Tyler Vigen's Spurious Correlations website provides hilarious examples of accidental correlations. For example, he shows that there is a near-perfect correlation between Google searches for UFO sightings and kerosene consumption in South Korea. Nicolas Cage film releases correlate strongly with swimming pool drownings, and so on. Of course, what makes these examples so obviously spurious is the lack of any sort of reasonable explanation for the connections.*
But in the business world, this often is not the case. In a Harvard Business Review article, Michael Luca shows how correlations can be completely reasonable and also completely misleading. In one of his examples, eBay's marketing team found a strong positive association between paid search ads and sales. Of course, we would absolutely expect this to be true. More paid ads drive more traffic, which drives more sales. As you can guess, this turned out not to be the case. What was actually happening was that most of the ads were being shown to people who were going to buy no matter what. But how easy would it have been to pour money into additional paid search?
When we see a pattern, we usually reach for the most obvious explanation and then stop looking. Instead of just asking "why" an association might exist, we should instead ask: what are all the explanations consistent with what we're seeing, and how do we decide which explanation is valid?
* Reasonable is the key word here. I've met some people in my life who would have explained that the South Koreans were using drones to scout kerosene refineries before the upcoming invasion.
← Back to Writing