How do you square Dunfermline with the national polls?

According to the latest round of surveys from all the main pollsters Labour is performing at or above what it got at the May 2005 General Election. Yet on Thursday its support collapsed and we saw such a massive swing to the Liberal Democrats – which has been recording drops of upto 10 points on last year.

From what I can gather Dunfermline was the first occasion in modern times when the governing party has lost a Westminster by-election when its poll ratings were ahead of what it got in the preceding General Election.

When Labour lost Brent East in September 2003 its ICM rating that month was 35% – seven points less than it recorded at the 2001 General Election. So clearly it had a struggle on its hands.

Labour had the same 35% ICM poll rating ten months later at Leicester South in July 2004 when the Lib Dems took the seat.

Going back through the Tory years I cannot find a single by-election loss by them to either Labour or the Liberals/Lib Dems/SDP when the party was recording higher poll ratings than at the preceding General Election.

Of course there are local factors in all by-elections but the national picture, as represented by the poll rating must count for something. And, indeed, this by-election was totally exceptional.

The answer, I believe, might lie with the way the pollsters have “processed” the Labour vote since the General Election and this needs looking at. The weightings given Labour declarers in their surveys might be too high. This is something that we will continue to watch.

Mike Smithson