Hugh Pickens writes:
Scott Adams of Dilbert fame writes on his blog that science's biggest fail of all time is 'everything about diet and fitness':
I used to think fatty food made you fat. Now it seems the opposite is true. Eating lots of peanuts, avocados, and cheese, for example, probably decreases your appetite and keeps you thin. I used to think vitamins had been thoroughly studied for their health trade-offs. They haven’t. The reason you take one multivitamin pill a day is marketing, not science. I used to think the U.S. food pyramid was good science. In the past it was not, and I assume it is not now. I used to think drinking one glass of alcohol a day is good for health, but now I think that idea is probably just a correlation found in studies.
According to Adams, the direct problem of science is that it has been collectively steering an entire generation toward obesity, diabetes, and coronary problems. But the indirect problem might be worse: It is hard to trust science because it has a credibility issue that it earned. "I think science has earned its lack of credibility with the public. If you kick me in the balls for 20-years, how do you expect me to close my eyes and trust you?"
The Wikipedia article gives the impression that there is always an underlying causality for a correlation, even if it is a causality from a third factor (all examples are explained with a third factor that causes both). However there are correlations which are not even related that way. One example of the "correlation is causation" fallacy was used in a statistics course on German TV: A correlation between birth rates and numbers of storks in a certain state of Germany, "proving" the claim that the stork brings the babies. There's no common cause I can see that explains the correlation (which was not just "the storks go down, and so do the birth rates" but also contained "at that one year, the stork population went up again, and so did the birth rates").
Sometimes correlation is just coincidence.
I think I read your example of the storks & babies in Box, Hunter & Hunter - Statistics for Experimenters. A classic.
The Wikipedia article gives the impression that there is always an underlying causality for a correlation, even if it is a causality from a third factor (all examples are explained with a third factor that causes both). However there are correlations which are not even related that way.
While you are right there are pure coincidental correlations, the Wikipedia article does not say there's always a cause for a correlation; read it in full.
I just linked to a specific section of Wikipedia article because the post I was responding examined the case of "correlations caused by a 3rd factor" and I only said I like better 2 of the examples the Wiki article cited in the section dedicated to correlations caused by a 3rd common causal variable (which is only one of the many article section)
the Wikipedia article does not say there's always a cause for a correlation
Please re-read what I wrote. I didn't claim that the Wikipedia article says it. I said that the Wikipedia article gives the impression.
Of the large section of examples, which make up the bulk of the article, not only does the "common cause" subsection take the largest space, but this is the complete table of contents of the Wikipedia article:
1 Usage2 General pattern3 Examples of illogically inferring causation from correlation
3.1 B causes A (reverse causation)
3.2 A causes B and B causes A (bidirectional causation)
3.3 Third factor C (the common-causal variable) causes both A and B4 Determining causation
4.1 In academia
4.2 Causality construed from counterfactual states
4.3 Causality predicted by an extrapolation of trends5 Use of correlation as scientific evidence6 See also7 References
7.1 Bibliography8 External links
You see, there examples of illogical inferring causation from correlation is only cases where there indeed a causation as root of the correlation, just that the real causation is different from the wrongly inferred one in those examples.
But yes, the article does indeed say that a correlation can be a coincidence. In exactly three sentences of the 27kB article.