Media & PR

Correlation implies causation rather more often than you might think

Chances are, you’re very familiar with the idea that correlation does not mean causation, as nicely illustrated by this cat:

It’s a point I’ve made a good few times on this blog too, such as in relation to web browsers and murder rates, not to mention mountain ranges and murder rates.

But as I point out in my new book, Bad News: what the headlines don’t tell us:

This wise advice quickly ossifies into misplaced repetition, too often deployed in situations where it is not applicable.

Consider salt (or pepper, if you prefer). When you season your food, it makes it taste saltier, right? I will go out on a limb and guess every reader agrees.

But why do you believe that? After all, the combination of putting salt on food and it being saltier is just correlation, and you know that correlation does not imply causation. So, you shouldn’t really believe that putting salt on your food makes it saltier (and therefore taste more delicious to you), should you?

Unless you’ve got a friend to blindfold you, split your food in two, put salt on one and not the other, and then see how you perform in a double-blind randomised trial – or unless you’ve hunted out the research of someone else who has done this – you are just relying on correlation implying causation.

And that is just the sort of daft error you are smart enough to avoid making, right?

Find out all about when correlation does imply causation, including the pioneering work of Austin Bradford Hill, who saved millions of lives, in Bad News: what the headlines don’t tell us, available from Waterstones, Hive or Amazon.

3 responses to “Correlation implies causation rather more often than you might think”

  1. It’s sensible to start off with the premise that correlation doesn’t imply causation but might do. You can then design work to establish whether there is a causal relationship, as happened in the case of smoking.

  2. There is a correlation between the installation of central heating in British bedrooms and the loss of the British Empire. Therefore, any country with an Empire that puts in central heating will lose it’s Empire. QED

  3. The cat video is brilliant. I now realise my mother’s cat, which loved watching TV rugby and snooker, had a profound effect on results.

    “Post hoc, ergo propter hoc” is one of the basic fallacies any trained historian will watch out for and it’s common enough. I’ve just reviewed, for next Liberator, Philippe Blom’s “Nature’s Mutiny” about the Little Ice Age and social change and suggested his argument is largely of this brand. However, “Ante hoc, ergo propter hoc” is obviously wrong except for Doctor Who and when historians look for the causes of something, they will sift through things that happened earlier for possible causes, testing them – for example, “Same thing happened in two other places/ at two other times and this result didn’t happen”. Also, of course, it’s a worry if you can’t understand HOW (a) could have caused (b), but dangerous to assume that because you can very easily see how it might have done, it did.

Leave a Reply

Your email address will not be published. Required fields are marked *

All comments and data you submit with them will be handled in line with the privacy and moderation policies.