An examination of the Berkeley Police Department’s stop data reveals significant disparities between stops involving white and Black individuals.

As racism in policing is called into question nationally, BPD has also been challenged for racial disparities in its stop data. Experts and the department, however, caution against being too quick to attribute the divides to racial bias on the part of the officers.

Since BPD began publishing stop data in January 2015, more than 58,000 stops have been recorded, as of press time.

White and Black people share nearly identical proportions of these stops, at 35% and 34%, respectively. Black people, however, account for only about 8% of the overall Berkeley population, while white people make up about 59% — a 51 percentage point disparity despite only a 1 percentage point difference in stop shares.

The disparities also track when stops are broken down by call type. For example, Black people account for 50% of all suspicious vehicle stops, while white people make up only about 22%.

A 2018 study of BPD stop data published by the Center for Policing Equity, or CPE, found similar disparities but warned that estimating bias by comparing a group’s population share in the city with its share of police stops fails to account for the portion of those interactions that are with individuals who are not city residents.

“It ‘points to’ racial profiling, but is not conclusive evidence of it,” said UC Berkeley public policy professor and CPE researcher Jack Glaser in an email.

Between 2016 and 2019, about 66% of all BPD traffic violations and 44% of arrests were incidents where the subject was not a Berkeley resident, according to Capt. Rico Rolleri from BPD’s Professional Standards Division.

According to Glaser, the rates at which illegal substances or weapons are discovered during a stop are a far better indicator of whether bias is present. If one demographic is frequently searched but not arrested, this may suggest that the group is often stopped without good cause.

While the data points tracked by BPD do not include specifics about whether a person was found to have contraband or a weapon, they do show that during traffic stops, Black people were more likely to be searched than any other demographic, as almost 19% of Black people stopped were searched. White people, on the other hand, were searched during only about 5% of traffic stops.

Of those stopped, Black people were cited or arrested for a violation about 30% of the time, while white individuals were cited or arrested more than half of the time. The rest of these traffic stops resulted in a warning.

“This is strongly suggestive of bias,” Glaser said in the email. “One has to consider that if POC are being searched more that they could be presenting valid indicators of suspicion at a higher rate; but if their searches are less likely to yield contraband or weapons, then those searches are less justified.”

According to Glaser, when minority drivers are issued a warning more often, it suggests that these stops may be “pretextual,” meaning that they may be used by officers primarily to justify a search.

Glaser added, however, that these higher warning rates may also reflect an effort by officers to reduce economic burden on marginalized communities by not ticketing or arresting them.

Rolleri attributed many of the disparities shown by the data to a lack of needed context in the current reporting system, which only requires officers to enter six data points after completing a stop.

“It’s just six characters that an officer types in after a stop,” Rolleri said. “It doesn’t take into account things like time of day or whether it was dark out.”

According to Rolleri, this lack of context has spurred BPD to work to implement new standards established by AB 953, a 2015 law concerning police bias, earlier than required. The law creates a system with 10 additional data points for officers to fill out after a stop, which Rolleri believes will better contextualize each incident.

Rolleri also said BPD is working toward reducing disparities with officer training.

Berkeley police officers are required to attend training sessions, including “Fair and Impartial Policing” and “Principled Policing” programs, which address implicit bias.

“I would challenge anyone to find a police department that treats people more fairly than the Berkeley Police Department,” Rolleri said. “We have a culture here of treating people with dignity and respect.”

According to Rolleri, BPD has hosted the 12-hour “Fair and Impartial Policing” training six times since 2010, with the most recent session taking place in 2016. The last time the department held the “Principled Policing” training was in 2018.

Due to turnover rates, however, only 67 officers who have completed these training sessions remain on the roster, and 103 current officers have not taken them. According to Rolleri, however, all officers are trained on implicit bias, as well as fair and impartial policing, in the academy.

Departments often look to training to mitigate bias, but according to Glaser, evidence suggests that it does not work because the psychology involved is difficult to permanently alter.

He does, however, support educating supervisors on bias, as they can incentivize practices such as frequent stops and searches, which sometimes result in biased policing.

“We train our officers to deal with and make good, solid enforcement stops based on reasonable suspicion and probable cause,” Rolleri said. “Police officers are human beings, and humans can make mistakes. That doesn’t mean every mistake is malicious or bias-based.”