When discussing hockey statistics, most people find it easy to rail against +/- as a bad stat. Fundamentally, people claim that it's a flawed statistic because it gives a player credit for something he might not have done, like getting a + while not helping to create a goal for. Critics say the fact that Mike Green and Alex Ovechkin were both near the top in +/- this year is proof that it stinks, since we know they don't play defense.
I've stated before that there are problems with raw +/-, and other stats, like adjusted +/- at behind the net, help to correct for it's deficiencies. But it's hypocritical to say that +/- is bad for the above reasons while pretending that every other stat in hockey doesn't have the same flaw. Rant time...
Goals: Goals are dependent on a ton of factors, like the quality of teammates and competition, and which zone you start in more often, among others. Easier competition means an easier time possessing the puck and scoring goals. Looking at players who played 40 games this year, 7 out of the 10 top ES goals/60 players had negative Corsi QoC's. Teammates have a similar effect, as good teammates can lure players away from you, thus giving you more time and space. They can also affect your goal total by being really effective passers, since killer set-ups aren't hard to finish. Looking at the same group, 9 out of 10 players had a positive Corsi QoT. Zone start is another factor in this equation, as players that start more often in the o-zone will score more goals. Once again, using the same group, 7 out of 10 players were used 50% or more in the offensive zone.
Assists: Assists are dependent on similar factors. Looking at the top 10 primary assist/60 players from this year, 6 out of 10 have a positive Corsi QoT. The perfect measurement for this would be looking at one's teammate's finishing ability; yet common sense still indicates that Crosby was hurt this year in the assist department because he doesn't get to skate with Mike Green, Daniel Sedin, or Nicklas Backstrom (like some other hart candidates). Zone start stats tell the same story, as 7 out of 10 in the same group had an OPCT of 50% or more.
Faceoffs: The NHL doesn't record faceoffs by clean wins and non-clean wins, so I've got to speculate here. Faceoffs are sometimes won not because the centerman did a great job, but because his teammates have quick reflexes and good sticks. It isn't terribly often that we see a draw won cleanly and straight back to the d-man. The scrambles that ensue after some (possibly most) are the result of a group effort, yet only the centerman gets the faceoff win. If scorers recorded how many faceoffs were won cleanly and how many weren't, then we might see that players who we thought were reliable in the dot might be benefiting from some good help.
Divisional Effects: Lastly, I want to cover divisional effects and how they affect point totals. Logically, it would make sense that playing in a weaker division would allow one to rack up points more easily. It's already been documented that players in the SE division get a boost in offensive numbers. Looking at the divisions from this past year, the two weakest divisions were the Southeast (- 42 goal differential) and Northwest (- 42). Looking at the top 30 total point players for this year, we see that 14 of those 30 players came from either the NW or the SE. Almost half of that group coming from 1/3 of the NHL's divisions is pretty good evidence that there's a bias depending on which division your team is in.
Ultimately, the next time you hear somebody claim that +/- is Satan's child, remind them that almost all hockey stats suffer from the same problems. Hockey is a team game, and since it's tough to separate an individual from his teammates, sometimes players will get credit for things they might not deserve. No stat is perfect, and +/- isn't alone.