Rants - Correlation and Causation
"Correlation is not causation." This tip has been passed around since ages ago, but it's still broken on a regular basis. And for good reason.
A while back, I googled around with my friend on the topic of hormones and love for a school assignment. We found that, while people are in love, their adrenaline levels (the excitement hormone) will be high. I interpreted this to mean that being in love is exciting, but my friends said it means we should do something exciting while making out. Who's in the wrong here?
It turns out, no one. If all you have is two trendlines that follow each other way too closely, the causation could be any of these three:
or even this:
That last one can usually be removed via controlled or statistical tests, but you're still left with three possible scenarios!
So how do we distinguish between them? With knowledge! Take the recent 5G tower saga, for example. Most of us agree that microwave emissions don't carry particles or cause significant health effects, so this is a very unlikely scenario:
The consensus says that this is likely what's actually going on:
However, if you don't know that radio signals can't spread viruses, either because we don't know as a species or because for some reason you want to start Science over from scratch, then the first scenario suddenly becomes equally likely! And we should do everything we can to protect ourselves and our children from this terrible, terrible pandemic, shouldn't we?
So, two trendlines following each other by itself doesn't mean one causes the other. However, it is still a finger pointing you in the right direction. There's gonna be some causation somewhere, but maybe at another level. So don't dismiss them entirely, just give them a few pinches of salt.