Understanding Correlation vs. Causation

We often hear the phrase “correlation is not causation,” which has been around for quite a while. It was notably articulated by David Hume back in 1739. To summarize his point: just because two things happen simultaneously doesn’t mean one is responsible for the other. Yet, despite this wisdom, it’s a mistake that many of us continue to make.

Common Misunderstandings

Here are a couple of noteworthy examples:

• For instance, the last two decades of research on gut biomes could very well be flawed due to the confusion between correlation and causation. Honestly, I’ve had a gut feeling—no pun intended—that there’s something off about this research.

• Also, the notion that moderate alcohol consumption is beneficial for health has been widely accepted. However, it turns out that its correlation with better health in certain groups doesn’t imply it actually causes it.

People often compile extensive data showing strong correlations—take the supposed link between rising autism rates and increased vaccinations. While it seems reasonable to conclude one causes the other, there’s actually no evidence supporting that vaccines lead to autism. It’s more likely that the diagnosis timing coincides with vaccination schedules and improved detection of autism.

Well-known examples illustrate this confusion, such as the curious relationship between ice cream sales and shark attacks. The rise in both during warmer weather might seem explainable, but we can’t say for certain that one leads to the other. Sometimes, different pieces of information simply don’t connect.

Consider the bizarre correlation depicted in a chart showing the ratings of “Two and a Half Men” directly correlating with jet fuel consumption in Serbia.

Looking for Patterns

There’s even a notable connection between online searches for “my cat just scratched me” and fruit consumption in the U.S.

I found this via Tyler Vigen’s site, which lets users create random connections and even fabricates amusing “research papers” to explain their findings.

In the case of cat scratches, an AI suggests that health-conscious individuals—those likely to eat more fruit—are also more cautious with minor injuries. So, while eating fruit doesn’t increase the likelihood of getting scratched, it might lead to more searches for treatment options.

Even if I find this reasoning a bit absurd, it still makes a kind of sense. That’s why this particular misunderstanding remains persistent; we have a tendency to grasp onto satisfying narratives. A neatly packaged explanation often feels more comforting than simply admitting uncertainty.

Misinterpreting correlation as causation can influence our choices, from diet trends to health guidelines, sometimes in harmful ways. The best approach is to remain skeptical when confronted with headlines claiming “X causes Y.” It might be worth treating such assertions as potential “cat scratches cause fruit consumption” situations until more solid evidence arises.