Monday 5 October 2009

Small Sample Polls And The News

I'm reading The Metro, a generally interesting free paper available on
a variety of public transport in the UK. I've got to page 4 and
already the paper has featured news stories based on polls, one of
which was the front page headline. Sample size? 1100 (for both which
suggests it was in fact the same poll).

How statistically significant or representative are the opinions of
1100 people out of a population of some 60 million? Not very. News
worthy? Fuck no.

What really pisses me off is that the writer and/or editor is passing
off the results of the poll as if it were representative of Britain as
a whole. 1100 opinions wouldn't even be representative of Glasgow.

These piddling pathetic polls shouldn't be used as news, certainly not
headline news. If you ARE going to pass comment on them then you
should stress that the small sample size is not large enough to
accurately represent a whole nation rather than mention it in passing
at the foot of the article.

It's bad reporting, or it's manipulative reporting designed to
influence opinion, which is worse than merely 'bad'.

It'll be interesting to see if other news outlets report on this and
how much significance they place on its findings.

What I would like to see (if it doesn't exist already) is a national
poll service one can subscribe to for both polling and viewing the
results, that would be much more interesting I tuink. I think I might
investigate that later.

freedoms_stain, losing respect for The Metro, out.

--
Sent from my mobile device

No comments: