefiting from the expenses scandal

BNP – last time got 5%, but no seats (their strong regions have a higher percentage threshold for election due to fewer MEPs); this time it’s possible they may get to 8-9% in London and get a seat; North West (Nick Griffin); Yorkshire and Humberside; West Midlands
Note 1:– it’s a tiny margin between getting NO seats and getting FOUR seats for the BNP
Note 2: latest YouGov poll puts them at 7%; ICN poll in the Guardian on Saturday put them at 1% – probably due to reluctance to admit to BNP over the phone; YouGov = online, so possibly more anonymous and honest

SNP / Plaid Cymru – will probably keep the same number of MEPs and similar share of the vote

Northern Ireland – different electoral system – single transferable vote, so likely to be similar in outcome to 2004, though likely with Sinn Fein with the largest share of the vote

Turnout – this was c.25% 1999, but there were no local elections; in 2004 it went up to 35% last time (almost certainly due to local elections – there were no local elections in Scotland in 2004 and turnout was only 30%)
Note: Because county elections are taking place the same time as EU ones this time around, turnout is likely to be above 30% again
Possible party effect: Local elections this year are mostly in Conservative areas (countryside not cities); Tory areas tend to have higher turnout than Labour anyway – so Tories may end up doing slightly better thanks to local election turnout boost