Thursday, December 28, 2006

A painful example of spurious correlation

Part of teaching is keeping an eye out for good illustrations. I'm teach methods, and this Christmas season has given me a great example of spurious correlation (at least I hope that it is spurious).

I started cooking dinner on Tuesday night at about 5:30. Before the meal was done, not one, not two, but three family members were in the emergency room!

Correlation? Absolutely--my cooking is highly correlated with rushing people to the hospital.

Temporal ordering? Yes--cooking came before hospital.

Plausible mechanism? Not sure. I didn't think I was *that* bad of a cook, but I could be wrong.

Here's what happened: Grandpa had chest/back pains (that ended up being a severely pulled muscle). My two sons had a trampoline accident. One lost half his front tooth, the other got a 13-stitch gash in his forehead. (Why they were jumping on the trampoline, in the dark, in the middle of winter is another story--being the sensitive father that I am, I now call them "chip" and "stitch").

This all happened in a 15-minute span, and we spent boxing day evening down as a family at the emergency room.

Everyone should heal up okay, but I warn you, I'm cooking again tonight.


Knumb said...

I hope Chip and Stitch recover well, and that Stitch doesn't scar badly.

Show Chip any picture of me smiling between the ages of 13 and 22 and you'll see ol' Uncle John Dog's chipped tooth, slightly discolored.

Then, show him any pic since and he will be able to see that they have improved greatly how well teeth get repaired.

And I turned out okay. (yea, right)

Poor guys.

weberms said...

Not really an example of a spurious correlation. It would be more of an example of coincidental correlation (correlation occurred by chance).

A spurious correlation occurs when two things happen at the same time (appearing to be related), but a third variable actually affects both.

An example would be the more fire engines that arrive at a fire, the more property damages that occurs. At first glance this could mean that fire engines cause property damage (variables appear to related), but in reality, the size of the fire (the third variable) affects both initial variables.