I would like some good concrete examples that demonstrate the phrase: Correlation doesn’t prove Causation.
I understand the phrase, I do not dispute it, but right at this moment am having some troubles coming up with some examples.
Are you interested just in cases where A does not cause B nor B A, or in cases where there’s no connection at all?
If the former, my favorite one is ice cream and snakebites. Both geographically and temporally, there’s a very strong correlation between sales of ice cream and the incidence of snakebites. But ice cream does not cause snakebites, nor do snakebites cause consumption of ice cream.
A good example would be where the underlying causes are the same leading to a strong correlation, but the two factors are not causes in itself.
For example one might expect to see a strong correlation between those who wear XXXL clothing and incidents of heart disease, but wearing XXXL clothing does not cause heart disease and neither does having heart disease cause one to wear XXXL clothing, instead they are linked by their common cause (obesity).
Dark skin and sickle cell anemia? People with dark skin are more prone to sickle cell because some of their ancestors came from regions where conditions favored both darker skin and malaria resistance, not because dark skin causes sickle cell or the other way around.
Symptoms of autism are often becoming evident to parents around the age at which vaccines are administered.
There are also variants of this that draw inference from vacuous or universal truths - as seen in The Dread Tomato Addiction - one point of which is: Everyone who has eaten tomatoes may be expected to eventually die - of course, this is just because everybody dies.
Two recent examples I’ve particularly liked are:
[ul]
[li]Chocolate consumption and Nobel prizes per capita[/li]
[li]Autism incidence and organic food consumption[/li][/ul]
The first because of the fact that it was actually published claiming a causal relation via the statistical correlation (and in the New England Journal of Medicine, no less), and the second one because maybe people finally stop listening to those anti-vax nutters, seeing how they’re now obligated to also become anti-organics…
There are dozens of factors that correlate with the results of presidential elections, from the height of woman’s skirts to whether the Redskins win on the previous Sunday.
Those are from this year’s campaigns. If you go back in time you’ll find a gazillion others.
Days of high umbrella sales see higher incidences of traffic accidents.
Ha! I haven’t seen that one in years.
All great examples. Thanks!
I got what I need, so we can drop it if you guys want; but - some of these are really fun, so we can keep 'em coming if you want too.
The late great science writer Stephen Jay Gould used the following example in The Mismeasure of Man:
“…Gould said that the measures of the changes, over time, in ‘my age, the population of México, the price of Swiss cheese, my pet turtle’s weight, and the average distance between galaxies’ have a high, positive correlation — yet that correlation does not indicate that Gould’s age increased because the Mexican population increased.”
One that consternates many, smoking does not cause lung cancer. There is a correlation, smoking is a key factor, but not the cause.
The per capita rate of homicides by firearms in the US is not correlated with lack of gun controls. (Or so will a lot of gun enthusiasts argue).
How about adopting does not cause an infertile couple to get pregnant?
My favorite is a study that showed a strong positive correlation between vocabulary and height (as in, tallness of the individual).
The kicker was that the subjects were all grade-school children.
The classic when I was in college was the study in the 50s (40s?) that showed a correlation between listening to opera and juvenile delinquency. Of course this was when there was a prevalence of Italian gangs.
Here’s another XKCD that has actual data in it that shows that cancer cases have leveled off and started to decline as the number of cell phone users rises.
And just to expand the xkcd-fest. Reading xkcd is (geographically) correlated with reading Martha Stewart living, and consuming furry porn, just because all three are correlated with “places people live”.