With the Super Bowl approaching, prepare for the perennial barrage of news stories predicting a spike in the number of men who beat their spouses or children on that day. Although studies have disproven that it actually happens, the concept of heightened violence on Super Sunday remains an urban legend.

But the link between watching football -- specifically college football -- and violence may not be a myth. A new study by researchers at the University of Colorado at Denver examines whether assaults and other forms of aggressive behavior increase when major college football teams play home games, and finds that they do. More strikingly, perhaps, incidences surge most when upsets occur -- whether the home team wins or loses.

Daniel I. Rees, an associate professor of economics at Colorado-Denver who conducted the study with a student, Kevin T. Schnepel, says that their inquiry (as part of a course on the economics of crime) began by looking into a potential connection between alcohol sales and violence at college games, but quickly discovered that that issue had become largely moot as many if not most had banned alcohol from their stadiums. Anecdotal reports like the one about Super Bowl-related violence and led them to explore existing literature on fan aggression, most of which produced "very inconclusive" results, says Rees.

So the two decided to examine the link between college football games and crime. For their study, they collected six years' worth of police data for five crimes (incidences of assault and vandalism, and arrests for disorderly conduct, driving under the influence, and alcohol law violations) from 26 towns that are home to universities that play NCAA Division I-A football. They compared the number of those crimes that were reported on days when home football games were played and found a significant uptick.

Wouldn't that be expected, since most college football games are played on Saturdays? Don't Saturdays naturally have more assaults, since people are less likely to be working and more likely to have leisure time on their hands with which to get into trouble? "Yes, but there is enough variation in when college games are played now, on Wednesday and Thursdays and what-not, that we can control for the day of the week and distinguish between a 'Saturday effect' and a real increase," Rees says.

But wouldn't you expect more crime and/or violence if you've got tens of thousands of people who wouldn't normally be there streaming into Clemson, S.C., or State College, Pa., for a Clemson or Penn State home game? "Yes, you can naturally expect that with the extra people, extra crowding, assaults are going to go up," he says.

That's where the outcomes of the games come in, the researchers say. If increases in the number of assaults or other aggressive acts were due solely to heightened numbers of people in college towns -- and in close proximity to one another -- incidences of the crimes would be the same whether the teams won, lost or tied.

When the researchers crunched the data, though, they found that the likelihood of assaults increased by 112 percent when the home team suffered an upset loss, defined as when an unranked team beat a ranked team or a lower-ranked squad defeated a higher-ranked one. Perhaps most surprisingly, there was a 36 percent boost in assaults on Saturdays when the home team won in an upset. For a typical police department in the study, the increases would amount to nearly 6.7 extra assaults after an upset loss, and 2.2 more after a surprise triumph. A spike also occurred in arrests for disorderly conduct.

The fact that you see these results tied to an upset, which a priori no one could have known was going to happen, suggests that the outcome of a game had some impact on assaults, instead of the entire effect being from there just being more people," Rees says.

What might explain why college football games appear to produce more violence in their communities? No one theory can explain the outcome fully, he says. The "social learning theory," which posits that fans act violently because they see violence on the field and mimic it, wouldn't explain the difference based on the outcome of games. And it's not just that fans don't like it, and are more likely to get angry and aggressive, when their teams lose, since the data show that "an upset win causes more violence than a loss," Rees notes.

Instead, the outcome seems "somehow to be tied in with expectations, with how fans react to upsets," Rees says, suggesting that future studies of fan aggression might be able to build on this study.

As for implications for colleges that play big-time football, the study suggests yet another layer of complexity to any calculation of the financial pros and cons of a university's sponsorship of big-time sports. An institution might put its ticket revenues and sponsorship money and vending sales on one side of the ledger and its employee and energy costs, cleaning crews and other expenditures on the other.

"But where do you account for the fact," says Rees, the economist, "that the football program is turning around and imposing six extra assaults every other fall Saturday on the community, which is bearing the cost in terms of crime?"