Join 3,521 readers in helping fund MetaFilter (Hide)


Divorce rate in Mississippi correlates with Murders by bodily force
May 9, 2014 7:32 PM   Subscribe

Spurious Correlations lets you dig through various data sources to find things that totally aren't causally related... or are they? [Related, previously]

Industrial manufacturing doctorates awarded (US) correlates with Per capita consumption of motzarella cheese (US).
Avg daily precipitation in Oregon correlates with Number of people killed by dogs
Passengers killed in collision with stationary object correlates with Per capita consumption of beef (US)
posted by moonmilk (26 comments total) 50 users marked this as a favorite

 
This one is more than a little creepy.
posted by Lazlo Hollyfeld at 7:42 PM on May 9 [5 favorites]


Oh dear lord, what fresh hell have you sent me to? I may never escape this rabbit hole! The Discover New Correlations thingy? Oh...I'm doomed. Curse you, moonmilk the square cow!
posted by dejah420 at 7:48 PM on May 9


The correlation between sunlight in South Carolina and golf revenues (nationwide?) shouldn't be too spurious: presumably a trend of good golfing weather in one part of the country would suggest a trend towards similar weather in the rest of the country, unless you had reason to believe that south Carolina's weather was regularly radically different from the rest of the country.

Likewise "people falling to their deaths vs. people falling from ladders" or whatever is actually a pretty straightforward correlation.
posted by Avenger at 7:51 PM on May 9 [1 favorite]


You can't tell me that the number of people electrocuted by powerlines dropping along with marriages in Alabama is mere correlation.
posted by Lutoslawski at 7:53 PM on May 9 [2 favorites]


The scary part is when you realize that while correlations do not equal causation, they do, in fact, equal correlation -- and that there may be some arcane connection between the price of tea in china and the homocide rate in Mobile, Alabama.

Statistics is black magic.
posted by Avenger at 7:54 PM on May 9 [2 favorites]


Sorry, you can't convince me there's anything spurious about the inverse correlation between money spent on pets and number of deaths by limb amputation.
posted by blue suede stockings at 8:18 PM on May 9 [3 favorites]


Import price of Uranium (US) correlates with Total revenue generated by arcades (US)

Shall we play a game? Wouldn't you prefer a nice game of chess?
posted by justsomebodythatyouusedtoknow at 8:22 PM on May 9 [3 favorites]


It's fun to look up things and pick the first result. Sometimes it makes sense (juvenile drug arrests correlates with juvenile arrests for marijuana? big surprise... but some... like the greater than 99% correlation between US spending on science, space and technology and deaths from hanging, strangulation and suicide... interesting to think about confounding variables.
posted by gryftir at 8:24 PM on May 9


Education saves lives!
posted by wildcrdj at 8:28 PM on May 9 [1 favorite]


Significant
posted by TedW at 8:34 PM on May 9 [3 favorites]


Am I doing it right?
posted by agentofselection at 8:40 PM on May 9 [5 favorites]


Engineers can't be trusted to walk down the hall.
posted by stevis23 at 9:29 PM on May 9


They misspelled mozzarella! Can we really trust a website that misspells mozzarella?
posted by flapjax at midnite at 9:42 PM on May 9 [3 favorites]


We can now safely guess that Niclas Cage is not to be blamed for dozens of fatal helicopter accidents every year.

Probably just a few. If I'm thinking of the same Niclas Cage.
posted by not_on_display at 10:46 PM on May 9 [3 favorites]


I miss Google trends correlation search by drawing. It no longer works (for me, anyway, in Chrome), but I wanted to draw spikes 6 months before major earthquakes and see what search terms popped up.
posted by jjwiseman at 12:10 AM on May 10


Mozzarella with a T. Nicolas Cage w/o the O? r2 is pretty close to 1 this guy doesn't have spellcheck.
posted by birdherder at 1:33 AM on May 10 [1 favorite]


This reminds me of the old Amarillo Slim line about how his hometown's population never changed, because every time a baby was born, a man left town.
posted by TheSecretDecoderRing at 2:33 AM on May 10 [1 favorite]


The slightly less violent version is also known as: Marketeers Exploring Google Trends.
posted by Nanukthedog at 4:01 AM on May 10


Tune in this summer for the X,Y files ... two FBI agents who investigate phenomena that may or may not be causally related!

"Mully, do you think these accidental poisonings by organic solvents and halogenated hydrocarbons and their vapours have anything to do with the number of lawyers in American Samoa?"

"Yes, Sculler, I think they're definitely correlated."

"But we need more data."

"No, don't you get it? The less data points we have, the easier it is to correlate them!"

"But that's meaningless!"

"Is it meaningless? Or just an unproven hypothesis?"

"You know what, next time you go get kidnapped by government agents at a UFO site you can just go rescue your damn self."
posted by RobotVoodooPower at 6:12 AM on May 10 [6 favorites]


The real story hidden here is that Vermont may be stealing spouses from South Carolina, possibly though some introduction to the state via needing maple syrup as a replacement for honey as a sweetener for tea.

I would like to hear jessamyn explain this obvious situation.
posted by GenjiandProust at 7:33 AM on May 10 [3 favorites]


The scary part is when you realize that while correlations do not equal causation, they do, in fact, equal correlation -- and that there may be some arcane connection between the price of tea in china and the homocide rate in Mobile, Alabama.

The key is to understand the possible ways they can be connected. Two big points that people usually get wrong when confronted with correlations like this:

From the wikipedia page on correlation not equaling causation = "As with any logical fallacy, identifying that the reasoning behind an argument is flawed does not imply that the resulting conclusion is false."

Spurious correlations. In statistics, a spurious relationship (or, sometimes, spurious correlation) is a mathematical relationship in which two events or variables have no direct causal connection, yet it may be wrongly inferred that they do, due to either coincidence or the presence of a certain third, unseen factor (referred to as a "confounding factor" or "lurking variable").
posted by srboisvert at 8:26 AM on May 10 [1 favorite]


The number of MeFi posts per day is inversely correlated with US GDP growth.
posted by humanfont at 10:56 AM on May 10 [1 favorite]


The all-time classic was years ago on the Flying Spaghetti Monster page: Number of pirates vs global warming.

http://www.venganza.org/about/open-letter/
posted by CrowGoat at 12:36 PM on May 10 [1 favorite]


MetaFilter: what fresh hell have you sent me to?
posted by homunculus at 5:43 PM on May 10 [1 favorite]


You ought to copyright "the X,Y files" RobotVoodooPower; that could totally be the new Carmen Sandiego.
posted by jamjam at 6:19 PM on May 10 [1 favorite]


Now we just need a companion page to automatically generate leading questions.
posted by ckape at 1:30 AM on May 11


« Older Introducing a perfect Metafilter storm: Ponyroguel...  |  The internet is flooded with h... Newer »


This thread has been archived and is closed to new comments