Does opinion really change by all that much?
Do we make too much of each minute swing?
Last week I had an enjoyable session with the YouGov boss and writer, Peter Kellner, when we talked about a new development his company is planning and, of course, the way the polls are going.
He made a point that seemed to ring very true – that people’s political allegiances do not change by anything like the magnitudes that are reflected in the polls. Things really are much more stable than they appear.
Yes we have events that seem to produce step changes but outside that there is a lot less movement than most people perceive. So just over a year ago we saw a step change with the new Tory leader and then there was another one in May when a series of calamities for Labour also seemed to come together at the same time. Aside from that not very much.
A lot of the problem is the media’s desire for a story. Polls that don’t change that much month by month don’t quite create the headlines that sensational moves do.
We have been conditioned a bit by the polls of yesteryear that did produce dramatic numbers and changes. Polling methodologies have moved on and the mechanisms designed to avoid sample bias seem to be working.
YouGov, of course, are in a strong position to be able to monitor how the views of different people change. Their polling is restricted to their polling panel on whom they have a vast amount of back data. These are YouGov’s last seven polls. The only major variation was at the end of September in the poll taken immediately after Tony Blair’s Manchester conference speech.
Date: CON LAB LD
22/12/06 37 32 15 +5
20/12/06 37 33 17 +4
30/11/06 37 32 16 +5
26/10/06 39 32 16 +7
29/09/06 36 36 16 0
22/09/06 37 33 18 +4
22/09/06 38 31 18 +7
So let’s see what the first poll of 2007 produces – that’s from Populus and is due in the Times tomorrow. A month ago the pollster reported the Tories down to 34% with Labour on 33%.