Economists parse the stats and find the correlation:
We find that college football games are associated with sharp increases in crime. For instance, assaults increase by about 9% when a community hosts a college football game, vandalism increases by about 18%, and DUIs increase by about 13%. We also find evidence that upsets result in larger increases in crime than games that do not produce an upset. For instance, an upset loss at home is associated with a 112% increase in assaults and a 61% increase in vandalism. We discuss these results in the context of psychological theories of fan aggression.
The most interesting aspect of the data, at least from the perspective of psychology, is the fact that upsets cause so much more violence. This implies that the fans aren’t simply imitating the spectacle of violence on the field. (See, for example, social learning theory.) Instead, their violence is triggered by a confluence of additional variables, one of which is their set of expectations. They are significantly more likely to act out once their expectations have been violated:
Although there is evidence that upset losses are associated with a larger increase in assaults than are upset wins, our results clearly indicate that expectations, and what happens to fans’ behavior when they are not met, should be explicitly built into future attempts to model the relationship between aggression and sporting events.
The best book I’ve read on the phenomenon of football/soccer and violence is Bill Buford’s classic text, Among the Thugs.