I noticed a strong positive correlation between social media engagement and in-store foot traffic during our last campaign. My boss wants to double the social budget, assuming the posts caused the traffic. How do I explain that correlation doesn't equal causation using business statistics? I’m worried there might be a "lurking variable" like a seasonal holiday affecting both.
3 answers
This is a classic "Lurking Variable" scenario. To prove causation, you’d need a controlled experiment, like running social ads in one city but not another, and then comparing the traffic. A simple correlation only shows that the two trends move together—it doesn't say why. It’s possible that your holiday sales event caused both the social posts (because people were excited) and the foot traffic (because of the discounts). If you double the budget without a clear causal link, you might just be spending money on noise rather than a driver of revenue.
Have you tried using a Scatter Plot with a trend line to see how tight the relationship really is? Also, what happens if you lag the data—does social engagement today predict traffic tomorrow, or do they happen at the exact same time?
Just show him a graph of "Ice cream sales vs Shark attacks." They correlate perfectly in summer, but ice cream doesn't cause shark bites. It’s the heat (the lurking variable)!
Haha, Karen! That’s the most famous example for a reason. It’s the easiest way to explain a complex statistical concept to a non-technical manager.
Steven, lagging the data is a brilliant move for a business analyst. If the social posts happen after the foot traffic peaks, then clearly the posts didn't cause the traffic—maybe the customers were just posting while they were already in the store! This kind of temporal analysis is essential for debunking false causal claims and protecting the marketing budget from inefficient spending.