Cartoon explains the difference between causality and covariation

Espen sez, "I thought you would appreciate this cartoon that explains the difference between covariation and causality. In English, the caption is 'During a convivial gathering there is talk of the unhygienic aspect of using galoshes. One of those present chips in: "Yes, I've also noticed this. Every time I've woken up with my galoshes on, I've had a headache."'" Link (Thanks, appliedabstractions)

[#1 posted by mrfitz , March 29, 2008 11:52 AM]

covariation=correlation?

you decide

[#2 posted by Doug Nelson , March 29, 2008 12:53 PM]

covariation>correlation

I have decided.

[#3 posted by ianm , March 29, 2008 1:51 PM]

Correlation != Covariation

Correlation: every time the bus drives by, the street lights go on. The bus happens to drive by at 7pm, the same time the lights go on. No agency in either the lights or the bus, no causal relation between the lights and the bus.

Covariation: def: correlated variation (dictionary.com) - seems to imply agency - inferring causal connections as best as possible without being able to readily identify the underlying causal factor due to the messiness of 'privileged access' we each experience as subjective consciousness.

Changingminds.org gives an interesting introduction:

Description

When explaining other people’s behaviors, we look for similarities (covariation) across a range of situations to help us narrow down specific attributions...

Example

If a manager yells at a person, we assume it is his nature if he is the only person to yell at that person (low consensus), he yells at other people too (low distinctiveness) and he yells at them often. However, if everyone else gets cross with the same person (high consensus) and the manager does not yell at other people (high distinctiveness), we assume it is something external—probably the person being yelled at. Finally, if the manager has not yelled at the person before, we assume that something unusual has happened (situational attribution).
So what?

Using it

Use this to help understand how others are thinking.

[#4 posted by js7a , March 29, 2008 1:54 PM]

covariation = correlation times the product of the populations' standard deviations

[#5 posted by Nick , March 29, 2008 2:04 PM]

Covariation = multiplicitive transformation of the pearson correlation coefficient.

(In other words, they mean the exact same thing; "correlation" is just the more common/lay term for the mathematical/philosophical concept of "covariation")

[#6 posted by Kinnaird , March 29, 2008 3:25 PM]

Was he wearing the galoshes on his head?

[#7 posted by Jake0748 , March 29, 2008 3:37 PM]

#6, yes, and then in the middle of the night someone snuck in and put them on his feet, and conked him on the head.

[#8 posted by heresiarch514 , March 29, 2008 6:44 PM]

Jake078 @ 7: Wouldn't they conk him on the head first?

[#9 posted by Jake0748 , March 29, 2008 7:30 PM]

Well, I dunno, I guess because a conk on the head is more effective when there's not a boot covering it? :)

[#10 posted by blogmother , March 29, 2008 9:05 PM]

Nick, they don't mean the same thing. correlation implies a relationship between the variables. Covariance does not---it's the word to use when the apparent relationship could be accidental / coincidental

[#11 posted by Nick , March 29, 2008 11:03 PM]

Blogmother - I'm not sure to what usage you're referring, but the statistical terms "covariance" and "correlation" are inextricably linked--if two variables covary to any degree (and thus have non-zero covariance), they will have a non-zero correlation coefficient.

In the philosophy of science, covariation is one of JS Mill's three criteria for finding causal relationships. He keeps covariation separate from his other criterion of "non-spuriousness," which is the part that deals with whether the realationship is real, or just due to coincidence or external factors.

[#12 posted by XuYu , March 30, 2008 8:08 PM]

Ummmmm... so he gets headaches when it rains the night before?

[#13 posted by Takuan , March 30, 2008 8:16 PM]

no no no ! He gets headaches because he puts his galoshes on his head and when they go to hit his feet they bash his noggin! Sheesh!

[#14 posted by pabo , March 30, 2008 11:41 PM]

Not sure how using galoshes are unhygienic, but I assume that waking up with a headache and galoshes on means that he got drunk the night before and didn't passed out without taking them off.

[#15 posted by murray , March 31, 2008 5:47 PM]

My wife has trouble differentiating these two things, which can make discussions (especially debates) with her difficult. It also causes her to just say some darn silly things. Or at least, I think that's what causes her to say silly things. But I might be confusing causality with covariation.