I don’t think you should be editorializing the content which gets posted. Instead come up with a clear definition/ test of what a “crappy correlation” is, it flies. People are going to post whatever they are going to post and trying to get your own perspective across (even if that perspective is “I want a break from that stuff”) is a political bias. Its bad moderation, and a political perspective in its own right. You don’t control what other people post is kind-of the point, and asking people to match your editorial bent seems inappropriate.
More importantly, whats a good definition for a high quality “crappy correlation” is?
It would seem to me that there may be tiers. I’ve seen some which are just line go up, line go down, not based on any real data. Those are pretty low quality imo. E-C tier.
Second there are those maybe based on real data, but the correlation is spurious, or attributed, but might actually have causal linkage. B tier.
Then there are those based on real data, and the correlation is ridiculous, and not possibly causal. However the correlation might be a low fit (R2 of less than .75). Those are A tier.
Finally there are those based on real data, and the correlation is ridiculous, and not possibly causal. However the correlation is a tight fit (R2 of greater than .75). Those are S tier.
a good definition for what a crappy correlation is is just the spurious correlations from the spurious correlations website that’s in the main link , and the whole point of them is that they’re funny. And by crappy we mean something that is mathematically correlated that it makes absolutely no sense. That’s why it’s funny, and that’s what makes it funny. If you do not make the moderator laugh, you will be executed. There’s no probation, we just go straight to execution. Oh I’m sorry, I meant crucifixion. Let’s do crucifixion.
I don’t think you should be editorializing the content which gets posted. Instead come up with a clear definition/ test of what a “crappy correlation” is, it flies. People are going to post whatever they are going to post and trying to get your own perspective across (even if that perspective is “I want a break from that stuff”) is a political bias. Its bad moderation, and a political perspective in its own right. You don’t control what other people post is kind-of the point, and asking people to match your editorial bent seems inappropriate.
More importantly, whats a good definition for a high quality “crappy correlation” is?
It would seem to me that there may be tiers. I’ve seen some which are just line go up, line go down, not based on any real data. Those are pretty low quality imo. E-C tier.
Second there are those maybe based on real data, but the correlation is spurious, or attributed, but might actually have causal linkage. B tier.
Then there are those based on real data, and the correlation is ridiculous, and not possibly causal. However the correlation might be a low fit (R2 of less than .75). Those are A tier.
Finally there are those based on real data, and the correlation is ridiculous, and not possibly causal. However the correlation is a tight fit (R2 of greater than .75). Those are S tier.
a good definition for what a crappy correlation is is just the spurious correlations from the spurious correlations website that’s in the main link , and the whole point of them is that they’re funny. And by crappy we mean something that is mathematically correlated that it makes absolutely no sense. That’s why it’s funny, and that’s what makes it funny. If you do not make the moderator laugh, you will be executed. There’s no probation, we just go straight to execution. Oh I’m sorry, I meant crucifixion. Let’s do crucifixion.