General, Tips

Ice Cream Sells Air Conditioners

Olumide Olasope Written by Olumide Olasope · 1 min read >


 If you have ever done any course in statistics, I am sure you are tired of hearing “correlation does not mean causation.” It is drummed into the heads of every student, and for good reason, it is a juicy trap to fall into.

Today’s fallacy we are discussing is the correlation/causation fallacy. The explanation is in the name; just because two events happen at the same time, does not mean one is causing the other. Even when two events occur so often together, they may have a link, but don’t immediately jump to the conclusion of causation.

What is Correlation?

It simply refers to how closely related two things are. It is represented by a single number and it explains the relationship e.g. as one increases, the other increases/decreases. When even statisticians spot two things so closely related, there is always that temptation to think that one causes the other.

The Fallacy

There are multiple ways we can fall prey to this fallacy, even people who are aware of the fallacy are not completely immune to it. The three ways this fallacy presents itself are:

  • Reverse cause
  • Third factor 
  • Coincidental
Reverse Cause

In this form of the fallacy, there is actually a causal relationship, but the person has it backwards. I can give an extreme example here; “When windmills spin quickly, there is a lot of wind; therefore, windmills spinning quickly causes wind.” Nobody needs to explain the flaw in the thinking here, but the reverse cause doesn’t always present itself glaringly in this way.

Another example that has always been a pet peeve of mine is the news that just keeps popping up in the media. It says that violent video games make children violent; but wouldn’t it make sense to instead infer that violent children are drawn to violent video games?

Third Factor

The explanation for this is in the name, it happens when you conclude that A causes B, when in fact there is a third factor “C” that caused them both. A good example here would be an increase in ice cream sales causing an increase in the sales of air conditioners.

Spotting the third factor here is easy, it is the weather; when it gets hot, people buy more of both. It is easy to spot when it is reduced to this point and laid out cleanly. But this third-factor issue has led to ideas like an air conditioning company advertising ice creams to increase sales; and because of correlation, they end up assuming that the tactic is working


This is the one that happens very often with people new to data analysis. There was no correlation in the first place, the data could have been bad or manipulated, the analysis could have been flawed or it was pure chance.

These issues happen, but they usually occur when you don’t have enough data to run a proper analysis of a population


To wrap things up, if you want to claim causation, remember that the burden of proof lies on you; you have to present facts or extra data to support your claims. Don’t pass this burden onto others, because you are playing into another fallacy we discussed earlier on this blog.

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.

%d bloggers like this: