Correlation Does Not Establish Causation

Imagine that every morning you get up early, the external environment still dark from lack of sunlight, and you look outside before the sun comes up.  You do this every morning, and every morning you've done this the sun has risen afterward.  Does this mean that you looking outside makes the sun rise?  You looking and the sun rising are definitely correlated in this example.  Yet this does not mean that the looking causes the sun to appear; instead, it at most proves that the looking precedes the sun showing itself.

Correlation simply means that things appear alongside each other.  For instance, there is a very strong, consistent correlation between me pressing the light switch in my room and the light in my room turning on or off--almost every time I press the switch the light bulb reacts as I remember it behaving in the past.  Causation means that one thing or event is actually responsible for bringing another into existence or into a certain state, like a cold temperature turning water into ice or an experiment creating a certain chemical reaction.

The scientific method is used to examine cause and effect relationships in the external world.  However, scientists need to be careful not to confuse correlation for causation.  Even beyond this, though, thoroughly rational minds will recognize that there is no way to prove that a certain cause and a certain effect in the external world are connected.  One can prove a correlation without establishing causation.  One can even prove a steady and long-term correlation without establishing causation.  How is this so?

Let's go back to my light bulb example.  The correlation between the switch movement and light bulb activity is very strong, but this does not prove that the former causes the latter, only that they are correlated.  It remains possible that an alien is manipulating the light in my room to correspond to me pressing the light switch up and down, or that something else is actually the cause of the light, although there is no supporting evidence whatsoever for these ideas.  Still, they remain possible.

Correlation could exist because there actually is a causal relationship between two things or events, but it could also exist for an unknown reason, a reason other than one of the correlated things causing the other.  This is why correlation does not prove causation.  It serves as possible evidence for a causal relationship, but a sound thinker will not simply accept a correlation as necessitating a particular cause and effect connection by default.  Indeed, the soundest of thinkers, when it comes to causality, will realize that all effects by necessity have causes, but this does not mean that they can know exactly which causes bring about exactly which effects.

I clarified above that particular cause and effect relationships in the external world cannot be proven, but it is crucial that I acknowledge that I can still prove that cause and effect relationships of some sort exist because something cannot come into existence without a cause and there are clearly effects that occur.  It is just that I cannot prove just by seeing one thing precede or follow another that there is a necessary causal connection between them.  Correlation exists and causation exists.  But correlation alone does not prove causation.