2. Regarding the point the last video made that “Correlation does not imply causation” do you think that the design used to determine a correlation can also be used to determine a causation, or should a different experimental design be used to determine a causation?

Experimental design, alternative to the one used for correlation, is required to determine causation. There needs to components of the causation experimental design showing which variable impacts the other.

Further, there are so many possibilities for association types, including:

The opposite is true: B actually causes A.

The two are correlated, but there’s more to it: A and B are correlated, but they’re actually caused by C.

There’s another variable involved: A does cause B—as long as D happens.

There is a chain reaction: A causes E, which leads E to cause B

I have an example of experimental design in product analysis. Imagine launching a new version of a mobile application and are analyzing whether product user retention is linked to in-app social behaviors. The product team implements a “communities” app feature.

To determine whether communities impact retention, the team creates two equally-sized cohorts with randomly selected users. One cohort only has users who joined communities, and the other only has users who did not join communities. Results show a correlation between joining communities and retention.

The next step is to add to the experimental design in order to determine whether there is causation. It is essential to “extensively test the relationship between a dependent and an independent variable before asserting causality” (Madhaven 2020). An example of experimental design used to identify causation with a product is split testing. It involves changing one variable and seeing what happens. Do something that prioritizes the input variable and increases it. If your outcome consistently changes (with the same trend), it is clear the variable being changed makes a difference. Also, when making a case that joining communities leads to higher retention rates, eliminate all other variables that could influence the outcome.

Source:

Madhaven, Archana. 2020. “Correlation vs. Causation: Understand the Difference for Your Product”. Accessed July 9 2020. https://amplitude.com/blog/2017/01/19/causation-correlation