But Catalano says these results might result from the data analysis used. Herms’ data combines Longwood and Cambridge campus statistics.

The institutions in the StalCommPol study are all private four-year universities located in a large city with 25 percent or more students in on-campus housing.

StalCommPol used statistics on violent crime that every campus police department is obligated to report to the Department of Education every year under the Clery Act.

Violent crime includes murder, non-negligent manslaughter, forcible rape, robbery and assault with a weapon.  The study does not include criminal offenses which took place within an on-campus residence hall because Herms says he prefers to focus on street crimes.

“These are the crimes that the FBI categorizes as violent crimes when they do their crime index,” Herms says.

Catalano, however, says he was not able to replicate StalCommPol’s data analysis until he met with students from the group in early March after which, Catalano says, Herms changed the information on his website.

But Catalano maintains that Herms’ inclusion of Longwood data leads to misleading results.

Herms’ crime statistics include incidents that occurred on “public” property, which Catalano defined as areas contiguous to and surrounding Harvard-owned property.

“When 99 percent of crime on Longwood occurs on public property, then what you get is increased crime rate for Harvard as a whole,” Catalano says.

To illustrate the discrepancies, Catalano says that when HUPD compares crime rates from Harvard’s Cambridge campus and Penn’s campus, excluding public property for both, Harvard had 34.4 violent crimes per 10,000 students, faculty and staff and Penn had 49.2 per 10,000 on average annually, based on data from 2000-2002.

When public property is included in these figures, Harvard had 96.1 violent crimes per 10,000 and Penn had 87 violent crimes per 10,000 based on data from the same period.

Though Catalano says grouping the public and the on-campus crime together is misleading, he says that disclosing the crimes that occur on public property is still important to HUPD because it raises awareness about Harvard’s surroundings.

“We draw our borders fairly liberally,” Catalano says. ” And we feel that’s appropriate. We want to avoid the appearance at all costs of hiding crime or downplaying crime. We just don’t do it. We could be more selective in the way in which we define public property but we don’t feel that’s appropriate.”

A source familiar with StalCommPol’s research describes it as “very sketchy.”

“It’s like he chooses the numbers he likes and twists them. He likes to interpret them in ways that are damaging to Harvard,” the source says.