The showdown! Correlation vs. Causation…

Okay…it’s not that dramatic, I admit. However, this is an important issue that I think gets glossed over by too many people. So let’s explore the difference between these two concepts, why they are commonly confused, and why that can be a problem for your business.

What is the difference?

If you don’t read the rest of this post, please at least read this. Correlation means that there is a relationship between the two variables, but that relationship does not mean that a change in one of the variables causes the change in the other. In fact, it could be that other variables cause changes in the correlated variables. Another problem is that correlation may actual indicate causation, but we can’t determine which variable is causing the other to change.

Examples are important. A good example of this would be cold weather. If you are running a toy store and did a correlation analysis between the temperature and your sales, you may find that colder weather is correlated with higher sales. If you assume a causal relationship between these two variable, you may project sales during colder periods with higher sales. Let’s say it gets colder during the spring (it literally snowed yesterday in Indiana in late March), but your sales did not increase. Why didn’t this hold true? Because there are other factors that may affect these results. Colder temperatures typically include the holiday season which is associated with higher consumer spend. While the variables are correlated – meaning that in general the colder temperatures rise and fall with sales – one is not necessarily the cause of the change in the other.

Why are they confused?

This is one of the statistics that I find are commonly confused by many. If you are in an industry for a while, it is easy to notice that changes in specific numbers are related. You may even have anecdotal evidence that the numbers are related. When you run statistics that prove that your variables are related, its easy to fall into the trap that one of the variables is causing the other. This is what psychologists call confirmation bias. This is the tendency to interpret information that supports your beliefs.

Why is this a problem?

The problem with this is that leaders are working to make data-driven decisions. If they are using correlational data to make these decisions without ensuring that they are exploring the root cause of the changes, they may implementing changes that do not have the effect that they intend. This can be the cause for a failed change in an organization.

To get statistical causation, a controlled study must be conducted. This is not feasible for most organizations. So how do you avoid the trap of making changes related to correlation? The answer is to do a thorough root cause analysis. There are multiple tools that can be employed to explore the root cause of an issue. These techniques use methods including divergent and convergent exploration and challenge mapping, among others, to identify the most likely cause of the issue.

Here’s the pitch!

Being trained and experienced in multiple process improvement and change methodologies, I can use these techniques, as well as my experience with research and statistics, to help you deep dive into the issues confronting your organization. The education, experience, and tools that I have to explore these issues will help you identify the underlying causes for the problems that are keeping you awake at night. Then we can work together to develop a plan to address those issues in the most efficient and effective manner.