Labour ahead in the paper that brought you the Harman poll cock-up

Back in 1992 they used to run polls like the one by Communicate Research that appears in the Independent this morning. That was the year of what Nick Sparrow of ICM calls “the great polling debacle”

The final opinion polls gave the Conservatives between 38% and 39% of the vote, about 1% behind. On the day itself exit polls were carried out and everything was pointing to a hung parliament. But when the votes were counted the Conservatives, under the leadership of John Major, had a margin of 7.6% over Labour.

Much work was done by pioneering pollsters like Nick Sparrow after that disaster and now all the firms, bar Communicate Research, carrying out telephone surveys use a system known as “past vote weighting” to ensure balanced samples.

For the big challenge is that the process of telephone polling almost always seems to produce many more people saying they voted Labour at the previous General Election than actually did so.

Thus in all the published data in the year after the 2005 General Election from the two polling firms that ask how respondents how they voted last time an average of more than 44% said Labour – compared with the 36% that the party got.

Thus, for whatever reason, getting on for a quarter more people say they voted Labour than did on May 5th 2005 and this pattern is repeated in poll after poll.

ICM and Populus adjust their figures to take part of this into account allowing for a level misremembering. Communicate Research, which has just started regular surveys for the Independent, does not.

When their first Independent survey appeared last month showing a 6% Tory lead I wrote: “This is quite remarkable given the findings. For the main impact of past vote weighting is usually to reduce the level of Labour support because telephone surveys, for whatever reason, invariably find many more people who voted Labour last time than actually did.”

All this is by way of preamble to today’s CR shares which are: CON 34%(-4), LAB 36%(+4), LD17%(+3). Clearly, as we saw in last week’s ICM poll, there has been a move away from the Tories in the past month.

I think it is a sad and retrograde step that the Independent should now be commissioning polls in a form that was so discredited a decade and a half ago.

I am emailing Andrew Hawkins of Communicate Research to ask if he wants to respond.

Mike Smithson