Causation and Correlation

There is wonderful data that definitively show that as the number of pirates decreased worldwide, global warming increased.  Obviously, the only hope for the planet is the immediate rebirth of the Pirates of the Caribbean.  Where is Johnny Depp when you need him?

Before I jump back into some of the more ludicrous examples of correlation being misinterpreted as causation, let me make an impassioned plea for at least one more pirate.  If Greenpeace were to purchase just one measly little attack submarine from the former Soviet Union and to use that sub to attack Japanese whaling vessels—while displaying a green Jolly Roger flag—I would happily send them a large donation.  

If we compare two dissimilar variables and find certain similarities, it is simply human nature to conclude that the events are linked.  This sounds preposterous, but luckily there are statistics and graphs enabling everyone to prove that anything is possible….As long as you don’t do the math (and you abandon common sense).

Humans are hard-wired to find patterns in unrelated data.  This is why we see shapes in clouds, we hear the phone ringing while we are in the shower, and we find a face on the surface of the moon.  Couple that with our tendency to only observe data that correspond to our prevailing beliefs and we are ready to jump to the wildest conclusions, all supported with “facts”.

Watch any Sunday morning news program and you’re likely to hear someone proclaim, “Correlation is not causation” in a futile attempt to win an argument.  Even though we know that simple coincidence doesn’t explain causation, this doesn’t prevent people’s continuing to make this common mistake.   Almost daily, the news is full of stories about public policy being based on the shakiest of mere coincidences.  

Take, for example, the Ivy League university professor who advised parents not to send their children to ACT and SAT prep courses because his extensive review of the data showed that the students who went through those courses averaged the lowest scores.  This noted scholar believed that these students would have scored better without the prep courses, ignoring the fact that the students who chose to take those courses were usually those who most needed them to make acceptable test scores.

If you were to graph the sale of ice cream by month and overlay the number of shark attacks per month, you will find that, in most years, there is a direct correlation.  Obviously, sharks are attacking people who taste like ice cream.  Either that or, as the temperature rises in the summer, more people eat ice cream and more people go to the beach.  Take your pick, but I prefer to believe that people who gorge on ice cream taste better.

Every year for the last century, the number of master’s degrees awarded by four-year universities is proportionate to the box office revenue of the motion picture industry.  This close relationship leads to the inevitable conclusion that the more the population is educated, the more it wants to see Rocky XI or Dumber and Dumberest.  An alternative (and ultimately, more boring) explanation is that both variables increase as the population of the country grows.  This would also explain why the annual production of nuclear power correlates closely to the number of pool drownings.

Recently, someone programmed a computer to shift through news reports to find correlated data.  Within minutes, the computer had found scores of examples and after running for several months, it had found more than 30,000 sets of correlated data, including the following:

The increase in imported lemons from Mexico inversely correlates with the drop in highway deaths.  As Facebook attracted new users, there was a matching correlation in the amount of national debt accrued by Greece.  The rise in cheese consumption exactly parallels the rise in the number of deaths caused by people getting choked in their sleep by their bedsheets.  The divorce rate in Maine mirrors the use of margarine. 

Naturally, my favorite example of correlation being mistaken for causation comes from history.  In 1830, as a cholera epidemic swept across Russia, the Tsar took extreme efforts to limit the spread of the disease.  Quarantines were imposed and the military was dispatched to the most stricken areas to impose order.  The Tsar even sent doctors to the worst afflicted areas to help the people in their suffering.

Unfortunately, the people noticed that wherever the soldiers and doctors went, the cholera epidemic was much worse.  Obviously, the soldiers and doctors were deliberately poisoning the peasants.  Desperate for the epidemic to be over—a feeling totally understandable to us now after the pandemic of 2020—the peasants launched a wave of violence that came to be called the Cholera Riots.  Mobs attacked public buildings, ransacked the state hospitals, and finally, in desperation, attacked and killed the doctors.

If this reminds you of anything recent, it is just correlation.