Salon weighs in on correlation and causation

In  This great piece: Lies, damn lies, correlation and causation Salon's Alex Koppelman takes on the issue of correlation vs causation,making fun of conservative bloggers who credit George Bush with bringing the cost of oil down yesterday. It's not that far from what media companies have been doing, assigning causality to popups and banner ads when in fact the reason why people purchase might be something entirely diferent -- running into a friend on the street that recommends the product for example.

Correlations can be tricky. Most of the time, we correlate social media activity with results and point out, for example, that from a revenue perspective, there is a closer correlation with social media than with traditional media. We don't say that social media causes more revenue, because there are all kinds of different reasons why revenue comes in. But we can point out those things that when aligned seem to work better.

And why don't we prove causality you ask? Because it takes a whole other level of research that few clients will support and because relative to the issue at hand, that research isn't necessary. Why spend $100,000 to research the effectiveness of something that causes $500 to do?

 

Comments

[Edward O'Meara]

So true... unfortunately, what many PR practitioners learned a long time ago was that some clients interpret correlation as causation. In these client organizations, as the spin spirals upward, it can become hard to control. "Invest more!" "Get Us More AVE!"

When you try to explain the difference (no pun intended for those research geeks) these exec's get frustrated with the lack of precision (or double-talk, as I've been accused) in the results...