In light of Schafer complaining about WCHA protecting their league and making stuff up, I took a look to see if there is any truth to this matter. A little analysis is dangerous, but here goes nothing. A few notes: includes 164 non-conference (nc) games with boxes available that are non-tournament games where the home team league is assumed to have provided the officials. Excludes common league opponents playing in NC games. PPO = power play opportunities.
@ Home Against NC Team NC Oppontents Apparent Protection Bias
Average PIM PPO / Game Average PIM PPO / Game Average PIM PPO / Game
AH 9.41 3.70 9.97 3.96 -0.55 -0.26
CCHA 12.72 4.47 14.04 4.07 -1.32 0.40
ECAC 12.93 5.03 14.38 3.81 -1.44 1.23
HE 10.98 4.74 11.48 4.45 -0.50 0.29
WCHA 13.28 3.87 14.20 3.44 -0.92 0.43
This would suggest the ECAC is the worst offender. The ECAC gives their league an average of 1.23 more PPO / game when playing a NC opponent. The only league that is not biased would be Atlantic Hockey which appears to hate their teams.
Ok, but is this any different from the way penalties are called in the league? "D" mean Delta in the header.
In Conference @ Home v NC Team On Road v NC Team @ Home v NC Team On Road v NC Team
PIM PPO PIM PPO PIM PPO D PIM D PPO D PIM D PPO
AH 10.96 3.89 9.41 3.70 11.43 3.79 -1.54 -0.19 2.02 0.10
CCHA 12.54 4.17 12.72 4.47 10.15 3.83 0.18 0.31 -2.57 -0.64
ECAC 12.82 4.17 12.93 5.03 13.71 4.19 0.12 0.86 0.78 -0.85
HE 12.87 4.37 10.98 4.74 16.56 3.58 -1.89 0.38 5.58 -1.16
WCHA 12.94 3.81 13.28 3.87 12.58 4.86 0.35 0.06 -0.70 1.00
This is a little weird. It suggests at home AH and HE schools are penalized less vs NC school whereas CCHA, ECAC and WCHA are penalized slightly more. All leagues except AH enjoy more PPO / game as the home team in NC games than they typically get in conference with ECAC being the worst again. As the visitor in NC games, CCHA and WCHA teams are penalized less than in conference, but HE, AH and ECAC are penalized more. The CCHA, ECAC and HE get fewer PPO / game on the road in NC games than in conference, but the WCHA and AH teams enjoy more PPO/game - including the WCHA that average 1 additional PPO / game as the visitor in NC games!
I'm sure there is some better analysis to be done (none of this takes the variance into account - how are PIM distributed anyway?!?!), but perhaps this is interesting to some people.
Here's some of the "raw" data (rows are visiting teams, columns are home teams):
164 Non-Conference Games Played
V/H AH CCHA ECAC HE WCHA Ind
AH - 11 19 7 2 -
CCHA 4 - 6 1 12 -
ECAC 7 8 1 16 10 -
HE - 2 9 - 9 -
WCHA 1 7 3 4 - 2
Ind 9 2 4 2 6 -
NC Visit Team PIM / Game
V/H AH CCHA ECAC HE WCHA Ind
AH - 14.36 10.21 12.14 9 -
CCHA 4 - 9 14 13.58 -
ECAC 9.86 16.13 - 10.75 18.1 -
HE - 12 21 - 16.67 -
WCHA 10 18.71 18.67 9.5 - 6
Ind 16 9 13 11 13.67 -
NC Home Team PIM / Game
V/H AH CCHA ECAC HE WCHA Ind
AH - 14.55 9.32 13.14 10 -
CCHA 7 - 10.17 12 11.5 -
ECAC 10.43 14.25 - 10.25 14.9 -
HE - 13.5 19.67 - 16 -
WCHA 10 15.29 16 9.5 - 20.5
Ind 10.22 6 9.5 10 14 -
NC Visit Team PPO / Game
V/H AH CCHA ECAC HE WCHA Ind
AH - 4.27 3.89 4 3 -
CCHA 3.25 - 3.33 5 3.75 -
ECAC 4.14 5.5 - 3.5 3.6 -
HE - 3.5 4.56 - 2.67 -
WCHA 5 4.57 4 4.75 - 6
Ind 3.44 2.5 3.25 5 4.17 -
NC Home Team PPO / Game
V/H AH CCHA ECAC HE WCHA Ind
AH - 4.55 4.42 3.57 2.5 -
CCHA 2 - 4 6 4.42 -
ECAC 4 4.88 5 4.13 3.8 -
HE - 3.5 - - 4.44 -
WCHA 5 5.43 8 4.5 - 3
Ind 3.78 4 3.75 5.5 4.17 -
Conference Games Played
V/H AH CCHA ECAC HE WCHA Ind
AH 65 - - - - -
CCHA - 72 - - - -
ECAC - - 48 - - -
HE - - - 56 - -
WCHA - - - - 82 -
Ind - - - - - -
Conf Visit Team PIM / Game
V/H AH CCHA ECAC HE WCHA Ind
AH 11.83 - - - - -
CCHA - 13.46 - - - -
ECAC - - 13.25 - - -
HE - - - 12.98 - -
WCHA - - - - 12.99 -
Ind - - - - - -
Conf Home Team PIM / Game
V/H AH CCHA ECAC HE WCHA Ind
AH 10.08 - - - - -
CCHA - 11.61 - - - -
ECAC - - 12.38 - - -
HE - - - 12.75 - -
WCHA - - - - 12.88 -
Ind - - - - - -
Conf Visit Team PPO / Game
V/H AH CCHA ECAC HE WCHA Ind
AH 3.68 - - - - -
CCHA - 4.01 - - - -
ECAC - - 3.94 - - -
HE - - - 4.27 - -
WCHA - - - - 3.63 -
Ind - - - - - -
Conf Home Team PPO / Game
V/H AH CCHA ECAC HE WCHA Ind
AH 4.09 - - - - -
CCHA - 4.32 - - - -
ECAC - - 4.4 - - -
HE - - - 4.46 - -
WCHA - - - - 3.98 -
Ind - - - - - -
A note on the raw data is that in EVERY league, the home team enjoys more PPO and fewer PIM!
Wow, too much time, but interesting.
Nice (and thanks for the explanations for those who had couldn't follow the data without a lot of study). Nate Silver, watch your rearview mirror. Someone is gaining on you.
Super cool! Thanks!!!
I think, and I could very well be wrong, that the AHA and ECAC use some of the same officials. Or at least they use to. What that might do to the conclusions I am not sure.
Quote from: MattSI think, and I could very well be wrong, that the AHA and ECAC use some of the same officials. Or at least they use to. What that might do to the conclusions I am not sure.
It makes things look even worse for the ECAC. Excluding AH-ECAC games from the analysis on the assumption the officials shouldn't have a league bias because they come from both leagues shifts the ECAC "apparent bias" to even more protecting - with an average of 1.4 additional PPO / game when an ECAC team plays at home against a NC opponent. In addition, AH appears to hate their teams even more. :)
Apparent Protection Bias
Average PIM PPO / Game
AH -0.93 -0.30
ECAC -1.58 1.40
Isn't part of the reason that AHA teams get more penalties at home NC games than the opponents get, the fact that the opponents are usually the better teams and do not need to hack as much?
Quote from: ursusminorIsn't part of the reason that AHA teams get more penalties at home NC games than the opponents get, the fact that the opponents are usually the better teams and do not need to hack as much?
There could be a similar effect due to home ice (correlating with "native" officials).
Thanks for the nice comments! Kinda fun to look at this and here's another take:
Each league appears to have a home team bias. Meaning home team is getting more PPO and less PIM per game. So, shouldn't this continue for NC games? Is it any worse when a NC opponents come to town?
In Leauge Home In-League Visitor In-League Home Protection Bias Home Team Tack-On Bias v NC
Average PIM PPO / Game Average PIM PPO / Game Average PIM PPO / Game Average PIM PPO / Game
AH 10.08 4.09 11.83 3.68 -1.75 0.41 1.20 -0.67
CCHA 11.61 4.32 13.46 4.01 -1.85 0.31 0.53 0.09
ECAC 12.38 4.4 13.25 3.94 -0.87 0.46 -0.57 0.77
HE 12.75 4.46 12.98 4.27 -0.23 0.19 -0.27 0.10
WCHA 12.88 3.98 12.99 3.63 -0.11 0.35 -0.81 0.08
By my calculation, this makes the ECAC stand out even worse. The home team bias nearly accounts for the NC bias in PPO/game in the CCHA (additional 0.09), HE (additional 0.10) and WCHA (additional 0.08), because those officials appear to just favor the home teams consistently in those leagues regardless if the visitor is NC. However, the ECAC averages an additional 0.77 PPO/game more for league teams against NC teams.
As @ursusminor pointed out, the discrepancy in AH bias might be more structural and related not to officials, but the talent level of the teams. This suggest that lumping together the different NC opponents isn't a very good thing to do, but small sample size makes analysis of individual pairs of conferences unreliable (for example, only 1 AH team hosted a WCHA team).
Edit: make clear the home and visitor columns are for in-league games. I think this is the analysis @Trotsky is suggesting in the previous post (this was started before I saw that post).
One possibility (that I have no real support for) could be that refs in some leagues more strictly enforce penalties than refs in other leagues. If the ECAC refs are less strict, ECAC players become conditioned to assume they will get away with more borderline actions. On the other hand, if WCHA refs are strict, WCHA players will focus more on avoiding penalties. As a result, ECAC team look less disciplined when playing WCHA teams with WCHA refs.
Quote from: jtn27One possibility (that I have no real support for) could be that refs in some leagues more strictly enforce penalties than refs in other leagues. If the ECAC refs are less strict, ECAC players become conditioned to assume they will get away with more borderline actions. On the other hand, if WCHA refs are strict, WCHA players will focus more on avoiding penalties. As a result, ECAC team look less disciplined when playing WCHA teams with WCHA refs.
The ECAC refs seem more random than the other leagues' refs in calling penalties.
Quote from: ursusminorQuote from: jtn27One possibility (that I have no real support for) could be that refs in some leagues more strictly enforce penalties than refs in other leagues. If the ECAC refs are less strict, ECAC players become conditioned to assume they will get away with more borderline actions. On the other hand, if WCHA refs are strict, WCHA players will focus more on avoiding penalties. As a result, ECAC team look less disciplined when playing WCHA teams with WCHA refs.
The ECAC refs seem more random than the other leagues' refs in calling penalties.
The ECAC refs seem more incompetent than the other leagues' refs in calling penalties
Quote from: ursusminorQuote from: jtn27One possibility (that I have no real support for) could be that refs in some leagues more strictly enforce penalties than refs in other leagues. If the ECAC refs are less strict, ECAC players become conditioned to assume they will get away with more borderline actions. On the other hand, if WCHA refs are strict, WCHA players will focus more on avoiding penalties. As a result, ECAC team look less disciplined when playing WCHA teams with WCHA refs.
The ECAC refs seem more random than the other leagues' refs in calling penalties.
Randomness by the refs could contribute to a more cavalier attitude by the players, I guess. If an ECAC ref only calls a certain type of penalty 60% of the time it occurs versus 90% by a ref in another league, the ECAC players would probably be more likely to commit the penalty because they think they can get away with it. (Although I'm sure css228 will tell me that I'm being generous by assuming ECAC refs get the call right more than 20% of the time.)