Get your degree at ACCU!

Welcome to ACCU!  We only have one curriculum here, and it's the study of how things relate to each other. 

That's it.

The Association Correlation Causation University was founded a few minutes ago when I decided to compose this post.  In that short period of time, we've educated no one.  But we have high hopes for the future.  VERY high hopes.

I hope that once you earn your degree at ACCU you will recommend it to your friends.  Now let's get started:

First please notice that the name, ACCU, is the same as the first four letters in "accurate".  That's a complete accident, and I just noticed it myself so I thought I'd point it out.  But it does bring us to the "A" in ACCU, which stands for Assocation.

Assocation is the state of being associated.  Not too helpful, but it's a start.  Associate means to join or connect together.  So to say two things are associated says they are connected somehow.  It doesn't say anything more or less. 

Let's look at an example:  Drinking water is associated with living.  In other words, there is a connection between drinking water and living.  But only a connection.  To say they are associated doesn't imply that more water makes anyone live longer.  Nor does it say less water will make someone live longer.  In fact, drinking water is also associated with dying.  Most people who die have at some point had a drink of water.  Not all, but most, so it's fair to say they're associated.  Again, just association.

Now if you start noticing trends in your associations then you might have a Correlation.  Correlation is a step beyond association. Merriam Webster said it beautifully:

"a relation existing between phenomena or things or between mathematical or statistical variables which tend to vary, be associated, or occur together in a way not expected on the basis of chance alone."

You're only seeking a Bachelor's Degree at ACCU, so let's keep it simple.  If one thing correlates to another, it means they vary positively or negatively with each other.  As one increases, the other increases is a positive correlation.  A negative correlation means one goes up the other down, etc.  An interesting example is the presence of fire trucks and fires.  There is a strong positive correlation between fire trucks and fire damage. The more fire trucks present at a fire scene, the more damage done at a fire scene.

Now, this is an accurate correlation.  The two things vary positively with each other.  But are the fire trucks causing the fires?  No!  Notice that a correlation only shows that two things vary together in a way "not expected on the basis of chance alone."  Correlation does not show Causation. 

Oops, it appears we're nearing the end of our curriculum, because that's the final C in ACCU!  Causation is the act or process of causing.  It's important to realize, as the example shows, that two things which correlate may or may not indicate causation.  It is a common mistake prior to attending ACCU to naively accept causation when all that has been observied is correlation.  In some cases, even simple assocation is enough to convince some people of causation.

I once had a friend who said that milk couldn't be bad for you because there were people who lived more than 100 years and they drank milk.  Pop quiz, what is this relationship?  It's just association.  Some people who drink milk live longer than 100 years.  It isn't even a positive correlation, which would take the form of evidence that lifespan increases with increases in milk consumption.  Yet this simple assocation convinced this person of causation.  Don't fall prey to the same error in thinking.

Which brings us to an additional point (don't worry, I'm only going to cover two of them) which is the idea that all of these relationships are assumed to be taken on average.  The presence of exceptions at the association and correlation level doesn't necessarily invalidate the relationship though it may shoot some holes in causal relationships. 

As an example, consider that higher IQ correlates with higher net worth. In other words, the smarter you are, the more wealth you accumulate.  There may be rich idiots and broke geniuses, but ON AVERAGE there is a trend in the association between IQ and wealth that is greater than can be explained on the basis of chance alone.
Let's just hit that next point as well and be done with it, which is the idea of a confounding variable.  Up till now our examples have only included two variables.  It is quite possible to establish a false correlation via confouding variables.  It is even possible to become convinced of causal relationships by ignoring confounding variables.

So just what is a confounding variable?  Let's imagine that we have a theory that drinking in night clubs causes lung cancer.  So we create a poll of 10,000 people and we ask them these two questions:  1) Do you frequently have drinks in a nightclub?  2)  Do you have lung cancer? 

Now let's imagine that 5000 people say they DO have drinks in a nightclub and of those 5000, 500 of them have lung cancer.  Of the other 5000 people who said they DO NOT have drinks in a nightclub, 100 of them say they have lung cancer.

The math is simple.  500/5000 people who drink in nightclubs have lung cancer and 100/5000 people who don't drink in nightclubs have lung cancer.  The next step is a headline on a newspaper or magazine which reads "Nightclub drinking makes you five times more likely to get lung cancer."

That's where those bizarre headlines come from, but it doesn't necessarily mean that because of two things.  First, this is a correlation based on observation, it does not prove causation.  Second, there might be confounding variables.  It might be that if we added a third question to our poll, "Do you smoke cigarettes?", then we might find that what we thought was a correlation between nightclub drinking and lung cancer is really a correlation between nightclub drinking and smoking, and it is the smoking that correlates to the lung cancer.

So it's time for your graduation from ACCU.  I'll deliver the commencement speech.  Here it is: 

Go forth and use what you've learned to listen critically to the flow of B.S. that occurs in the media, in books, and in conversation.  The next time you hear a headline that reads "Study shows that increased camel feces consumption lowers heart attack risk." find out if the study really showed causality or if it just observed a correlation.  Read the actual text of the study and see if causality is demonstrated or if correlation was observed.  Ask yourself what confounding variables the researchers might not have considered.  And just in general, don't be a dolt.

Throw your little square hat thingy in the air now and you don't have to come to class any longer.