[superice]

I get it, you like stories of individuals. We should figure out a way you can listen to more stories, so you can form an even better opinion! Perhaps we write them out, shorten them a bit. Or perhaps group them by similarity. And then if we count the types of story per category… and boom, we’ve invented statistics!

[coldtea]

And ...boom! we lost the nuance, and shoehorned together disparate elements into a bunch of measurements, as if those explain everything (as opposed to needing explanations and a working theory themselves, and often fitting multiple theories about how they came about).

Not to mention the cases where the numbers are collected or analyzed in bogus ways (from wrong methodologies and false reporting, to p-hacking), and people are asked to cargo cult respect them anyway...

[mensetmanusman]

I know how to use statistics to tell nearly any story though, and I would love to trust statistics…

It has to do with unknown unknowns.

[superice]

And I know how to tell any story, it's called lying. Whether you are lying with statistics or lying in a story does not matter. In the end it all comes down to whether you can validate what you've been told. Most people however will skip validation in favor of 'this sounds reasonable' and most people have a worse intuition for statistics than for stories. That's on them though, let's not blame statistics for that shortcoming.

[coldtea]

I'd prefer a lie with a story over a lie with statistics.

At least the former doesn't pretend to be scientific or objective, it's just a story.

[mensetmanusman]

The point is that if you have over 100 correlates to assess a situation, any particular story you try to tell is probably a lie even with the best intentions.