We’ve all heard the phrase, correlation does not imply causation. Unfortunately, we rarely understand this phenomenon and are often mistaken by what we often label as evidence after examining the data. Sometimes both are referred to as cause and effect, which is potentially misleading. What exactly do we mean by correlation and causation?

Let’s examine one interesting fact: In the United States, the number of ice cream cones sold is directly related to the amount of home burglaries. In other words, there is a correlation between the number of ice cream cones sold and the number of thefts from home residences. If more ice cream cones are sold in a given month, you can safely expect the number of home thefts to increase as well. However, this does not imply that there is a causation between these two events. The sale of ice cream in no way motivates an increase in crime. Why then is there a correlation? Simply put, there is a shared variable between these two events. During the summer, more ice cream is sold. Also, during the hot summer months, more people leave their windows and doors open, and as a result, more people can easily gain unauthorized access to one’s home.

How about another example, this time from the United Kingdom? In a small village in England, residents noticed a strange pattern: For every household that had a new baby, storks were found resting on the roof. Superstition aside, we have a correlation between the number of births and the number of storks appearing on rooftops. However, without additional research and evidence, we cannot say that there is a causation between the two variables; babies, after all, do not cause storks (nor the other way around). We strongly suspect a shared variable exists between these two scenarios. Indeed there is: When a baby is born in the village, the house is kept warmer by placing more logs in the fireplace more frequently. Warmer homes equals warmer roofs. Storks, given the option to rest somewhere cold versus someplace warm will often choose the latter.

Often, this comes into play when dealing with marital therapy.  There are a lot of theories out there about how couples can better communicate, however research is often lacking.  One of the areas where research does exist is with Gottman Method Couples Therapy. It’s a great collection of research, well defined, and empirically based.  However, even their findings fall into one unescapable conclusion:  The results are causal, not correlational. Now, for the Gottman Method this doesn’t mean that we abandon our research; instead, it means that we need to be very clear on what factors we tend to see in poor communication between couples versus what are good predictors of successful marriages.

Even when a causation is found between two variables, this might not be reciprocal. Situation A might cause situation B, however situation B might have no influence on the chances of situation A happening. Take our stork example and let’s assume that, as unlikely as it might be, a causation does exist (it does not, however if proven wrong, please contact the author who will promptly retire and become rich). One might discover that having a baby does indeed cause a stork to arrive. However, upon further examination, having a stork land on one’s rooftop might have very poor probability of a baby being born in the house. This is an example of a one-way causation. Were storks to cause babys to be born and babies to cause storks to appear, we would have a bidirectional example of causation (not to mention a sudden jump in real estate prices).

Human beings are easily duped. We are fallible, gullible and often want to believe in something. However, our predilection for belief must be measured carefully against the rigorous discipline of statistical analysis and research. Otherwise, pigs might, indeed, fly.

Photo credit:  Pixabay/stevepb