Unbelievable correlations?
July 15, 2016 11:57 PM   Subscribe

I'm looking for things that have been proven to be correlated directly but which aren't expected. So something I wouldn't want: people who have sex more are happier overall. What I would want: people who have sex more are more likely to choose grape soda over orange when offered both ( I wish I didn't have to say I made that up :)) anything you think applies to this, bring it on.
posted by jitterbug perfume to Society & Culture (21 answers total) 12 users marked this as a favorite
You've seen the spurious correlations site right? That's a classic.
posted by FireFountain at 12:03 AM on July 16, 2016 [19 favorites]

There's a debunked data mining story relating sales of beer and diapers
posted by willem at 3:28 AM on July 16, 2016

This is the entire schtick of correlated.org
posted by jaymzjulian at 4:22 AM on July 16, 2016 [2 favorites]

I think OP is looking for valid but surprising correlations, not spurious and debunked ones.
posted by workerant at 4:46 AM on July 16, 2016 [8 favorites]

Relevant XKCD
posted by b33j at 4:56 AM on July 16, 2016 [2 favorites]

I don't have the reference for this but the example my econometrics professor always used was ice cream consumption and murder rate. They do track together, but only because they both track with hot weather.
posted by telegraph at 5:26 AM on July 16, 2016 [6 favorites]

OKCupid had a whole blog dedicated to the trends they found in their data. It hasn't been updated for two years now, but if you go back through the entries, you'll find lots of fascinating correlations. Some of them are not surprising but many of them are. For example, non-religious people are more likely to be offended by spelling mistakes.
posted by yankeefog at 5:33 AM on July 16, 2016 [13 favorites]

People with blue eyes are apparently significantly more likely to develop alcohol addiction.

I went looking for this because there's a point of view that I think leads to the conclusion that people with blue eyes should be more likely to develop substance addictions of all kinds.
posted by jamjam at 9:48 AM on July 16, 2016 [2 favorites]

Response by poster: Workerant is right - I'm not looking for random things that so happen to correlate in a bizarre, unexplainable manner as much as I am for things that have some sort of reason behind the correlation - one which wouldn't be obvious. Like the example someone posted of ice cream sales and murder on hot days. Additionally, I *am* looking for things even more random, as long as they correlate somehow for a reason, the more bizarre and I-never-would-have-thought-of-that as possible! Thanks so far and please keep them coming!
posted by jitterbug perfume at 10:47 AM on July 16, 2016 [1 favorite]

Length of marriage and baldness or grey hair, because if you have been married 50 years, you aren't 22 anymore.

That was from a stats class. I vaguely recall another about race and milk consumption, iirc. I guessed that it was because you see more lactose intolerance in some groups but the real answer is that some groups are too poor to afford more milk.
posted by Michele in California at 10:56 AM on July 16, 2016

Being tall is associated with higher risk of cancer, all else being equal. (All else is often *not* equal, so it's *not* true in general that short people have lower rates of cancer than tall people. But if people have similar socioeconomic status and health habits, the tall person is more likely to get cancer than the short person.)

The proposed reason is that being tall requires more cell divisions than being short, and each additional cell division increases the risk of causing or uncovering a mutation that causes uncontrolled cell division.
posted by clawsoon at 11:18 AM on July 16, 2016 [5 favorites]

Another cancer correlation: Age of first childbirth and risk of breast cancer are strongly correlated. A woman who has her first child at 15 years old has about half the breast cancer risk of a woman who has her first child at 30.
posted by clawsoon at 11:26 AM on July 16, 2016 [2 favorites]

That's very interesting, clawsoon -- and you remind me that I've read that short men are more likely to develop heart disease, for reasons that aren't immediately clear.
posted by jamjam at 11:26 AM on July 16, 2016 [1 favorite]

And women with Down syndrome are about one twentieth as likely to develop breast cancer as age matched controls.
posted by jamjam at 11:30 AM on July 16, 2016 [1 favorite]

in early elementary school years, astrological sign is correlated with IQ, but this correlation weakens with age and disappears by adulthood. source (explanation left as an exercise to the reader)

(wwii) bombers were more accurate when attacked by enemy fighters. source (fighters were only effective when cloud cover was low).

[both from searching for amusing examples of confounding variables]
posted by andrewcooke at 11:34 AM on July 16, 2016 [4 favorites]

in early elementary school years, astrological sign is correlated with IQ, but this correlation weakens with age and disappears by adulthood. source (explanation left as an exercise to the reader)

This reader thinks that the key to that explanation might rest on knowing whether the AS/IQ correlations reverse themselves in the Southern Hemisphere.
posted by jamjam at 11:42 AM on July 16, 2016

There is also a really high and unexplained incidence of cancer in some places. My recollection is that the county just north of San Francisco has anomalously high breast cancer rates that no one could figure out the cause of, at least at the time that I lived in the San Francisco Bay Area.

This may be "too obvious" for what you are looking for: There is also a correlation between height and rank in the military, and probably elsewhere. Generals tend to all be over 6' tall. Even female generals tend to be at least 5' 9" tall or taller. Tall people are perceived as more authoritative and it makes it easier for them to be promoted to leadership roles because it just fits with the general social expectation. At least, that's the theory.

Weight for white woman is inversely correlated to salary. Heavier white women make less money. This did not hold true for other races in the U.S., IIRC.

I also recall hearing that beauty and IQ are correlated in women: More beautiful women tend to have higher IQs. I don't know if there is a theory for this one. My theory is a) "beauty" is shorthand for "appealing traits" and it is unsurprising that intelligence is an appealing trait, so not weird that smart women would have whatever physical traits others view as "beautiful" and b) "beauty" is at least partly a culturally derived value and brighter people can figure out how to dress, do makeup or whatever in a manner that is more culturally acceptable. So competence at reading the expectations of other people and replicating the desired result would be about being smart and would translate to appearing more physically attractive.
posted by Michele in California at 12:08 PM on July 16, 2016

How to Lie with Statistics has a couple of (old, from the '50s) examples in Chapter 8, my favourite being the close correlation between the salaries of Presbyterian ministers in Massachusetts and the price of rum in Havana. Does that mean that the ministers were buying rum or selling it?
posted by clawsoon at 12:28 PM on July 16, 2016

Freakonimcs is full of these sorts of things. See especially the section on parenting.
posted by jojobobo at 1:49 AM on July 17, 2016 [2 favorites]

« Older What's your favorite visual novel?   |   Equivalent of Magic (concierge service) in the UK? Newer »
This thread is closed to new comments.