November 17, 2009

NFL and domestic violence?


In their latest NBER paper, David Card and Gordon Dahl tried to explore one of the most perplexing forms of violent behavior--violence by men against their partners and children. Two common interpretations are put forth by economists: family violence may be interpreted as instrumental behavior; family violence may on the other hand arise simply because men like to commit violence and women tolerate it in return for higher transfers. Ouch. Well, Card and Dahl offers an alternative: family violence simply "represents expressive behavior that is triggered when conflictual interactions escalate out of control." Their key hypothesis is:

"[T]he risk of violence is affected by the gain-loss utility associated with salient emotional cues. Specifically, [they] assume that negative cues--based on realizations relative to expected outcomes--increase the risk of violence."

So how did Card and Dahl test their hypothesis? They look at the effects of National Football League (NFL) game outcomes. They analyzed data from states that are home to six different NFL teams, focusing on the effect of wins and losses of the home team on police-reported family violence incidents. But why focus on NFL? Two things:

"On the one hand, many fans feel a strong emptional attachment to local teams. Home games on Sunday afternoons typically attract a quarter of the television audience in markets with a local NFL team. Team affinity is reinforced by extensive pre- and post-game coverage by local news and entertainment media. On the other hand, the existence of a well-organized betting market allow us to infer the expected outcome of each game. Assuming that a game outcome is random, conditional on the pre-game [Las Vegas] point spread, we can interpret any difference between the rate of family violence following a win or loss as a causal effect of the outcome of the game. Moreover, we can easily define the strength of the emotional cue associated with a win or a loss in a given game."

Four interesting findings are found:

1. An "upset loss" by the home team (a loss when the team was predicted to win by more than 3 points) leads to an 8% increase in the number of police reports of at-home male-on-female intimate partner violence (this is after controlling for location fixed effects, the pre-game point spread, sesonality and weather factors, and the size of the local viewing audience).

2. An "upset win" (a win when the team was expected to lose by more than 3 points) have at most a small effect in reducing family violence.

3. Upset losses in more salient games (those involving a traditional rival, or occurring when the home team is still in playoff contention) have a larger effect on the rate of intimate partner violence, as do unexpected losses after games involving an unusual number of sacks, turnovers, or penalties.

4. NFL game outcomes have little or no effect on police reports of female-on-male intimate partner violence.

All in all, Card and Dahl are finding that:

"[T]he strong impact of random external factors on the rate of family violence provides compelling evidence that at least some portion of family violence arises through a breakdown of control, rather than as instrumental behavior driven by purely rational choice."

I know it seems that this new research somewhat points to a [small] negative impact of the the greatest sport in America. But look on the bright side--coaches can use this study in their pre-game motivational speech to their team: "Listen guys!!! We're expected to win this one. If you don't want to be responsible for thousands of domestic violence out there, you better shape up and deliver!!!"

Go figure.

By the way, which NFL teams are in the study? Carolina Panthers, Denver Broncos, Detroit Lions, Kansas City Chiefs, New England Patriots, and Tennessee Titans. Well, at least we won't be seeing spikes in domestic violence after Lions or Chiefs games.