No, this is not an April Fool’s post.
I found the argument about mexican lemons at Derek Lowe’s In the Pipeline (if you’ve not got it on your RSS aggregator, get thee behind me) and thought it was a better way to celebrate the day of fools than by doing something fake like “I’m going to work at Elsevier as head of intellectual property!” or some such.
You see, we get fooled all the time by data. And frequently we really, really want the data to tell us something specific – something we’ve been looking for, for a very long time. Or we want to have the data tell us something coherent – to tell us a story. Eric Schadt uses the phrase that “we are wired for storytelling, not for statistical uncertainty,” and I tend to think he’s right.
This means we’re wired in some ways to find causation when all we’ve got is correlation.
But no matter how badly we want it…mexican lemons just ain’t the reason highway deaths are going down.