media & PR

Single-sourced national polling stories should be as suspect as any other single-sourced stories

Sometimes a single sourced exclusive story is so important, and the truth so hard to dig out, that it’s a justified journalistic gamble. Only sometimes, however, which is why the norm is to work hard to find other sources, and certainly to wheel in corroborating evidence.

Yet when it comes to reporting changes of support in national voting intention polls, especially when it’s a media outlet reporting on polling it has paid for itself, the more outlandish the poll results are compared to others, the bigger the splash it gets. All those caveats about random fluctuations, it only being one poll get buried in the 48th paragraph (if you’re lucky) as the banner headlines are rolled out.

The reality is most national voting polls are boring. It’s rare to have a poll that shows a movement which is clearly outside the normal margins of error – and of course as the random statistical noise means some polls will by bad luck be out by even more (roughly 1 in 20 on the usual calculations), then even the poll showing an apparently statistically significant move may well just be a rogue.

That means you can almost never say of a single national poll ‘party X has surged’ or ‘party Y has slumped’ (constituency polls are different as they more regularly show big movements, especially during by-elections). The boring reality is that what you should say is ‘something may have happened but we don’t know yet and will have to wait and see’. That doesn’t make such a good headline, however.

The risks of this was neatly illustrated at the weekend. The Sunday Times gave over a huge front page splash (followed by more stories on p.2) to the news that Labour had moved into a four point lead over the Tories following Thursday night’s TV ‘debate’. And then a few hours later out came another poll, this time showing that the very same Tories who were four points behind Labour were also four points ahead of Labour.

This exercise is Escher polling is not required due to a paucity of opinion polls. In fact, we’re on course to hit 2,000 published national voting intention polls this Parliament.

It’s due to the media’s rush to pontification on one poll as if it’s the equivalent of a multi-sourced, solid story. It’s not. It’s the equivalent of a questionable single-sourced story and should be treated as such.