Confounding Confounders!

“Let me never be confounded”

Ben Goldacre does a great job explaining why you can’t always take a headline about a research study at face value. Readers need to understand the concept of something called “confounding variables” – not always so obvious factors that can explain away associations that looks causal but in fact are nothing more than coincidence.

Why does he care? Because the media still keeps reporting unadjusted findings in headlines, only to use the last sentence in their article to say “Never mind – there’s a whole ‘nother reason why what we said in the headline really isn’t true”.

Goldacre uses the theoretical example of how drinking is linked to higher rates of lung cancer – until one takes into account the fact that drinkers tend to smoke more. Then – Voila! The association between drinking and cancer disappears and the real troublemaker – nicotine – is unearthed.

Required reading for science writers and lay readers alike.

And that video up there?

It just may probably be the best choral song ever written, and the most fun I have ever sung – In Thee Have I Trusted, from Handel’s Dettingen Te Deum. (If you don’t have time to listen to it all, start around 2:40 – the end is the best part…)

It has just two lines –

Oh Lord in Thee I trusted
Let me never be confounded.

“Let me never be counfounded.” Best song line ever. It is my mantra.

Leave a Reply