What do you think of the “2nd best 2nd worst rule”?

With so many polls showing such different pictures I’ve now adopted my own “rule of thumb” for working out what the big picture is. Basically I take the latest surveys from YouGov, Mori, ComRes, ICM and Populus and take the second lowest share for Labour and the second highest shares for the Tories and Lib Dems.

Where two or more pollsters are showing the same high figure I take that and do the same with the low ones.

The reason for this is that the pollsters have a long-standing tendency to understate the main opposition parties and to overstate Labour. Even though the formula compensates for this it would have over-stated Labour in the past four general elections but not by as much as most individual polls.

One of the issues is that many respondents say Labour when they actually mean “Anti-Tory” – an element that affects ratings for both parties. Also with nearly 10% of all seats in Lib Dem hands there are going to be even more tactical considerations next time, whenever that is.

In 2005 the approach would have produced CON 33% (correct): LAB 37%(+1): LD 23% (correct).

Taking the second highest and second lowest cuts out the odd rogue or old survey.

Based on the latest polls from the five pollsters the current “Second best – second worst” calculation produces CON 34%: LAB 39%: LD 18% – which seems about right.

Mike Smithson