Last weekend I finally got around to watching The Last Gladiators, a documentary about the enforcers, “goons”, in ice hockey. The story follows that of Chris Nilan, and showcases the sometimes devastating price of being a “tough guy”. If you’re interested, and I really think you should be, the documentary is available on Netflix.
I’m not going to analyze the content, partly because it’s already been done so well: Jouni Nieminen wrote about the documentary for Helsingin Sanomat (unfortunately in Finnish, but scroll down for links for English coverage). But the documentary got me thinking. Fighting is one of those topics that pop up every season, at least once a season. And it’s one of those things that almost everyone who has an opinion feels strongly about, one way or the other. There’s sites and blogs promoting fighting, such as hockeyfights.com where you even get to vote on the winner of the fight. And there’s sites and blogs against fights, such as It’s Not Part Of The Game which tries to rid the game of fighting, arguing that, well, it’s not part of the game.
I have to insert a disclaimer here: While I’m the last person to be called “a flower-hat lady” (from the Finnish word “kukkahattutäti”, meaning a schoolmarmish Miss Judy goody two shoes) I am against fighting when it’s just for the sake of fighting. Because why would you fight? There’s no point!
And I can just hear the uproar… The arguments for fighting are seeded deep in hockey tradition. “Fighting helps the team win more; it motivates the team.” “Fighting cleans the game by policing out the dirty play.” The two most common arguments heard in favor of the fights. Too bad they’re not true.
How much are we fighting, then?
The lovely people at hockeyfight.com keep good statistics on fights in the National Hockey League. So let’s see how much fighting happens in the NHL. The following figures are based on the data from the 2000-2001 up to and including the current season, unless otherwise stated.
- Were you to watch every game in a single season, you’d see, on average, 640 fights. (This average excludes the shortened 2012-13 and current season.)
- On average, 39% of games in a season have at least one fight.
- 132 games, on average, in a season, have more than one fight. That’s approximately 10% of the games. (Excluding both the 12-13 and current season.)
- The average number of fights per game range from 0.35 (current season) to 0.65 (2001-2002). There seems to be a declining trend:
* The current season, 2014-15, statistics are recorded at 231 games played.
Do fights win games?
In their study The effect of Home Advantage, Momentum, and Fighting on Winning in the National Hockey League (Journal of Sports Economics, 12, 5, 2011) Benjamin Leard and Joanne M. Doyle used data for fights in the 2007-2008 season. (Leard and Doyle used dropyourgloves.com for their data, another fan-based site like hockeyfights.com. The numbers are slightly different between the sites. Hockeyfights.com counts fights where at least one player involved receives a fighting major. I couldn’t find a definition for a fight at dropyourgloves.com.) During that season 41% of games had at least one fight taking place. Out of the total of 722 fights, almost half took place in the first period.
This means, should this distribution between periods hold for other seasons as well and there’s really no reason to assume otherwise, that approximately 44% of fights happen in the first period. And around 70% of fights take place in the first two periods combined.
Leard and Doyle studied the effect of a won fight on the outcome of the game using game-level data. They acknowledged the possible endogeneity bias between fighting an game success as doing poorly in the game could trigger a fight. Of course, given the fact that only 25% of the fights took place after the second period, that doesn’t seem a very prominent reason for fighting.
To minimize the possible bias Leard and Doyle used only the games where there had been fights only in the first period. Turns out, there is “no evidence of the motivational effect of winning fights”. They also experimented with different slices of the data; using both first and second period fights made very little difference. Overall, they conclude, there is “no significant evidence, positive or negative, of winning fights on the likelihood of winning the game”. (Emphasis mine.)
So, fighting makes no difference to the outcome of the game. Why do it then? As The Last Gladiators made so plain, it’s hardly a healthy way to play hockey.
Does fighting clean up the game?
If fighting doesn’t help us win, then it must at the very least help us protect our players, right? If we put a known enforcer, a bad-ass fighter on the ice, the other team will think twice before landing a dirty hit on our superstar. Right?
Well, it doesn’t look like it. Adam Gretz at Regressing wrote about this a year ago. Going over two seasons worth of games he looked at hits that “resulted in a suspension, fine, or match penalty” and compared that with whether or not the team taking the hit had an enforcer in the lineup that night. Out of the 106 incidents, Gretz found that 52 of the teams did not have a fighter in the lineup. That means 54 teams did, and the hit still took place.
SkinnyFish (really? Okay.) over at Pension Plan Puppets studied the number of “non-obstruction penalties”, or “naughty behavior” as he calls it, drawn by a team with respect to fighting majors taken. His argument was, that a team that fights a lot would draw less penalties, that is, other teams would play cleaner hockey against them than against teams that don’t fight. He found out there’s no correlation between fights and the level of “naughtiness” that goes on on the ice. (Nice jab at Toronto in the graph, though.) Having a fighter present doesn’t influence the level of dirty play you get from the other team.
So we’re not winning more games with fights, nor can we use them as a deterrent to foul play. And I ask again: why bother?
Are fights untouchable?
I’m beginning to wonder, are fights an isolated aspect of hockey, something that isn’t affecting much anything, but also something that isn’t affected by much anything. Jac C. Heckelman and Andrew J. Yates, in their ingeniously titled study And a Hockey Game Broke Out: Crime and Punishment in the NHL (Economic Inquiry, Oct 2003, 41, 4), used the natural experiment of NHL exploring the possibilities of using two referees in 1999-2000 seasons to study how the number of referees effects both the number of penalties called and the number of rules infractions committed by players.
Borrowing from the economic theory of crime, they isolate two different mechanisms: the monitoring effect and the deterrent effect. Monitoring effect would simply mean that two referees catch more infractions than one, while the actual number of infractions is the same. Deterrent effect on the other hand suggests that the increased risk of getting caught keeps players from committing the infraction in the first place.
The theory suggests, that “when the number of referees is increased, there are differential effects on major and minor infractions. … For major infractions, there is only a reaction effect to the other teams committing fewer minor infractions (due to the other team’s own deterrent effect).”
85% of the total number of penalties were minor. And as was to be expected, a significant monitoring effect was found. Adding a second referee means more infractions are caught. However, no deterrent effect was found for minor, nor major, infractions. Fights aren’t the type of infractions one usually tries to hide, quite contrary, so the monitoring effect doesn’t apply to them. But there was no deterrent effect found either. What does affect them?
Heckelman and Yates argue that infractions are more “crimes of passion” than conscious decisions, and as such are born out of the heat of the game, which would account for the lack of deterrent effect. Either way, it doesn’t seem like fights are motivated by many the variables they studied: offensive and defensive prowess, relative physical size of the teams, power play track records of the teams to account for the costs of getting a penalty..? Only opponent shooting percentage was significant (positive effect), own team power play defense had a significant and negative effect, presence of goons in teams was significant and positive, as was the age difference (teams with relatively older players fought more), number of times the own team had been shorthanded in the past 15 games had a positive and significant effect (implies a track record of rough game). Looking at the results, and besides that fact that if you put fighters on the ice, they’re going to fight, none of the above really explains fighting.
It won’t clean up the game. It won’t help you win more games.
So why fight?