Tuesday, September 29

Correlation vs. causation and hockey

Here's a good example of mixing up correlation and causation, for those sports fans who happen to be econ geeks. Our profs in empirical courses always harp on this kind of stuff.

I have the third overall pick from this year's NHL draft, Matt Duchene, in my hockey pool. So I was looking for analysis about how he might perform this season when I came across a posting on Mile High Hockey tackling the question of whether Duchene could benefit from another year of junior hockey before turning pro. The blogger does this by analyzing how other high draft picks faired in their rookie year, making this conclusion:
"Granted, this is an extremely small sample size, but the argument can be made here that the statistical output of the rookie season for these high draftees wasn't enhanced by an extra season in juniors. The 4 who made the leap averaged .53 points a game. The 6 who didn't? .43 PPG."
But there's an important oversight. Sure, there seems to be a negative correlation between playing an extra year of junior and output during one's rookie NHL season. But it's quite likely that the reason these players were sent back to junior for another year was because they were less skilled to start with. There's nothing to indicate that the extra year of junior made these players worse than they would have been had they jumped straight to the NHL.