2015-03-14

Correlation Vs.Causation

Nowhere more than in politcal discussion is there greater evidence of our misunderstanding of what it is to make a correlation as compared to arguing, not to mention proving, causation.

I thnik that a lot of folks don't even know the difference, and the greater majority of the remainder don't bother to make the distinction when discussing an issue.  So?  What is the difference?  And why does it even matter?  What is the big deal?

As usual, I will start with a rediculous example:  This morning, while getting a bit of fresh air on the front stoop, I noticed two garter snakes soaking in the sun on the rocks in the front garden.  This is the fouth day in a row that I have seen these pretty little creatures in the same general vicinity at approximately the same time of day.  In the  eight years we have lived here, this is the first year that I have seen snakes out this early in spite of the fact that we have had similar warm weather this early in the year.

Now, using only correlation, I could draw all sorts of conclusions, anything from there is a growth in the population of gartern snakes to a warming climate and on to food supplies and the desire for slithering creatures to be around someone like me.  I have verifiable evidence for any one of these conclusions including dated photos on my cell phone.  However, I have no proof that there is a direct link between anything I might conclude and the fact that the snakes are suddenly appearing at the aforementioned times.  There is no evidence of causation.

I have recently witnessed some of the most rediculous examples of mistaking correlation for causation that can be immagined in the news, on facebook, on the radio, almost anywhere I look or listen.  Many of the examples of this confusion are tightly coupled with political issues.  Some of them simply defy common sense.  Others are understandable because the issues are tightly coupled to our daily lives, whther in how we believe or in how we operate.  Issues like climate change, where we are only now beginning to be able to gather and assess data that might indicate causation.  Other less tangible issues like LGBT rights, or race and gender equality have plenty of data which indicates how things are and have been, but not how to change them if we wanted to.

Don't get me wrong:  There is nothing wrong with correlation!  It is how our minds recognize patterns, and pattern recognition is a huge part of our ability to think.  But a big part of thinking is asking questions about what the patterns we recognize might mean.  Thinking is our biggest asset as humans.  And yet by drawing conclusions based on observations we have not asked fundamental questions about, we shortcut the thought process and often as not, based on what I have been seeing recently, make asses of ourselves.

Theorizing about correlations we make is good, drawing conclusion based on them without providing real evidence of a causal model is self destructive and destructive to the process of exchanging ideas and learning from that exchange.

Might I suggest we at least understand and distinguish between correlation and causation?  What can it hurt?

Post a Comment