Tips on Arguing: Correlation versus Causation

Brandon T. Bisceglia

When two events are shown to be related in some way, they are said to correlate. When one event makes the other happen, the first is said to cause the second.

These two terms may at first appear to be similar, and they are. But the distinction is vital: two events can be related without having any kind of causal relationship. When you burn logs in a fireplace, for example, both smoke and light are produced. Though they occur together, neither causes the other – combustion is the cause of both. The smoke and light are therefore correlative, but not causative.

Causation can be a difficult phenomenon to pin down. Scientists will generally only claim such a relationship after many experiments have been performed, replicated, and reviewed. Even when a causal relationship is established, it can always be challenged by the introduction of new evidence.

Causation is often assumed without warrant in both public and private spheres. Perhaps you’ve heard that drug use is a cause of crime? While both do often happen in conjunction; however, poverty is actually a better predictor of both.

Perhaps you’ve seen An Inconvenient Truth and have been led to believe that Hurricane Katrina was a result of global warming. In actuality, we don’t know if there is even a correlation in this case. Such devastating hurricanes are hardly unprecedented near the Gulf Coast. They have been happening for centuries, long before humans started pouring extra carbon into the atmosphere.

On the other hand, the germ theory of disease was generally accepted after Robert Koch’s Postulates were published in 1890 based on his work showing that anthrax was caused by the bacterium Bacillus anthracis. Most viruses were too small to see until the invention of the electron microscope in 1931, but scientists nevertheless postulated (correctly) that they existed and caused illnesses.

When presented with an argument that presumes causality, it’s useful to first ask yourself: is there some other factor that could explain this? Is there a mechanism that could explain a causal relationship? If not, the most you can say is that two events are associated.

The key is caution. Jumping too quickly to causative arguments can lead you frustratingly far down a false path.