Is the idea that “governments always recover” a myth?
We are hearing it all the time at the moment from journalists, politicos and pundits who should at least have the basic ability to check some simple facts before publishing wild statements – the idea that governments “always” recover in the polls as we get close to the general election.
One of the problems, as discussed on the previous thread, is that many recall the unreformed polls of the 1990s and compare them with today’s more measured approach. As I’ve argued here before the only pollster that you can make such historical comparisons with is ICM – which is why I’ve reproduced above all its polls for the Guardian in the fifteen months or so before the 2005, 2001 and 1997 elections.
Spend a couple of minutes casting your eye down the table and you’ll find it very difficult to discover evidence to support the swing-back theorists. In the run up to 2001 there was a bit of a blip in September 2000 caused by the petrol crisis and that’s really about it.
What’s surprising about the numbers is the sheer consistency of the Tory shares, whether in government or opposition. There are one or two obvious outliers which you would expect but the overall picture is one of very little variation and the percentages hardly ever deviate from the margin of error.
For Labour the table is not good news for before each election its shares seem to be mostly ahead of what it got when the votes were counted.
There might, indeed, be a Labour swingback before polling day but there’s nothing in modern polling history to suggest that this will happen.
As an addendum it should be stated that on the last two occasions that the Tories came to power, in 1970 and 1979, all the pollsters over-stated Labour.