Correlation Does Not Imply Causation

Correlation measures the relationship between two variables and can help predict one from the other. The page warns that a strong correlation does not by itself show cause, because the link may be accidental or shaped by a hidden factor.

The ice cream and sunburn example shows how temperature can drive both at once. The same caution is applied to software work, where adding developers does not mean a project will move proportionally faster. The page also connects this to systems thinking, which looks at the larger set of interactions behind the data.

Reading notes

#

• Correlation describes how strongly changes in one variable predict changes in another.
• Two variables can be correlated without one causing the other.
• Spurious correlation is a statistical relationship with no real-world connection.
• Ice cream sales and sunburn both rise with higher temperatures, which explains the apparent link.
• Adding more developers does not make software work scale in a simple one-to-one way.
• Causation is the direct relationship in which one event influences another.
• Systems thinking looks at the components and interactions behind the observed data.
• Correlations can point to places worth investigating, but they do not explain reality on their own.