Is John Curtice right about it being a myth?

One of the biggest errors that many pundits make is when they compare the numbers from one pollster with another and then conclude that there is a trend. You see this all the time and it is wrong.

The six pollsters that regularly carry out voting intention surveys in the UK each have their own bespoke methodology which is unique to them. The only valid comparisons are when you compare like with like and that means looking at each pollster separately.

Reproduced above, from UK Polling Report, are all the published surveys this year from ICM – the firm which has been polling for longest using broadly the same methodology.

Looking at the table and you see Labour’s share dropping down to 25% in June but then recovering so that in each of the two August polls it was on 29%. It then moved up to 32% in a survey immediately after the Labour conference but has since slipped back to 30% where it has stood for the past three polls.

There’s been no sign of any “bail-out” bounce there – just a substantially better Labour position than June, a notch up on August but worse than September.

The Tories, meanwhile, seem to have settled into a position in the low to mid 40s and have barely moved out of that range. The biggest variation with ICM seems to be the Lib Dems.

For an indication of the consistency of ICM over the past twelve years check out out the firm’s performances from 16 months ahead of the 1997, 2001 and 2005 general elections. It’s impressive particularly because ahead of 1997 and 2005 elections its Tory shares this far out proved to be a very good guide to what would happen.

So the weekend analysis by the prominent psephologist, Professor John Curtice “Exploding the myth of Brown’s bounce” seems to have foundation.

Maybe tonight’s November Populus poll for the Times, taken after the Glenrothes result, will show something different – but Labour needs to see the Tory percentage well below the 40 mark before it can start cheering.

Mike Smithson