Correlation Versus Causation

TheHonestAtheistDecember 15, 2015

xkcd Correlation and Causation

So, often in my conversations with people, I find some who misunderstand the difference between "correlation" and "causation."

For instance, on one occasion, I was met with the claim that "vaccinations cause autism."

Rational skeptic I am, I asked the woman claiming this why she thought that. Her reasoning? When she was younger, she did not personally know anyone who was autistic, but now she did and vaccinations had been on the rise. My mind did a somersault. What???

Her bad reasoning aside (which would be an excellent lesson in critical thinking), her claim was unsubstantiated.

From my brief time in class studying statistics (which I hated and found extremely boring), I learned quite quickly and clearly the difference between correlation and causation.

When I began to think about it, I realized I had probably equated the two before taking that statistics class.

I now wish everyone had to take statistics, just to understand this difference.

Correlation represents that which is seemingly correlated, that which is typically concurrent.

Correlation can be both positive (A and B increase and decrease proportionally) or negative (A and B increase and decrease inversely proportional).

Causation represents event A causes event B, that performing A will result in B.

Causation is a subset of correlated--if something is a causation it is also a correlation (but not the converse).

As it turns out, causation is very hard to prove (see here if you are interested).

The main problems with the misunderstanding between the two, is that many companies still make use of correlation to back their decisions (such as Bing deciding on a specific blue for hyperlinks).

The takeaway is to be careful where we assume, only based on correlation that thing one causes thing two, as such reasoning can lead into the logical fallacy: post hoc ergo propter hoc or other silly notions (pirates & global warming).