The full story, with graphics and charts, can be read here

On Sept. 6, 2013, four plainclothes San Francisco police officers approached a 55-year-old African American man as he parked in front of his home in the Bayview district. They ordered the driver and his friend out of the car and onto the sidewalk, according to a grievance the man filed with the city. The driver complied, handing the officers his license and registration.

After checking his documents, officers asked to search his vehicle, according to the driver, whose identity was withheld by the city’s Office of Citizen Complaints. He consented, and officers found a small amount of marijuana both in the vehicle and in the passenger’s backpack. The police did not cite either of the men, telling the passenger they were looking for weapons. None was recovered.

In his complaint, the driver said police never explained why he had been stopped and alleged that his consent for the search had been coerced, though he did not say exactly how. He also accused police of targeting him because of his race. One of the officers interviewed disputed the claim, telling investigators the search was predicated on incorrect information from a dispatcher that the car might have been stolen.

None of the allegations was sustained by the Office of Citizen Complaints.

‘Red flag for bias’

This stop and search, and the dispute about whether police acted properly, are emblematic of issues raised by a Chronicle examination of three years of data on traffic stops in San Francisco.

From the beginning of 2013 to December 2015, SFPD officers chose to search black and Latino drivers at much higher rates than whites or Asians after traffic stops, either by invoking probable cause or requesting the driver’s consent. Those searches, however, were much less likely to result in officers uncovering evidence of crimes than less frequent searches of white or Asian drivers.

Police say that factors including where they patrol and the demographics of criminal suspects are at the root of those racial disparities, but many social scientists, criminologists and civil rights advocates believe they signal racial profiling. The higher search rates of blacks and Latinos suggest police are more inclined to search drivers of those races, they say, and a lower evidence recovery, or “hit” rate, implies those searches are more often unwarranted.

“A lower hit rate for ethnic minorities is a red flag for bias,” said Lorie Fridell, an associate professor of criminology at the University of South Florida who participated in a previous analysis of SFPD stop data in 2007. “It’s not conclusive, because nothing is, but it is certainly a statistic that should lead to questions.”

Importance of stop data

Because traffic stops are the source of most interactions between the police and the public, they have become a growing area of study and an impetus for police reforms.

In Rhode Island, racial disparities in traffic stop data prompted the state to revamp its police training policies in the mid-2000s. Connecticut recently began tracking a variety of detailed metrics on police stops and searches. And in Oakland, achieving racial parity in stops and searches is one of the last hurdles that city’s Police Department must clear before being freed from more than a decade of federal oversight.

The California Legislature passed a bill last year requiring all police departments in the state to collect traffic stop data and send it to a newly created branch of the attorney general’s office starting in 2019, where it will be analyzed.

Neither San Francisco Police Chief Greg Suhr nor any deputy chief was made available for comment on The Chronicle’s findings. But other SFPD officials attributed racial disparities in stops and searches to higher crime rates, and subsequently a heavier police presence, in minority communities, as well as the disproportionate number of crime suspects identified as black and Latino.

Cmdr. Ann Mannix, who heads SFPD’s traffic division, said officers do not racially profile drivers.

“I don’t say, ‘There’s a white guy, I’m going to go pull them over. There’s an African American guy, I’m going to search them,’” she said. “I pull someone over because their registration is expired, they’re going too fast, they’ve got a headlight out. And then, if I see something else, I probe further.”

Critics, however, say the data raise concerns about racial bias at a time when police-community relations in San Francisco have grown tense, in part because of the officer-involved shooting deaths of Mario Woods and Alex Nieto, and derogatory text messages shared by SFPD officers.

Last week, Public Defender Jeff Adachi asked California Attorney General Kamala Harris to open a civil rights investigation of the SFPD. The American Civil Liberties Union of Northern California has been lobbying the U.S. Department of Justice to open an investigation of the department for two months.

“It reflects systemic racism and bias in dealing with black people,” Amos Brown, president of San Francisco’s NAACP chapter, said of The Chronicle’s findings. “We’ve been working with the Police Commission, Chief Greg Suhr, the mayor’s office to improve (community-police relations), but it’s still a major issue that needs to be addressed.”

Seeking driver’s permission

San Francisco police made almost 320,000 traffic or suspicious-vehicle stops from 2013 to 2015. The vast majority of drivers were given a warning or ticket and sent on their way, but cars were searched in about 12,700 of those stops — roughly 4 percent, a rate close to the national average.

Justifications for a search after a traffic stop are expansive. If someone smells like marijuana, their car can be searched. If they are on probation or parole, they can be searched. If police have reason to believe the driver is armed, officers can pat them down for weapons.

A police tactic often questioned by critics is a consent search, which requires asking a driver’s permission. Unlike probable cause searches, which require police to articulate some evidence of a possible crime, consent searches can be based on just an officer’s hunch. A 2008 research paper from Boston University said consent allows “an open-ended search with virtually no limits.”

Due to the degree of officer discretion involved, many observers, including the ACLU, consider consent searches a reliable barometer for discerning bias — both conscious and subconscious.

“Consent searches are what people complain about when they say police just pulled them over for no reason,” said Supervisor Malia Cohen, who represents the Bayview district and sponsored legislation last year requiring the city to more thoroughly analyze and report its traffic stop data. “There’s a compelling argument that they’re obsolete, unfair and incredibly biased.”

San Francisco’s top prosecutor and public defender both have reservations about consent searches. District Attorney George Gascón, formerly the city’s chief of police, said such searches are useful for law enforcement but that some drivers who are stopped do not realize they can decline an officer’s search request. Adachi said his office’s clients often say their consent was coerced — something that is difficult to prove in court.

“Too often we hear that a (defendant) said, ‘No, I don’t agree to a search,’”Adachi said. “And the officer still went into their trunk, searched the floorboards of the car, searched the glove compartment.”

SFPD reported just under 1,800 consent searches from 2013 to December 2015, or about a dozen per week. By comparison, police in Charlotte, N.C., roughly the same size as San Francisco, conducted 2,500 consent searches last year alone. About 37 percent of all vehicle searches in Connecticut were conducted with the driver’s consent from Oct. 1, 2013, to March 31, 2015, a proportion more than twice as high as San Francisco’s.

Despite the low volume, there was a large racial divide in whom police asked to search. Over that three-year span, African American motorists were eight times more likely to be searched with consent than white drivers after a traffic stop. Latino drivers were searched at almost four times the rate of white drivers.

Crucial tool for police

In all consent searches during that period, police found criminal evidence less than 13 percent of the time. Just 85 of the individuals searched were ultimately arrested; 46 percent of those arrested were African American.

Experts say the low search-recovery and arrest rates suggest that consent searches may not be the best way to detect crime. “Within the span of a couple years, thousands of people were subjected to a humiliating ritual,” said Janice Nadler, a law professor at Chicago’s Northwestern University. “And I’m guessing many of those people were innocent.”

Certain law enforcement agencies and jurisdictions have temporarily or permanently banned consent searches after class-action lawsuits or constitutional challenges. The California Highway Patrol has not conducted them in more than a decade. The ACLU suggested SFPD end the practice in 2002, after that organization’s analysis of stop data uncovered similar racial disparities.

Law enforcement experts maintain that consent searches are crucial for catching criminals in the absence of tangible physical evidence.

“Consent searches give us the ability to do our jobs efficiently and go on to the next call for service,” said Mike Durant, president of the Peace Officers Research Association of California, a statewide lobbying group. “We don’t always have the ability to completely articulate” reasonable suspicion.

Training on stops, profiling

While consent searches can involve a high level of discretion, SFPD officials say new hires receive ample training on fair and impartial policing. Recruit officers go through a four-hour workshop on racial profiling at the police academy, 12 hours of search and seizure training, and a city-ordered 40 hours of “cultural immersion” training, where they meet with leaders of various minority communities.

Veteran officers also return to the academy every few years to receive additional instruction on racial profiling, which covers correct use of probable cause for stopping and searching suspects.

San Francisco police officials attribute racial disparities in stops and searches largely to higher crime and a heavier police presence in minority communities. The disproportionate number of crime suspects identified as black and Latino is another factor that could affect the overall numbers, they say.

Because traffic stop-and-search data can be influenced by such social factors, many experts believe the most telling sign of potential racial profiling isn’t whom police choose to search. Instead, they focus on the outcome of searches, particularly those predicated on reasonable suspicion or probable cause, also referred to as evidence-based searches.

From 2013 to 2015, SFPD conducted 3,450 evidence-based searches after traffic stops, many based on officers’ reporting the potential sight or smell of drugs, alcohol or weapons. Like consent searches, most took place in the Bayview district, downtown and the Mission.

Black drivers were searched at a rate more than four times higher than white drivers after being pulled over, according to SFPD data. Latino motorists were searched almost 2½ times as often as white drivers.

Results with whites, Asians

Despite the high search rates, officers found guns, drugs or other contraband just 32 percent of the time after an evidence-based search of African American motorists, and 41 percent of the time when the driver was Latino. By comparison, when police searched white drivers, evidence was recovered 74 percent of the time; the recovery rate for Asian drivers was almost 80 percent.

“It implies that when police search whites and Asians, they’re pretty darn sure they’re going to find something,” Florida criminologist Fridell said. But the data suggest “there’s a wider net being cast and a lower level of proof (required) before initiating a search of African Americans and Latinos.”

While racial disparities in traffic stops might seem minor compared with serious incidents of police misconduct, they can have a profound influence on community-police relations. “That sort of intrusion is very meaningful in a person’s life,” said University of Kansas Professor Charles Epp, an expert on investigatory police stops. “It leads to suspicion of the police and to stories about disrespectful treatment.”

District Attorney Gascón, who put together a task force to investigate bias within the SFPD last year, said the impact of racially skewed stops and searches reverberates throughout the criminal justice system, and they challenge the notion of San Francisco as a progressive city.

“If a significant segment of the community believes the system is rigged, that’s going to impact every portion of our work,” he said. “Because we’re a progressive city, we often fall under a false sense of security that we can’t have the problems other places do when it comes to racially biased policing.”

Seeking to rebuild trust

Last year, Supervisor Cohen sponsored legislation requiring the Police Department to release reports on its traffic stop data. Despite the fact it has collected the information for more than a decade, the city seldom analyzed it, past reporting by The Chronicle has shown.

The ordinance was one of many measures geared toward rebuilding public trust in the Police Department after what Cohen characterized as a tumultuous few years.

“Bias exists within our law enforcement agencies, particularly SFPD, and these data about stops and searches is another example of a justice system that’s failing,” she said, while noting that many of the department’s problems are caused by a handful of officers. “We are bringing some tools to the table to help rectify historic past wrongs. ... We have a rare opportunity to make major tweaks to a department that’s falling short.”