In the debate over the police tactic known as stop-and-frisk, both sides agree there's nothing inherently wrong with officers stopping more black and Hispanic than white residents, at least in cities where violent crime is concentrated in minority neighborhoods.
The question is: At what point does stopping a disproportionate number of minorities cross the line into illegal, race-based policing?
When does a legitimate, proactive tactic become the wholesale harassment of communities?
Determining those limits has been one of the more controversial topics in big-city law enforcement - and a question that often has gone before the courts to be answered.
In Philadelphia, where police embraced an aggressive stop-and-frisk policy nearly three years ago in response to rising gun violence, the debate is about to begin in earnest.
Civil rights lawyers filed a federal lawsuit earlier this month, arguing that Philadelphia police have been targeting people based solely on race.
In 2009, officers stopped 253,333 pedestrians, 72 percent of whom were African American, the suit says.
But David Rudovsky, one of the lawyers who filed the suit, acknowledges that racial profiling can't be proved "solely by the numbers."
"The key question is: If you control for factors like police deployment and crime rates, would that explain the disparity?" he said.
Rudovsky said a better benchmark would be to look at the "hit rate" - how often the stops result in arrest or the discovery of a weapon or other contraband.
The lawsuit says just 8 percent of the stops in 2009 resulted in arrests, often for "criminal conduct that was entirely independent from the supposed reason for the stop." Beyond the overall number of stops by race in 2009 and the arrest percentage, Rudovsky and his cocounsel lack data to search for patterns and figure out precisely who gets frisked and why.
As part of a 1996 settlement of another civil rights lawsuit, the police agreed to provide such information and allow the monitoring of vehicle and pedestrian stops. Rudovsky and other monitors looked at hit rates through 2005.
"It turned out, interestingly, that the hit rate was higher for . . . whites," he said.
The current suit - filed by Rudovsky's firm, Kairys, Rudovsky, Messing & Feinberg, and the American Civil Liberties Union of Pennsylvania - details the experiences of eight African American and Hispanic plaintiffs who say they were repeatedly stopped by police without legitimate reason.
The attorneys, however, do not have extensive police data on pedestrian stops since 2005. They hope to collect the data through the lawsuit, and Rudovsky said he expected to seek an expert report analyzing the information.
Academic studies published on stop-and-frisk data in Los Angeles in 2008 and New York City earlier this year found that minorities were being stopped unconstitutionally.
But both studies were following city-commissioned reports that found no pattern of race-based policing. The Los Angeles study was commissioned by the ACLU of Southern California. The New York report was prepared for the Center for Constitutional Rights, which has a federal lawsuit pending against the New York Police Department.
Jerry Ratcliffe, a Temple University professor of criminal justice who advises the Police Department, said arrest data are not the best way to examine stop-and-frisk.
The tactic, in which officers can stop and frisk pedestrians as long as there is "reasonable suspicion" of illegal activity, was designed to drive down the number of illegal guns on the street.
Officers want to "increase the sense of certainty of police intervention if you carry a gun," Ratcliffe said.
Fewer arrests and seizures of weapons could indicate the tactic is working, he said.
Homicide and other violent crime have decreased since 2008, when Mayor Nutter took office and declared a "crime emergency."
Nutter and Police Commissioner Charles H. Ramsey have credited stop-and-frisk with helping to drive down violent crime.
"If done legally and with some sensitivity to being targeted to the areas that need it the most . . . the research shows it is effective," Ratcliffe said.
The problem, Rudovsky said, "is very few of these stops have anything to do with violent crime." Typically, he said, they are more mundane interactions, such as police stopping "the three kids standing on a corner."
When monitors examined data from 1996 to 2005, Rudovsky said, they found about 20 percent of the stops were done without sufficient cause. He suspects that pattern continues.
The suit asks the court to order more police training, supervision, and monitoring.
City attorneys have not yet filed their response to the lawsuit, and police officials have declined to comment on the pending litigation.
Ratcliffe noted there was "no easy way to reduce violent crime" in a way that doesn't inconvenience those who live in high-crime areas.
"How much inconvenience are we prepared to tolerate for being safer? Do we want to be stopped by police a couple times a year, or do we want to increase the chance of our kids being shot?" he asked. "Unfortunately, that's a question that has to be answered by a minority of the population."