When someone mentions correlation, they imply a linear relation between the correlated variables and a range of values for which that relation is valid.
Only when that model is further validated correlation suggests causation.
Only when that model is further validated correlation suggests causation.
Yeah that's what correlation means?
> Only when that model is further validated correlation suggests causation.
No! It suggests causation already. There may not be causation but it definitely suggests it.
For example if I find that eating fast food correlates with BMI then there's a pretty clear suggestion that eating fast food causes increased BMI. It might not be the case! But it's clearly very suggestive.
1. There is little correlation, but there is actual causation (e.g: a cricle)
2. There is correlation, but it turns out to be a fluke later during validation.
So, yes. It suggests but in both cases I'd rather continue checkcing