Internet Explorer 11 is not supported

For optimal browsing, we recommend Chrome, Firefox or Safari browsers.

Oakland Overstates Reductions in Crime Due to Faulty Data

The Bay Area city had recently touted a 33 percent year-over-year decrease in crime. But a review of police data found that the city overstated the improvements and has been using incomplete information for years.

In recent months it seemed that Oakland, Calif. — a city famously grappling with a rise in crime — had course-corrected in remarkable fashion.

Statistics published by the police department showed year-to-year crime had dropped by 33 percent overall by the end of April, a dramatic shift after last year's spike. Mayor Sheng Thao praised the city for turning a corner. Gov. Gavin Newsom quoted the 33 percent figure in a news release touting Oakland's partnership with the California Highway Patrol. The news spread through social media posts and optimistic headlines, including stories in the Chronicle.

But a new Chronicle review of Oakland police data finds that the city overstated the improvements actually seen on the streets. More troubling, the analysis found a persistent problem in the Citywide Weekly Crime Reports published by Oakland police, which compare incomplete year-to-date figures from the current year to complete year-to-date figures from past years.

As a result, these counts inevitably, and at all times, create the impression that Oakland's crime trends — up or down — are better than reality.

Given currently available data, it isn't possible to know just how much the 33 percent figure overestimated Oakland's overall reported crime drop. A Chronicle analysis of Oakland's underestimates in previous years suggests that though overall crime was almost certainly down in Oakland through April, that reduction may fall to 20 percent, possibly less, when the data is fully updated.

The Chronicle's reporting follows on the work of Oakland resident Timothy Gardner, who exposed the problem in his Substack newsletter, the Oakland Report.

"My motivation is simply to stop the misinformation," Gardner said in an interview.

Indeed, at a moment when public safety is one of Oakland's most divisive political issues, driving efforts to recall the mayor and the Alameda County district attorney, voters do not have access to a clear picture of crime rates in their neighborhoods.

While acknowledging the flawed comparison and percentage changes in the reports, police blame technology constraints and say they are trying to balance the public's appetite for crime data with their duty to be as accurate and transparent as possible.

They include a footnote in each weekly report, explaining that the data may be partially counted: "Statistics can be affected by late reporting, the geocoding process, or the reclassification or unfounding of crimes. Because crime reporting and data entry can run behind, all crimes may not be recorded."

Police officials, though, do little to correct the record when politicians and media outlets cite the inaccurate data and trends.

Asked about the faulty data, Thao's office provided a statement. "I appreciate OPD's efforts to provide weekly transparent crime data to the community," the mayor said. Her office declined to answer additional questions about the issue.

A spokesperson for Newsom said the office had cited data from the police and would not comment beyond that.

The Oakland Police Department's year-to-date statistics for homicides and violent crimes are mostly accurate. But the reports only partially count burglaries and other nonviolent crimes, which are a lower priority, generally not investigated right away and usually compiled from reports filed online by victims.

Those online reports need to be verified and imported into a records management system, causing a lag of up to six weeks, police said.

"There can be a delay in the overall number of reported property crimes based on our community's use of the online reporting system," an OPD spokesperson said in a statement. "Each of these online crime reports must be reviewed and verified by a member of our staff before being included in crime statistics."

The impact of the backlog can be seen in the crime report published after the week of July 1-7. It tallied 17,969 incidents of what are known as "Part 1" crimes for the year — murders, assaults, robberies, rape, burglary, theft, auto theft and arson. That was an apparent 33 percent plunge from last year's count of 26,903 crimes through July 7.

However, the corresponding report, published one year earlier, had different 2023 figures: It logged 24,244 crimes as of July 9. That's because the count was not yet complete.

Year-end tallies for 2023 were closer to accurate, percentage-wise, but were still incorrect due to the undercounting of crimes like burglary and theft. Similarly, the year-end report for 2022 showed that overall crime had risen by 7 percent, though a complete report published a year later documented the actual rise of 20 percent.

Spokespeople for the police department insisted that other jurisdictions experience similar delays caused by online reporting, and attach "comparable disclaimers" to their data.

Yet a Chronicle review of crime data from San Francisco, Berkeley, San Mateo, Fremont, Hayward and Vallejo showed that other cities either do not release year-to-date comparisons, or have lags that are less severe and include more obvious disclaimers.

Lt. Barry Donelan, head of Oakland's burglary and general crimes detail, said the department is not trying to deliberately mislead the public. Rather, police staff are hobbled by "decrepit IT infrastructure" and different software packages that do not communicate with each other efficiently. A report management system issued to Oakland police in 2006 has yet to see an upgrade, he said.

"Do we know there are shortcomings in the numbers? Yes," Donelan said. "Is there a desire among professional law enforcement to fix that? Oh, yes."

Over the past few years, more Oakland crime victims have used the online portal to file reports, and the old computer systems haven't kept pace, Donelan said. It's frustrating not only for residents and policymakers who want accurate information, he said, but also for police who need real-time data so they can strategically deploy officers.

Apart from the weekly reports, the year-end crime data that Oakland submitted to the state Department of Justice several months ago appears to have a major error.

It states that in 2023, aggravated assaults more than tripled compared with 2022 — from 3,329 to 11,169 — after remaining steady in prior years. No one in the Oakland Police Department could explain the implausible rise. In the city's year-end crime report, assaults rose from 3,222 in 2022 to 3,531 in 2023.

Rik Belew, a computer scientist who worked for years to make OPD's data more accessible to the public, views the unreliable weekly reports as a snapshot of widespread deficiencies in technology and record-keeping.

Inaccurate reporting from the police force has many downstream effects, Belew said. Real estate sites rely on the data to create crime maps for prospective home buyers. Residents and merchants lack key information about when and where crime happens. People are deprived of metrics to evaluate the police chief and other top law enforcement officials.

Nonetheless, the weekly report format is deeply entrenched, and some critics of the technology still find value in the data, as do police supervisors, Donelan said.

Crime is going down in many categories, including burglaries, as police devote more resources toward investigating them, he said. That's led to more arrests and prosecutions.

Whether the department will upgrade its software anytime soon is unclear.



(c)2024 the San Francisco Chronicle. Distributed by Tribune Content Agency, LLC.
TNS
TNS delivers daily news service and syndicated premium content to more than 2,000 media and digital information publishers.