Experts in Territory, Location & Field Force Optimisation

Causation vs Correlation – What’s the Difference and How Can a Mix Up Cause Serious Problems?

Did you know that there’s a direct relationship between the number of people who die from being tangled in their own bedsheets and the consumption of cheese in the United States? Looking at the data below, the evidence is clear, unrefutable and statistically sound. After all, the trend matches almost perfectly for a full nine years.

Or could we be looking into the eyes of coincidence?


If we actually look at this with even the smallest amount of common sense, we know there’s actually no relationship between the two trends. There’s simply a correlation and there are many examples of where this phenomenon exists.

Causation vs Correlation – Why is it Important?

Decision making in business often finds us investigating issues in order to find the source, which of course makes perfect sense and is entirely justified. But once we’ve discovered what may be causing the issue, what we have to do is establish the causal link.

A Worked Example.

Mark is the Head of Human Resources at a call-centre in which there’s a team of over 50 staff who spend their day consuming coffee and making outbound sales calls. It has been brought to his attention that 10 members of the sales team have suddenly left the firm in quick succession.

The first thing Mark realises is that a new Team Leader was recently appointed from outside the company and has been known to rule with an iron fist. What is Mark’s first thought?


Consequently, the new team leader is instructed to change his supervisory approach.

The following week, Mark is surprised to find that a further five employees have left the company and everybody under the new Team Leader is statistically selling less every single day. What comes into mind for Mark this time?


A memo arrives on Mark’s desk highlighting that a new competing call-centre has recently opened less than a mile away and offers a much more attractive commission structure and has been head-hunting members of the team. Those leaving the team are all now employed by the new call-centre.

In reality, those leaving the company were simply moving to another company in search of better commissions.

Because of Mark’s mistake in attaching causality to correlation

  • A perfectly good Team Leader was made less effective
  • Sales fell as a result of the Team Leader being micromanaged
  • The opportunity to retain good salespeople was lost

Start Asking Why

Next time you draw conclusion from data analysis based on correlation, ask yourself the question “why?”. This is of course a simple exercise but it will help draw the causal link between two trends. Ask yourself if there’s another way the data could be correlated. Ask yourself what else could be causing the effect.

When a correlation exists, it’s important to look at other factors that could be causing the outcome. In the above example, when Mark found out people were leaving the company quickly, instead of jumping to conclusions, he should have been looking for all the possible reasons why people may be leaving the team and not jumping immediately on his first thought.

Identifying correlation is a stimulus not for encouraging you to find an immediate conclusion but instead for encouraging you to do further research. It’s simply one of many different statistical relationships data can have.


PHP Code Snippets Powered By :