Credit where credit’s (mostly) due: Sunday Telegraph and polling

The internet seems to have got a bit excited about the latest ICM poll, something which I try not to do, but given my past complaints about the media and polling it seems only fair to point out that the Sunday Telegraph looks to have given up its brief flirtation with a non-British Polling Council firm (a flirtation I criticised at the time).

Moreover, the paper’s report sensibly compares the poll results with the previous ICM poll, even though that was commissioned for (shock! horror!) a different newspaper. The tendency to airbrush out polls carried out for rivals when reporting your own poll is still a weakness of British newspapers, but fair’s fair – that wasn’t done this time.

Only two cheers out of three though because the report does fall foul of the popular exaggeration of accuracy in claiming that you can turn its figures into a precise numerical majority in the House of Commons based on those levels of support. You can’t. Swing calculations (uniform, proportional, adjusted, unadjusted, computer, envelope, whatever) aren’t that precise.

It hinders, rather than helps, understanding of polls when results or implications are reported to spurious levels of accuracy.