Tuesday 9 November 2010

Correlation does not imply causation

It's statistics 101, the most basic thing anyone who looks at numbers for a living needs to remember - just because two variables have a correlation, it doesn't mean that variable A is causing variable B to happen. The number of men wearing hats in Britain has declined since the 1920s, has as overall air pollution in cities. But this does not mean that wearing hats causes pollution. There are other factors in play that need to be taken account of, such as clean air legislation.

It's a very simple concept but one that journalists are forever struggling with, as Ben Goldacre and Atomic Spin never tire of pointing out. The Mail (who else?) have just come up with yet another example:
Teenagers who text up 120 times a day 'are more likely to have had sex, use drugs and drink alcohol'

Teenagers who text 120 times a day or more are more likely to have had sex or used alcohol and drugs than kids who don't send as many messages, according to provocative new research.

The study's authors say they have seen an apparent link between excessive messaging and this kind of risky behaviour.

First things first, the headline is supremely sloppy - it talks about kids who text "up to" 120 times a day (which would include everyone from those who send one SMS a day to those who send 120), when in fact, as the story makes clear, the study talks about kids who text MORE THAN 120 times a day. It's only a minor point, but it just shows that the writers are being rather lazy and doesn't fill you with confidence.

The opening two paragraphs exhibit a clear suggestion that researchers have found a causal link between sending a bonkers number of texts and "risky behaviour". It fits nicely with the Mail's general narrative about no-good young people (see also happy slapping and Facebook). Unfortunately the researchers have found nothing of the sort, as the Mail itself points out in the very next paragraph:
While researchers say they aren't suggesting that 'hyper-texting' leads to sex, drinking or drugs, the study concludes that a significant number of teens are very susceptible to peer pressure and also have permissive or absent parents, said Dr Scott Frank, the study's lead author.
Right. So everything you said in the previous paragraph about researchers finding "an apparent link" between the two factors was actually a great big fib? Do go on, anonymous Daily Mail Reporter:
'If parents are monitoring their kids' texting and social networking, they're probably monitoring other activities as well,' said Dr Frank, an associate professor of epidemiology and biostatistics at Case Western Reserve University School of Medicine.
And that's the final nail in the story's coffin. The causal factor is "permissive or absent parents". Two symptoms of permissive or absent parents are their offspring indulging in risky behaviour AND sending lots of texts becaus their parents aren't sat next to them saying "stop sending text messages every eight minutes, and while you're at it stop having sex". Despite what the headline implies and what the second paragraph states as fact, there is NO link between sending texts and having sex.

Perhaps I should be celebrating because the Mail has actually included all the facts for a change. But instead I'm feeling unreasonably annoyed because, yet again, the paper has twisted the facts to fit its own agenda and stuck in a headline and intro that is totally inaccurate.

Oh well. At least they didn't say that texting gives you cancer.

1 comment:

Jon said...

Another great post. It's good to see analysis of tabloid duplicity within a single article, as well as across a number of pieces.