Will the pollster perform better than it did in Scotland?

A new London survey tonight from a firm that is not listed as a member of the British Polling Council provides a big boost to Ken as he seeks to defend his mayoralty.

The firm, MRUK, found first preference figures of Ken 45%: Boris 44%: Paddick 9%. After second preferences are taken into account Boris and Ken are on 50:50.

The survey took place from April 7th to the 14th and the results were weighted to“match the electorate”. There is no indication whether likelihood to vote was factored in or whether the firm sought to ensure a politically balanced sample. The fact that this is not mentioned leads me to assume that these elements were not part of the methodology.

We last saw MRUK at the Scottish elections last year when they over-estimated the Labour and SNP shares and grossly underestimated the Conservative and Liberal Democrat positions. This was their prediction in the constituency section with comparisons on the actual result: LAB 34% (actual 32.2%): SNP 38% (32.9%): CON 11% (16.6%): LD 13% (16.2%)

When the firm’s Scottish poll was published in April 2007 I offered them a wager of £1000 that they would be overstating the Labour vote. Sadly for my bank balance they, nor the paper that published the survey, took me up. I am prepared to repeat that with this London poll. How about it Mr. Sunday Times?

These latest London numbers are very much in line with what MORI found last week – BEFORE the firm applied its turnout filter. After that calculation had taken place MORI reported a 6% deficit for Ken on first preferences.

If I am right about the methodology this survey is good news for Boris and bad news for Ken. The real issue with an election like the one on May 1st is turnout and unless this has been measured and applied to the figures then this looks like one of those old fashioned surveys that got the 1992 general election result so wrong.

It is a real shame that the Sunday Times has not commissioned a firm that is a British Polling Council member which would have given us greater confidence in its findings.

UPDATE: The story about the poll has now appeared on the Sunday Times website and does suggest that the question of likelihood to vote was put. What we do not know is whether this was factored into their figures.

There’s a quote from a director of MRUK that “Turnout looks like the key – Ken can win if his natural followers make it to the polling booth, whereas the support for Boris seems slightly more solid in terms of likelihood to vote”. But how was this applied to the figures that have been published? Did they ask whether respondents were even registered to vote – something that the other pollsters will be doing in the next few days?

If MRUK was a member of the BPC then the full detail would have to be made available within two working days and we could come to a proper assessment. Let’s hope the firm does make information available.

It is perhaps worth reiterating that every single telephone poll ahead of the 2000 and 2004 mayoral elections overstated Ken’s position by a considerable degree.

Mike Smithson