The Science of Why: a brief introduction

Understanding how causation works and how to work with it

Arthur Mello
5 min readJul 30, 2021

“Correlation does not imply causality”. You have probably heard this before, and by now the difference between those two should be pretty obvious, right? Well, not really. Although we understand they are not the same thing, causality isn’t that easy to define.

Moreover, it is also hard to measure and identify. In this article, we will try to provide a definition for causality and, perhaps most importantly, how can we identify and measure it.

What is the difference between correlation and causation?

When two things usually happen together, we say they are correlated. For instance, when it’s Christmas, weather is usually cold in the Northern hemisphere (and warm in the Southern). Those things are correlated, but does that mean the cold weather causes Christmas (or that Christmas causes the cold weather)? Well, that does not make much sense, right? That is because causation implies that one thing would be less likely to happen (or would not happen at all) if it were not for the other. For example, when it rains, people use umbrellas. Rain and umbrellas are correlated but there is also a relationship of causation: rain makes people get their umbrellas. If it wasn’t for the rain, people would probably not be using their umbrellas.

Causation is defined as “the relationship between cause and effect”, whereas cause

--

--

Arthur Mello

Data scientist and educator. I write about data analysis and machine learning applied to marketing.