Correlation and Causation


I used to work in public health and then I got a history degree. As a result correlation and causation are very important to me. And just like when I was 20 years old, working on cancer cluster research it boggles my mind when people don’t understand the difference between the two or confuse them.

Correlation is when two things happen simultaneously that may or may not be related. Causation is when something happens and causes something else to happen.

For example, everyone who gets cancer drinks water. Does this mean water causes cancer? No, it means people need water to live. The two correlate, water isn’t the cause. And before you argue that not every person on the planet drinks water, yes they do, it may not look like water or taste like water, but it is still water. It doesn’t matter if it’s tea, coffee, soda, beer, Kool-Aid, fruit juice, V8, Gatorade, sparkling water, tap water, or distilled water, it’s all water. Even milk and wine is primarily water.

Now, if you live somewhere like Herculaneum, Missouri, there is a very, very high chance that drinking the water will give you cancer. Because Herculaneum, Missouri is a superfund site: a location where the soil and groundwater has been contaminated with toxic chemicals due to improper disposal of hazardous materials.

But most of us don’t live near a place like Herculaneum, Missouri and the water isn’t going to cause us cancer. It’s a correlation. The two things happen simultaneously, because people need water to live and people get cancer.

Here’s another correlation, that does seem to be completely random. Nearly everyone I know with CRPS owns a dog. When I first started experiencing pain and problems with my hands and forearms, I owned a border collie named Frisky. Did Frisky cause my CRPS? No. I have two possible injuries that caused my CRPS and neither was dog related: injury one I stood up on a teeter totter and the person on the other end jumped off, I broke my wrist. Injury two I accidentally shoved a woodburner into my thumb and caused a 3rd degree burn right down to the bone. Either of these are possible causes. Burns and broken bones are the most common triggers for CRPS. Here’s the shitty part of that, they aren’t the only triggers. You can develop CRPS after spraining an ankle, breaking a nose (no bones in the nose), having a surgical procedure, or childbirth. There is a correlation between owning a dog and having CRPS, but not a causation. And plenty of people own dogs that never develop CRPS.

False and misunderstood correlation and causation is the reason there’s an outbreak of measles in the state of Washington. There is a correlation between vaccination and autism. Vaccinations became more common in the 1950s and 1960s. Autism was diagnosed for the first time in 1933. However, it did not become a “common” diagnosis until the 1980s. The diagnosis of autism, has nothing to do with vaccines and everything to do with our understanding of mental disabilities and the desire to differentiate and define them better.

I’ll give you an even clearer example of misunderstanding of the two. 1952 is the first year that heart disease was the number one killer of adult Americans. There is a definite reason for it, but not a cause. We didn’t suddenly have a massive increase in the number of people dying from heart disease. From 1870-1951 the number one killer of adult Americans was tuberculosis. In 1949, we found the first antibiotic that cured tuberculosis. In 1950 and 1951 the use of streptomycin to cure TB became widespread across the US, Canada, and UK. As a result, 1952 is the first year millions of Americans didn’t die of tuberculosis. There wasn’t an increase in cases of heart disease, there was just a decrease in the mortality rates of TB.

However, if you look at the data searching only for causation, it appears curing tuberculosis made heart disease more deadly. More confusingly, correlation can mask causation. When you start searching for a cause, a strong correlation can be misleading. For example in the 2000s violent crime in Detroit, Michigan dramatically decreased. Gun sales also decreased. The two did not cause each other. The two were caused by the same factor and correlated very strongly as a result. I remember reading a sociology paper about lower gun sales leading to less violent crime at the time. However, gun sales and violent crime decreased because the population of Detroit decreased. In 1990, Detroit had a population of 1.2 million people. In 2000, the population had dropped to 900,000. By 2010 it was down to 650,000 people. The drop in population means fewer criminals and fewer people available to legally buy guns.

I’ve used simple examples, but causation can be very complex. There might be six or seven factors at work in a causation, some of which may not be obvious. The black death was one of the worst plagues to ever hit the human population. But bubonic plague isn’t actually that contagious. There were multiple factors at work to make the Black Death as formidable as it was: drought caused crop failures, crop failures lead to even higher rates of malnutrition, malnutrition causes higher susceptibility to diseases and illnesses, Bubonic Plague mutated to become communicable person to person without the parasitic vector (no fleas needed in other words), also malnutrition leads to fewer antibodies to fight a disease once you have it, making bubonic plague nearly 100% fatal. And suddenly, Bubonic Plague a common illness in the 1300s kills tens of millions across Europe and Asia.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s