Hacker News Reader

Comment by Zikes | Hacker News Reader

Isn't it a fallacy to say that the correlation between dropouts and decreased success means there's a definitive link between them specifically? It's easy to imagine that people going through a very rough period wound up dropping out as a result, but that the dropping out is not what put them into a bad spot in the first place.

View on HN

I would say a lot depends on reason why you dropped out of colleague. Even Zuckerberg was very reluctant to do that, but Facebook growth took too much of his attention. That much different than lack of self discipline/just getting bored.

While correlation isn't causation, it is certainly evidence for causation.

Plus, in many cases, careful analysis alone can reveal the presence (or absence) of causation with great certainty.

Correlation is not evidence for causation.

Why do you say that? It seems clear to me it is. Do you disagree that A being caused by B usually means A and B correlated[0]? That's all that's necessary to say correlation is evidence for causation.

0: I know there are some exceptions, but they seem rare. http://stats.stackexchange.com/questions/26300/does-causatio...

evidence != proof

He is simply stating that correlation can indicate a need to look deeper into the data, which is IMO true.

Yes it is. If A and B are correlated, and C and D are not correlated, then absent other information, it's much more likely that A and B are causally related.

I think people get confused because there can be evidence for things that aren't true.

Reply on news.ycombinator.com