Correlation vs. Causation in the Digital World
I’m not sure if it was a psychology, sociology or journalism ethics class when I was introduced to the correlation vs. causation conundrum we all face. During this pandemic, correlation is often mixed with causation. It seems research has taken a back seat.
So C vs. C has been on my mind. It seems like for some time, the digital trade press was freaking out about how much time is going to mobile. Many called it the “death of desktop search.” Except throw in a pandemic and up goes the desktop search numbers. Causation or correlation? Don’t know. Need research.
And now, the story is that people have “turned away from browsing the web on their phones because they are using apps more.” Wait, what? A media distribution company I have respected for many years says, “Mobile web is already dying as people are spending more time in apps like Facebook and Twitter.” Wait, wait, what?
In this age of too much data, we have forgotten one thing: Correlation and causation are not the same thing. The book and radio blog Freakonomics is full of false or spurious cause-and-effect relationships. For the record, causation means A causes B; correlation is that A and B were observed at the same time and there may be some causation, but we don’t know that.
The fact that I spend more time on Twitter than I do searching in my browser is not related. In fact, the more efficient and accurate search becomes, the less time I’ll need to spend searching for what I want. I know a lot of people who spend a lot of time in game apps. Are those related to less time in search? Absolutely not.
I read in a blog by Chris Taylor for Wired magazine that, “If big data came in a box, it would be stamped, ‘Warning: Correlation does not imply causation.’ “ This is especially important when you are making decisions based on the data. Make sure you are not making leaps of faith based on comparing two data sets or two outcomes. Trust, but verify.