Why you cannot make comparisons with the run up to 1997

The above is reproduced from Anthony Wells’s excellent UK Polling Report site and shows the polls in the final month before the 1997 general election which Labour won with a margin of 13%.

Just compare the above figures with the final shares of CON 31.4%: LAB 44.4%: LD 17.2%

I refer back to this because time and time again you will read or hear commentators and politicians say that Labour was doing much much better in the run-up to 1997. Yes they were but polling was different in those days and the Labour share was repeatedly over-stated by what turned out to be very big margins by all but one of the firms.

The exception, as can be seen from the above, was ICM which then and now had the system of past vote weighting to ensure politically balanced samples.

Since 1997 polling has changed dramatically. Firms have come and gone and now four out of the five regular pollsters carrying out monthly surveys have followed the ICM lead and use some system to make sure their samples are politically representative. The other one, Ipsos-MORI, now reports headline figures based on those “certain to vote”. Then it didn’t.

Blair in 1997 went on to win a massive landslide with 44% of the vote. He was helped by a lot of tactical voting driven by a big desire to get the Tories out.

I don’t think pre 1997 conditions exist in the UK at the moment and my best guess is still that the Tories will be short of a majority. But comparing polling between now and then is totally misleading.

Mike Smithson