The Sun says: “Every August poll carried out before a spring election since 1996 has predicted the result to within one per cent.”
Reality says: MORI 24 August 2001 2000 – Con 29% (out by 4%), Lab 51% (out by 9%), lead 22% (out by 9%)
Although this is the most striking example, overall for example eight out of the 12 August polls since 1996 had the Labour lead out by more than 1%. Similarly, only three got the Labour share to within 1%.
Not exactly a case of “every”.
UPDATE: Himmelgarten Cafe points out that the graphic accompanying the Sun story isn’t exactly a piece of accurate representation.



6 Comments
“The Sun says: ”Every August poll carried out before a spring election since 1996 has predicted the result to within one per cent.”
Reality says: MORI 24 August 2001 …”
Eh? Surely it means an election the following spring?
I suspect this is based on the Political Betting analysis of ICM polls in the August ahead of an election (see http://politicalbetting.com/index.php/archives/2009/08/26/could-cameron-win-a-victory-of-1997-proportions/). Looks like The Sun are being a bit slack with their facts (not for the first time).
Presumably they only mean general elections, compared with the poll in the previous August. Which means a sample of precisely three elections. In other words, (surprise, surprise) utterly meaningless.
HB – just got a panicked call from Mark asking me to update the post as he erroneously put 2001 where he meant 2000…
Sun – facts. Facts – Sun. Perhaps you might like to talk to one another one day?
You can’t seriously have expected this rag to be accurate and unbiased?! Anything that will boost “Dave” ‘s chances between now and the GE is all they’re interested in!