The False Cause Fallacy: Correlation Does Not Equal Causation

When we see that two things happen together, we may assume one causes the other. If we don’t eat all day, for example, we will get hungry. And if we notice that we regularly feel hungry after skipping meals, we might conclude that not eating causes hunger. A good deduction!

But does the same logic always apply? Are two things that seem to occur together always related? Or is this sometimes an error? In this post, we look at correlation and causation to help you understand – and hopefully avoid – the false cause fallacy in your academic writing.

Correlation and Causation

A correlation is a mutual relationship between two or more things. Typically, this is a statistical relationship where two variables are interdependent:

• A positive correlation occurs when two or more variables seem to increase or decrease together. For instance, there is a clear correlation between the variables ‘foot size’ and ‘shoe size’ because people with bigger feet reliably have bigger shoes.
• A negative correlation occurs when one variable increases as another one decreases. For example, the variables ‘speed of vehicle’ and ‘duration of journey’ are negatively correlated because a faster vehicle will typically complete a journey in less time.

Correlations like this can be useful because they can help us spot a connection between two things. In some cases – including the examples we’ve used here so far – you can even identify a causal relationship between the variables. For instance, few would deny that skipping meals can cause hunger, or that a faster vehicle can reduce journey time.

But we must be careful when drawing this kind of conclusion. Correlation does not always imply causation. And if we misinterpret a correlative relationship, we might fall into the false cause fallacy.

The False Cause Fallacy

The false cause fallacy arises when we wrongly assume that one thing causes something else because we’ve noticed a relationship between them.

For instance, if one thing happens after something else, we may assume that the first causes the second. However, following from or coinciding with something is not the same as causing it. And if we are too quick to conclude a causal relationship, we might end up with a false cause.

Two major hazards here are reverse causation and spurious correlation.

Reverse Causation

When looking at a correlation, we may misunderstand the relationship between the variables. And this can lead to mixing up a cause and an effect.

For instance, based on a correlation alone, it would be just as reasonable to believe that windmills cause wind as it would be to believe wind causes windmill blades to turn. All we know is that the two things happen together, increasing and decreasing at the same rate.

For anyone who knows anything about windmills, this is obviously a false cause: windmills catch wind to create rotational energy, not the other way around. Thus, a correlation can only tell us about a cause if we know how the variables are related. And if we get this relationship wrong, we can end up with reverse causation.

A similar error is the post hoc ergo propter hoc fallacy (i.e. after this, therefore because of this). This involves assuming that the order of events implies causation.

Spurious Correlation

The false cause fallacy can also occur when there is no real relationship between variables despite a correlation. For example, there is a genuine statistical correlation between films released featuring Nicolas Cage and the number of people who drown in US swimming pools each year.

If correlation implied causation, we might assume that Nicolas Cage films are deadly around water. But this would be at best a hasty conclusion.

As with the windmill example above, correlation alone is not proof of causation. If we truly wanted to say that one of these variables caused the other one, we would need to explain how Nicolas Cage films are related to pool deaths. And we’d need evidence that the two things were connected.

Without this, we’re left with a spurious correlation (i.e. two things that coincidentally overlap in some way). And we cannot draw any useful conclusions from this kind of relationship between variables.

How to Avoid False Cause Fallacies

So, then, how do you avoid the false cause fallacy in your own work? We have a few tips that you might want to follow:

1. Remember that correlation does not equal causation. It is fine to report a correlation in your data, but you cannot assume a cause and effect relationship from that alone.
2. Always consider how variables in a correlation are related. Think about non-causal explanations, such as pure coincidence. Is there enough data to suggest a strong correlation between two variables?
3. Consider whether other variables could explain the correlation. For example, ice-cream sales and hospital admissions for heat stroke are positively correlated because both are influenced by a third variable (i.e. high temperatures), not because eating ice-cream causes heat stroke.
4. If you are going to argue that a correlation suggests a causal relationship between variables, back this up with evidence.

Don’t forget, too, that having your work proofread can help you express yourself clearly. And the more clearly you can set out your arguments, the easier it will be to avoid false cause fallacies.