Frances Woolley is a professor of economics at Carleton University
The video game NHL 10 is programmed to make fighting worthwhile. According to producer David Littman, “if you're down 3-0 in a playoff series and you're getting killed, or if your team isn't playing well, you can start a fight and it gets all the energy back for your team. Your team's energy goes back to 100 per cent.”
But in the real world, are teams that win fights more likely to win games? In a forthcoming paper in the Journal of Sports Economics, economists Benjamin Leard and Joanne Doyle use detailed, game-level data from the 2007/08 NHL season to estimate the impact of fighting on the probability of winning.
It’s not easy to do. Teams that fight might be less likely to win because losing causes brawling, that is, players start fights when they’re behind and they’re frustrated. Leard and Doyle eliminate this problem in two ways. First, they focus on first and second period fights, ones that might be expected to build momentum and influence the course of a game. Second, to isolate the impact of fighting, they control for other factors that influence the probability of victory: differences between the teams in the number of power plays, the number of face-offs won, the ratio of shots on goal to the number of possessions, goalie save percentages, and home ice advantage.
Leard and Doyle get their information on fights from two sources. First, they use statistics from dropyourgloves.com to calculate the number of punches thrown and landed by fighters from each team. The difference between the two teams’ number of punches thrown (or landed) is then used as a measure of fighting success. The researchers find that punching is no way to win hockey games: a team that throws or lands more punches has no significant advantage over its opponent.
The second measure comes from hockeyfights.com, where viewers watch fight videos and vote on who won. When Leard and Boyle measure victory using hockeyfights.com results, there is some suggestion that winning fights gives a team an advantage. My reading of their results is that winning a fight might have no impact on the probability of victory, or it might increase in the probability of success by two percentage points. The sample of 2007/08 games with first and second period fights is too small to decide between these two alternatives.
Leard and Doyle place more emphasis on the punches thrown and landed measures than the hockeyfights.com results, because they argue that punch counts are more objective than Web polls. This leads them to conclude that “fighting success does not appear to have any measurable effect on the probability of winning.”
Leard and Doyle’s results are unlikely to convince those who see fighting as part of the code, a way to police the ice. Yet players risk their future well-being when they step up and take one for the team. As a hockey mom I can’t but ask: for what?
Follow Economy Lab on twitter