There's been a lot of complaining about the referees lately on Marquette messageboards. Certainly, there is a common expectation that MU is going to go into the Kohl Center in a few weeks down a few fouls just to start the game. This invariably begins the conversation regarding the foul discrepancy for Bo's teams, especially against Marquette. I decided to investigate the matter to see what the numbers contained.
The Data and Initial Results
Over the weekend, I pulled up box scores for UW games going back to the 2002-2003 season. The following data was compiled: Year, Opponent, Home/Away, UW Fouls, Opponnent Fouls, Wisconsin Score, Opponent Score, Conf/Non-Conf, Ranked/Non-Ranked. This generated secondary data of Wisconsin Win, Win Margin, Foul Disparity, and Total Points. Obviously, Foul Disparity is the key issue in question.
First, an initial pass of the data. UW averages just under 70 ppg, and their opponents average just under 60 ppg. The average number of fouls committed by Wisconsin was 15.5, and the average number of fouls committed by Opponents was 20.1. There is no question that, on average, Wisconsin commits fewer fouls than their opponents.
Analysis 1 - How important is Foul Disparity to UW Winning?
However, I wanted to see how relevant the Foul Disparity (FD) was on the likelihood of UW winning. Therefore, I ran a regression of the likelihood of Wisconsin winning against Home/Away, FD, Total Points, Conf/Non-Conf, and Ranked/Unranked. Only the Home/Away and Foul Disparity variables were significant, but the FD variable was significant!
95% confidence interval listed in parentheses
Intercept 0.55 (0.13, 0.97)
Home/Away 0.30 (0.18, 0.42)
FD 0.03 (0.02, 0.04)
In other words, Wisconsin is 55% likely to win the game just by being Wisconsin, they're 30% more likely to win the game if they play at the Kohl Center, and every Foul of Foul Disparity increases odds by 3% (holding all other variables constant). Better said, if Wisconsin's opponent has 5 more fouls, then Wisconsin is 15% more likely to win the game. A subsequent regression of Win on just FD and Home/Away had consistent results. This makes sense. Wisconsin's been pretty good, teams are more likely to win at home, and if their opponent has more fouls then they're more likely to win. So what?
Analysis 2 - Does Wisconsin receive a home court advantage on Foul Disparity?
If Wisconsin really does just play different defense, then the FD result should be consistent whether or not the team is home or away. Therefore, I ran a second regression of the Foul Disparity on Home vs. Away and Win Margin. Both variables were statistically significant. Again, with 95% confidence interval listed in parentheses.
Intercept 1.53 (0.54, 2.53)
Home/Away 3.34 (1.84, 4.85)
Win Margin 0.10 (0.05, 0.15)
Win Margin numbers say that for every 10 points Wisconsin wins by, that is likely to result in one additional foul for the opponent. Recall that UW wins games by an average of 10 ppg. But Holy Crap! The numbers are saying that Wisconsin starts with a 1.53 Foul Disparity advantage, but that increases by 3.34 Fouls if they play at home. This means that Wisconsin starts with a 5 Foul / game disparity when they play at the Kohl Center, or a 15% increase in their chances of winning. That is some good home cooking right there.
How has this played out for Marquette?
Year Foul Disparity Location Winner
02-03 -2 Bradley Center Marquette
03-04 11 Kohl Center Wisconsin
04-05 0 Bradley Center Marquette
05-06 12 Kohl Center Wisconsin
06-07 4 Bradley Center Wisconsin
These results are a nice little microcosm of the FD advantage when UW plays at the Kohl Center. At the BC, MU has managed to keep the foul disparity even. It's the problem of visiting the Kohl Center that has not turned out favorably.
Bottom line - Wisconsin does commit fewer fouls than their opponents, but that foul disparity is even more evident when UW plays at home. This does not bode well for Marquette on 12/8, meaning that it's not paranoia if someone is really out to get you!
Imagine how productive you could be if you applied this kind of effort to things other than MU basketball... ;)
Did you do the MSAE program? I'm in it right now. You should forward your results to Nourzad, he would absolutely love it. In fact, you probably already have.
Great work!
Wow, I'm very impressed.
This kind of analysis is something that the JS or even the MU Tribune should be doing for a very good interest piece.
Again, impressed. A very good read.
Being able to explain stuff like this is the reason I love econometrics. Nice job, Henry Sugar.
If you are/were an econ major, you make Clark, Nourzad and Toumanoff proud.
Quote from: marqptm on November 26, 2007, 03:03:14 PM
This kind of analysis is something that the JS or even the MU Tribune should be doing for a very good interest piece.
Ha! The tribune only does opinions when it comes to men's basketball. The day something even remotely close to this appears in the trib I will take them somewhat seriously.
Agreed though! Excellent work H.S.!
Unless you are making the outrageous claim that there is referee bias against Marquette as an institution, you should realize that what you're claiming is that Bo Ryan is a more forceful and effective coach than Crean.
Because you cannot honestly begrudge Ryan or Coach K for trying to gain advantage for their teams. At least I hope not.
I'd like to see what other teams cannot muster a decided advantage in foul calls in their home arena.
home court advantage is just that. I would almost expect that a team has a better foul disparity at home than anywhere else. The fact that UW does isn't really news to me as most of us knew it on some level anyhow - the analysis gives us a statistical argument to quite convincingly back up that feeling. A better comparison of foul disparity than home vs. away would be UW @ home vs. other teams of interest @ home (like MU/Duke/Pitt/etc.) and/or ncaa average as a whole @ home. This would give us an indication of how "fair" this home court advantage is for UW.
This is not to diminish the work you've done at all.. kudos for it as it is. I'm just thinking that a different type of apples/apples comparison would give us better grounds for being peeved. :D
Quote from: PuertoRicanNightmare on November 26, 2007, 03:37:48 PM
Unless you are making the outrageous claim that there is referee bias against Marquette as an institution, you should realize that what you're claiming is that Bo Ryan is a more forceful and effective coach than Crean.
Because you cannot honestly begrudge Ryan or Coach K for trying to gain advantage for their teams. At least I hope not.
I'd like to see what other teams cannot muster a decided advantage in foul calls in their home arena.
Once again you are projecting your dislike of Crean onto an unrelated topic.
Nowhere in that analysis was there any mention of how effective Crean is at gaining a home court advantage or a foul discrepancy. Duke or Coach K weren't mentioned at all. In fact, a comparison of Ryan and Crean was explicitly avoided.
Good for Bo Ryan leveraging the home court to his advantage. Maybe if someone is aware of it, the home court bias won't be as noticeable in the future (like say, 12/8)
Quote from: spiral97 on November 26, 2007, 03:43:05 PM
the analysis gives us a statistical argument to quite convincingly back up that feeling.
That's what I was going for. Note that not every team does manage to turn home court advantage into a foul disparity. Note the following link for a quick and dirty analysis (it's for Big Ten only - 2006)
http://www.spartantailgate.com/forums/msu-red-cedar-message-board/240032-there-up-date-foul-discrepancy-somewhere.html (http://www.spartantailgate.com/forums/msu-red-cedar-message-board/240032-there-up-date-foul-discrepancy-somewhere.html)
Wisconsin has a pretty high disparity in comparison to other Big Ten teams for Big Ten play.
Quote from: spiral97 on November 26, 2007, 03:43:05 PM
A better comparison of foul disparity than home vs. away would be UW @ home vs. other teams of interest @ home (like MU/Duke/Pitt/etc.) and/or ncaa average as a whole @ home.
That's a lot of data to collect. How about a regression of Foul Disparity on Conference / Non-Conference, Home / Away, Ranked / Unranked?
Intercept 1.3 (0.1, 2.5)
Conf / Non-Conf 2.0 (0.7, 3.3)
Home / Away 4.2 (2.8, 5.6)
Ranked / Unranked -1.9 (-3.8, 0.1)
In other words, UW starts with about a 1.3 FD advantage, they get an extra 2 fouls in discrepancy against Non-Conf opponents, 4.2 fouls for playing at home, and they lose 1.9 fouls when matched up against a ranked opponent.
Here's a question .. what about MU's Bradley Center FD advantage? Is there one? Is it statistically significantly different than UW's home FD advantage?
I think you'd need to see numbers from a few other teams to determine if it's UW that's getting a consistent advantage, or "any team that's playing at home" getting that advantage.
Quote from: Henry Sugar on November 26, 2007, 03:58:16 PM
Once again you are projecting your dislike of Crean onto an unrelated topic.
I'm simply pointing out that your vast receptacle of information on referee bias is nothing more than a Pandora's Box.
Quote from: Henry Sugar on November 26, 2007, 04:13:58 PM
Quote from: spiral97 on November 26, 2007, 03:43:05 PM
A better comparison of foul disparity than home vs. away would be UW @ home vs. other teams of interest @ home (like MU/Duke/Pitt/etc.) and/or ncaa average as a whole @ home.
That's a lot of data to collect.
Agreed but still the numbers that would give us any reason to be upset. Easy counter claim for UW is to say "every team gets about 5 more fouls in their favor when playing at home". Not sure what data you are collecting (just foul disparity for each game and game location or additional info too) but if it's not much then I would be happy to look up the numbers for a few teams and let you crunch them through.
Additionally, but separately, the difference in foul disparities experienced between home and away is what truly allows comparison between teams home advantage in terms of foul calling. Some teams have a style of play and/or proficiency of play that leads to more or fewer fouls being called against them (depending on how aggressive they are, how well they screen, how much their defensive mindset is to try and take charges, whether they are overmatched in terms of height, etc.). Most of those factors are nullified when comparing the deltas in foul disparity. In an ideal world, those deltas would be fairly consistent across the basketball landscape.
Quote from: PuertoRicanNightmare on November 26, 2007, 03:37:48 PM
Unless you are making the outrageous claim that there is referee bias against Marquette as an institution, you should realize that what you're claiming is that Bo Ryan is a more forceful and effective coach than Crean.
Because you cannot honestly begrudge Ryan or Coach K for trying to gain advantage for their teams. At least I hope not.
I'd like to see what other teams cannot muster a decided advantage in foul calls in their home arena.
An honest question here before this turns into a thread about Bo vs Crean.
PRN, can you please define what you mean by forceful and effective?
I completely understand Henry's line of logic and thinking because he shows the research and clearly defines all of his assumptions and conclusions.
I'm afraid your conclusion here is a little vague due to your terminology, or I'm just not understanding where you are going with this.
What do you mean by forceful and effective?
Interesting stuff Henry...here is some more from the web:
Purdue blog on Big 10 referee ratings...a few familiar names...and a bit funny on how they came up with all this:
http://www.goboilers.com/fanguide/refs.htm
Second is from Pomeroy on home cooking: Big difference on Big 10 and BE home winning percentage rank.
http://kenpom.com/blog/index.php/weblog/rabid_crowd_theory/
We can pretty easily conclude that referees/umpires are a major part of any sport (the pros know this and use all sort of means--whether via technology or supervisor analysis-- to grade each official for every game). A coach needs to plan for the ups or downs of a game (practice flops like Duke, foul low like Wisconsin, keep away from the hand, palms up like dcube, go to zone more on the road, to hand check or not to hand check, etc.). Conference refs call games differently even though teams play the same style--home or away: adjust, adapt, influence.
Bo Ryan's reaction to your findings (as if he doesn't already know!) (http://www.youtube.com/watch?v=1TajSyEm-Qg)
Nice analysis. The thing that I would like to add is that there are some things that home cooking does that might escape statistical analysis. The first one is that if you call two fouls one a teams best player early in the game, this will have a dramatic impact on the game (I think this happened to Wade the last time he was in Madison). Second, if a team goes into an environment where this happens, they might actually try to change their strategy. This is what worries me sine we are an aggressive defensive team. If you are Crean, I think you might tell a player like Jerel McNeal to play a bit more conservative early in the game so that he does not pick up two quick ones early in the game (I think this happened to McNeal last time in Madison).
Quote from: Pardner on November 26, 2007, 08:22:45 PM
Interesting stuff Henry...here is some more from the web:
Purdue blog on Big 10 referee ratings...a few familiar names...and a bit funny on how they came up with all this:
http://www.goboilers.com/fanguide/refs.htm
Second is from Pomeroy on home cooking: Big difference on Big 10 and BE home winning percentage rank.
http://kenpom.com/blog/index.php/weblog/rabid_crowd_theory/
Thanks for sharing, Pardner. The GB.c stuff is hilarious, and it's refreshing to see that there are other people on the Interwebs that waste time.
Pomeroy's analysis is interesting as well. Seems to me that he really seems to discount any "on-court" advantage.
Finally just wanted to say thanks to folks that provided some critical feedback. I realize that the analysis needs further investigation and is still incomplete.
Quote from: 1990Warrior on November 27, 2007, 07:57:58 AM
Nice analysis. The thing that I would like to add is that there are some things that home cooking does that might escape statistical analysis. The first one is that if you call two fouls one a teams best player early in the game, this will have a dramatic impact on the game (I think this happened to Wade the last time he was in Madison). Second, if a team goes into an environment where this happens, they might actually try to change their strategy. This is what worries me sine we are an aggressive defensive team. If you are Crean, I think you might tell a player like Jerel McNeal to play a bit more conservative early in the game so that he does not pick up two quick ones early in the game (I think this happened to McNeal last time in Madison).
Great point... obviously that's the "game inside the game" type of stuff. There are a ton of tiny variables in this research (strategy, score, tempo, style of play, etc. etc.)
The larger the sample size, the more these small variables will be minimized (ie if you looked at 25 years worth of data)
Overall, this is still great information and does show that MU fans aren't just being paranoid... UW does seem to have a greater foul ratio at home. Why they have it is a whole different debate that includes a ton of those little variables... but one of them certainly has to be the fact that it's simply "The Kohl Center" and they play well there.
Good analysis. I'd also like to point out that maybe there are more fouls called on the opponent at the Kohl Center, where the Badgers win a ton, because the opponent was losing down the stretch and needed to "foul" try and catch up. That's gotta be good for at least a couple fouls per game. You won't see as big a discrepancy on the road if the Badgers since they tend to lose more on the road, and you'd expect them to foul when losing as much as they get fouled when winning.