Dave @ Cornell suggests using a goals-allowed-per-minute stat instead
of a goals-allowed-per-attempt for power play/short-handed stats. I, too,
thought this new stat would give "truer" rankings of teams' PP/SH teams
until I saw the stats from the NHL.
 
From the 80 game NHL 90-1 season (stats from The Hockey News):
 
Rank   PPG/Attempt     PPG/Minute
  1      Calg           Calg
  2      NYRang         NYRang
  3      Pitt           Chi
  4      Chi            Pitt
  5      Bos            Bos
  ... (too lazy to type them all in) ...
 17      Edmon          Edmon
 18      Van            Van
 19      Det            Det
 20      NYIsle         NYIsle
 21      Que            Que
 
Basically, teams come out the same in both lists. Granted, I haven't
done a statistical test to see if the two rankings are significantly
different, but my eyeball tells me there is not enough to worry about.
 
The two biggest differences were LA (#6, #10) and Win (#11, #7), btw.
 
The short-handed tables looked alike, too.
 
And the first one is SO MUCH easier to compute, so why bother?
 
Keith