Why Labour should ALWAYS assume the worst
Before we get into the days big story about whether Brown can hang on let’s have a quick look at the pollsters which, once again, have shown that the Golden Polling Rule still operates.
This, as we’ve discussed here before, is that in CON-LAB contests the survey showing Labour in the least favourable position is the most accurate one – something that has happened on every occasion when polls have been tested against real results right back to the 1980s and is a key element to remember as we move into the general election campaign.
The final GB EU shares votes were
(excluding the Western Isles) (excluding Scotland): CON 28.6: LAB 15.3: LD 13.9: UKIP 17.4: GRN 8.7: BNP 6.5. The final polls can be found here.
The Scottish numbers will shift Labour a bit but not by very much.
In terms of the CON-LAB battle the winner on the figures we have so far is Populus with its May 28th poll for the Times which had CON 30: LAB 16: LD 12: UKIP 19: GRN 10: BNP 5. Only it and YouGov had Labour as low as 16%.
At the other end of the scale the loser was ComRes which had a CON-LAB split of 24-22 – a lead for Cameron’s party of just two points. This firm, extraordinarily, abandoned past vote weighting for its EU election polls. This is the measure that is used to ensure a politically balanced sample by asking how respondents voted at the last general election and weighting accordingly.
If ever there was a demonstration of the importance of past vote weighting the overnight numbers provide it.
The Golden Rule, incidentally, is one of the reasons why I am opposed to poll averaging.