2 ms·
You interjected “random” and that seems odd. Correlation doesn’t mean causation but that also doesn’t mean correlation is random noise. It just means there’s a
by eyelidlessness 5y ago
You interjected “random” and that seems odd.
Correlation doesn’t mean causation but that also doesn’t mean correlation is random noise. It just means there’s an unknown relationship between facts.
- XorNot 5y agoThe point is that a random data set will spontaneously produce clustered groups of apparent events that are not due to any underlying cause.
- eyelidlessness 5y agoAnd a non-random data set will produce clusters of correlations which don’t have causal explanations. I’m well aware of this axiom which is why I found it important to address.
- ChrisLomont 5y agoAnd if there were such clusters beyond pure random noise, they'd show up in studies. They don't, hence they're random noise. How many heads in a row would you expect from an unbiased coin if you flip it long enough? That is your "clusters of correlation."
- ChrisLomont 5y agoCorrelation most certainly does not imply an unknown relation between facts. In fact, there are more completely unrelated correlations than there are correlations caused by related things. This is simple statistics. Proof? Consider M things related that show a correlation. Suppose some other thing randomly by chance correlates, and that thing has N items correlated by some other set of related facts. Then all the not causally related cross correlations between the M and N items vastly outnumber the causally ones. So no, correlation absolutely does not imply an unknown relation. It points to a place to look.