When phone polling was first introduced in the UK it attracted flak over its accuracy (least remembered about those ASL polls the better), but also gained popularity through both its lower costs compared to the then dominant face to face polling and also its greater flexibility. The same pattern has been seen again with the spread of internet polling the UK. Just as face to face polling used to be the gold standard and phone polling the upstarts, now phone polling is the gold standard and internet polling the upstarts.
The merits or otherwise of YouGov have been much debated elsewhere but there are also many much newer internet polling firms, some of whom had poor 2010 general elections. A newbie since then is Survation, whose record has yet to be tested against actual results in that way.
Its record in Parliamentary by-elections, always tricky given the big movements that can happen in the last few days of a campaign, has however shown some primse. A big under-estimate of the Lib Dem vote and over-estimate of the Conservative vote in Leicester South but across the board very close to the final result with its figures in Feltham & Heston and Barnsley. Survation is also a member of the British Polling Council, abiding by its ethical and transparency standards.
Apply therefore a couple of pinches of salt, but not the whole packet, to its polling figures for the London Mayor and Assembly elections. In Survation’s favour on the Assembly elections is that their polling tries to replicate the actual voting experience, taking people through the three different ballot papers in turn.
Here are the results from their fieldwork April 18-24 and with changes shown from the 2008 result:
Mayor first preferences
Johnson 42% (-1%) – wins 55% – 45% in the second round
Livingstone 31% (-5%)
Paddick 10% (n/c)
(Everyone else 5% or under)
Labour 33% (+6%)
Conservative 28% (-6%)
Lib Dem 10% (-1%)
Green 8% (nc)
UKIP 7% (+5%)
(Everyone else under 5%)
Striking how the poll shows Labour’s vote up but Livingstone’s vote down, reinforcing the comments that many – including Labour members – have made about how Livingstone had changed from a candidate who used to be able to appeal beyond Labour supporters to one who struggles even to hold on to Labour supporters.
The poll also has breakdowns for individual GLA constituencies, but as the samples sizes in each of them are so small the figures mean very little. (The random sampling margin of error alone on each constituency figure is around +/- 10%.)
However, what this poll does throw up is how close the Assembly seat results are likely to be with. With a 5% threshold having to be passed to win any seats via the top up list, a small fluctuation in support for UKIP in particular could be the different between 0 or 2 seats, and if it (and any other parties) fall just below the threshold, that means extra seats to go around amongst those above it.
Moreover, if the poll is accurate, it suggests the Conservatives could lose slip below the one-third of votes necessary to get through the Mayor of London’s budget. (The GLA can only reject the Mayor’s budget by a two-thirds vote.)
The way of course to make sure these possibilities are all more likely to come down in a good way for the Liberal Democrats is to help maximise the vote for Brian Paddick, Caroline Pidgeon and colleagues in the last few days of the campaign.