Vulgar Statistics: Crowd Atmosphere Matters

This topic is back.  I have given it new life just so I can kill it so personally and so emphatically that it will never be back again.  That’s right, I’m here to talk about crowd noise and booing and how that factors in to home ice advantage.  We’ve already had this debate, and I’ve already given my opinion, as has Dan.  The crux of it is that since the fans want better hockey, and that booing doesn’t bring better hockey, that booing is stupid.

Of course this leads to numerous justifications, that the fans deserve a good on ice product (wrong!  especially since tickets are so cheap), that the fans can do what they want (well sure, but that makes complaining about bad hockey seem kind of stupid since the fans are helping cause it), and that the players just need to sack up and impress us, or that booing doesn’t matter anyway.  About those last two…

This is where I start to get angry and do things like spending over a month compiling roughly one hundred excel tabs worth of home ice data for the past eight seasons. But my German Shepherd-like focus is not alone.  A quick Google search reveals many studies linked to the effects of home crowds.

Zeller and Jurkovac (1988) noted that while professional baseball teams as a whole won 7.2% more games at home than away, teams whose home stadiums were domed won 10.5% more games at home than away. This difference was attributed to the noise level difference between normal and domed stadia, as the domed stadia made the crowd support more prevalent by holding in the noise.  (Lerra 2003)

One potential source of bias by referees is social pressure (influence of the crowd).  Dohmen (2008) finds that architectural conditions play a key role in the refereeing bias observed, namely: the size of the crowd (absolute size), the attendance-to-capacity ratio (relative size) and the proximity of supporters to the pitch (the presence of a running track). He finds that there is more added time in close matches when the crowd is physically close to the field of play. Also, home teams are significantly more likely to be awarded a disputed penalty, with the physical distance between the crowd and the playing field important to this decision. Petersson-Lidbom and Priks (2007) find similar results for Italian football following the Italian government’s decision to enforce clubs with sub-standard stadiums to play home games behind closed doors.  (Dawson and Dobson 2008)

Ryan Boyko, a research assistant in the Department of Psychology in the Faculty of Arts and Sciences at Harvard University, studied 5,000 English Premier League games from 1992 to 2006, to discern any officiating bias and the influence of home crowds. The data was published in the Journal of Sports Sciences suggested that for every additional 10,000 people attending, home team advantage increased by 0.1 goals.

Sports Illustrated, in a 17 January 2011 report, reported that home crowds, rigor of travel for visiting teams, scheduling, and unique home field characteristics, were not factors in giving home teams an advantage. The journal concluded that it was favorable treatment by game officials and referees that conferred advantages on home teams. Sports Illustrated stated that sports officials are unwittingly and psychologically influenced by home crowds and the influence is significant enough to effect the outcomes of sporting events in favor of the home team.

While that last one seems to debunk my argument, it actually reinforces it.  Even if you believe that the players should be (or are) tough enough to blot out any influence the crowd might have on them, the home team still sees an advantage linked to crowd noise.  That’s all well and good, but it’s also very general.  What of the product that we have before us now, professional hockey in the post-(04-05) lockout era?

This debate suffers because it is impossible to quantifiably measure positive crowd support other than soliciting a deluge of personal anecdotes.  While selling out might not directly translate into overall crowd atmosphere, it’s about as close a connection as we can get.  More on that later.

This journey begins with understanding home field advantage, and where the averages lie.  From 2005-2006 until the past season, the NHL has had an average home points percentage of .611, and an average home winning percentage of .553 versus a .506 points and .447 winning percentages on the road.  That means that the average NHL team will earn 20.6% more of it’s points at home than on the road, or 60% of it’s points at home and 40% of its points on the road, roughly.  To add to that, the average NHL team will score 3.2 goals per game at home, and allow 2.9.  Obviously the best teams do much better than this, and the worst teams do much worse.

The best team in the aforementioned time span is the Calgary Flames, seeing a 38% increase in points percentage while at home, or amassing 69% of their points at home and 31% on the road.  The worst team in that same span is the Boston Bruins, sitting at +6.3%, or 53% / 47% roughly.  Both do well at the gate (98.68% attendance for the flames and 93.69% for the Bruins), and while you can certainly claim that Flames crowds are more positively boisterous than the long jilted Bruins fans, you can’t really prove it.  Make what you will of stereotypically happy and passionate Winnipeg (+36.8%) and Minnesota (+34%) being #2 and #3 and stereotypically negative and passionate Buffalo (+11.2%) and Philadelphia (+7.4%) being #28 and #29 respectively.  There was no correlation between performing well at home and selling out.  That isn’t necessarily damning though.  As back-to-back laden Buffalo fans are well aware, all schedules are not created equal.  Full spreadsheet is available here.

Goals provide a much better means of approaching this problem as they allow for a larger sample size, taking into account to what degree games are won and lost.  NHL teams are, on average, +100 goals (goals scored minus goals allowed) at home, and -100 goals on the road.  These numbers were a little off on my sheet because I scaled Winnipeg and Atlanta to allow for direct comparisons to the rest of the NHL.  Winnipeg (+402, almost certainly inflated due to the small sample size), Calgary (+306), and Tampa Bay (+248) took the top honors, and Edmonton (+117), Philadelphia (+110), and Toronto (+75) took the bottom spots.  While there was no correlation between goal differential and attendance, there was a slight correlation between goals scored and attendance.  This certainly makes sense as a larger and (presumably) louder crowd will bring energy and increase the offensive output of both teams, but there is little that can be done to confirm this outside of taking a decibel meter to each arena hundreds of times.  It also makes sense that the correlation will be small as there are numerous other external factors that influence an NHL game besides crowd size and noise.  That there is any correlation is, I think, telling.  Full spreadsheet is available here.

The goal is, in knowing that multiple things make up a home ice advantage, to attempt to nullify the ones that aren’t related to our purposes, those being fatigue from traveling and familiarity with a given arena.  One of the easier ways to do this is to look at overtime games.  In match-ups that go to extra time, both teams are fatigued, and the focus has been on the game itself long enough for any uncomfortability due to a lack of arena familiarity to have dissipated.  The expectation would be, given limited influences on the home ice advantage, that the overtime winning percentage in that span would be pretty close to .500, or at least significantly lower than the NHL average of .553.  It is .543.  Full spreadsheet is available here.

But there is another way to nullify travel fatigue.  Many times during the course of an NHL season, a team will play an away / home back-to-back.  Since the home team in that second game had to travel the night previously, and the away team obviously had to travel to reach their away game, this (mostly) eliminates travel fatigue as a variable.  One would expect that points percentage to be lower than the NHL average of .611, and it is at .586, but I’d wager that the difference is smaller than most fans expect.  Unfortunately in that span, the sample size is too small to read into the figures for any one team with Chicago playing the most away / home back-to-backs (59) of any team, and Edmonton (4) playing the fewest.

We can also mete out arena familiarity by isolating divisional match-ups from that data set.  Aside from rookies and a few new players, the majority of NHL rosters are comfortable with the cities, the arenas, and probably even the flights to their most-played opponents.  Given this, one would expect the points percentage to decrease even further as we are eliminating (at least a portion of) the travel fatigue and arena familiarity aspects of home ice advantage.

Since 2005-2006, home teams in divisional games in which both teams traveled the night previously have a winning percentage of .640.  Unlike previous parts of this study, there seems to be a very direct correlation between filling up the building and winning games.

I made my sellout cutoff 99% because I’ve found ESPN.com and Wikipedia arena size listings to be slightly off at times.  In any event, the 99% < X <100% games accounted for maybe a half dozen out of hundreds.  They had little to no impact on the overall results.

Right now you’re probably pointing out that the above numbers are the case because good teams sell out more, and good teams also win at home more. That’s a reasonable assumption so I investigated to see if that was the case.

And while I do not find the following chart to be particularly useful because I think we grow dangerously close to splitting the data too many ways, it nevertheless reinforces the point.

Full spreadsheet available here.

Where does that leave us?
Obviously we know that teams play better at home and from the above we can attribute that (generally) to scoring more at home. We know that there are multiple variables that factor into home ice advantage and we know that crowd noise is definitely one of them. Though we can’t provide a quantifiable link between crowd positivity and on ice performance, enough of a correlation exists between crowd noise itself and performance that it’s certainly within the realm of believability.

If you don’t think crowd noise has an effect on performance, you’re wrong.

If you don’t think crowd noise has a significant effect on performance, you’re wrong.

If you don’t think booing has a negative effect (or conversely a positive crowd has a positive effect), you’re probably wrong.  (Again this will probably never be proven, but I for one am not willing to go against the direction roughly 9,000 NHL games is trending.)

Of course, as a free citizen, you’re perfectly able to boo if you want, but the truth is that booing brings worse hockey and by extension makes your experience that much more miserable.  Congratulations.