Oakland’s crime data mess: How politicians are playing shell games with public safety

Cherry-picked stats and untested reforms have left Oakland’s public safety in disarray.

Photo by Levi Meir Clancy on Unsplash

Author Notes

• This opinion was first published in a shortened form in the SF Standard.

• This article was updated at 11:00 pm on 19 July 2024 as described in footnote 2.

Oakland politicians have a simple playbook on crime: When crime is rising, they wave it off as a broader state or national trend, or they blame it on the previous administration. When crime goes down, they take full credit. They close with a virtuous spiel on violence prevention and policing alternatives as the future of a more just public safety system.

But this playbook is not grounded in data about the reality on our streets or law enforcement. Politicians and elected officials cherry-pick incomplete data that suit their political objectives. Much of the public accepts their spin at face value.

If the electeds did take a deep look, they'd find out the data were full of issues that could impair correct interpretation, because data gathering, management and curation are difficult undertakings, and they are systemically underfunded.

Instead, they base public safety policy on political expediency, which in recent years has disrupted and dismantled the resources needed for fair and just policing. 

Take the Oakland Police Department’s crime-reporting system, which is more than twenty years old. It’s so degraded that the department can no longer use it to centrally track police investigation documents. Instead, employees use Microsoft Word, saving files on desktop folders and sharing by email.

That system is supposed to automatically transfer uniform crime reporting data to the state for its annual crime compilation. But that too is broken. The state recently released 2023 data showing 11,169 aggravated assaults in Oakland in 2023—a 3.4-fold increase over 2022. But that’s an error. OPD’s own report shows about 3,500 aggravated assaults.¹ The state data is flawed due to an unresolved bug in data transmission discovered a month ago.

Police sources tell us that OPD is now “beta testing” a new $12 million computer-aided dispatch system for 911 operations, funded by a state grant. The new system has a records module that could replace OPD’s current system, but police can’t use it because it won’t communicate with the myriad of archaic OPD data and reporting systems required by local, federal and state oversight policies.

As record-keeping gets worse, it invokes a downward spiral. Trust plummets as a skeptical public demands transparency but does not get it. OPD is inundated with more than 10,000 public records requests each year, which are fulfilled by the same overtaxed staff—primarily its investigative staff of about 40 officers and administrators (excluding the homicide division) who are also tackling a caseload of over 40,000 crimes.

Poor records and technology infrastructure are just the tip of the iceberg. In the past decade, policy changes and decisions have upended Oakland's policing. These include reducing police stops by about 80%, limiting criminal pursuits, intensifying oversight, limiting crowd control practices and closing the local jail. In parallel, the city reduced police staffing to the bare minimum required by law and may go further. 

One of these policies—the pursuit policy limiting the speed and type of chase—is not inherently a bad policy. It aims to protect the lives of innocent people who are frequently harmed by pursuits. These tragedies also cost the city tens of millions of dollars in lawsuits—a cost cut in half with the new policy.

However, this policy has consequences; it means non-violent and most violent criminals escape arrest. If we curtail this tool of law enforcement, we also need to replace it with effective technologies, strategies and resources to track and intercept offenders without a direct pursuit. But the city has not done that. So, in essence, they have issued a ticket to freedom for criminals.

The city has made all these changes like an uncontrolled clinical trial using Oakland residents as guinea pigs—except this “trial” has no principled design, no measurement of outcomes and no guardrails to ensure the experiment does no harm. In 2020, a small but prescient group of individuals foresaw this; they implored the Re-imagining Public Safety Task Force not to replace proven policing methods until first proving that alternatives work.

It’s clear the experiment is not working. The state of lawlessness is so severe that Oakland has become a destination for crime tourism.

Adding to the pain, the police department has been under a federal monitor’s control for more than two decades, the longest federal oversight of any department in the country. 

Its 50+ requirements have fundamentally reshaped the department's operation. Officers are constantly scrutinized by the monitor, an investigator general, a police commission, the Community Police Review Agency, a media machine, a public primed to expect misconduct, and by those who may profit from maintaining the public perception of misconduct. 

The department was just three months from ending its oversight when Mayor Sheng Thao fired Chief LeRonne Armstrong in February 2023. Subsequently, an independent arbiter cleared the chief of wrongdoing. But the oversight continues.

Many in the city believe in hopeful theories about police and criminal justice reform—theories based on comforting labels like “violence prevention” and “police alternatives,” or on shame-inducing phrases like “mass incarceration” and “police militarization.” But these are words, not data, nor evidence, nor rigorous analysis of effectiveness. 

Controlled studies of Oakland’s Department of Violence Prevention (previously called “Oakland Unite”) showed that the department’s practices did not reduce arrests, convictions, probation, or victimization of program participants.² Yet the city has generously expanded its funding since, without any demand for performance accountability.

The argument for these alternative policing programs is that they better serve the communities that are hurt by the criminal justice system. However, this approach disregards the impact of unchecked crime on those same communities. Analyses done in 2013 and 2017 found that 67% to 78% of Oakland’s homicide victims are Black. The city’s own surveys show that Black residents, those most severely impacted by crime, prioritize policing and police data transparency more than any other demographic.

Certainly, there are deep, inscrutable sociological factors contributing to crime that deserve to be addressed with public action. But those factors aren’t unique to Oakland. Oakland’s crime problem is much worse than in nearly any other city in the country because we don’t enforce the law.

Criminals will exploit that gap in law enforcement no matter how much we spend on alternatives.

As a community, we chose this path because we voted for the leaders and ballot measures that implemented those policies. 

We can have a safe Oakland, but we have to choose it.

1. Note that undercounting of crimes in Oakland due to tabulation error has been less than 5% in recent years, and generally less than 3%, for violent crimes. So the final, fully-tabulated data for year-end 2023 would likely be at-most about 3700 crimes. Certainly nothing close to 11,169 reported in the state data.

2. This paragraph originally stated: “Oakland’s Department of Violence Prevention (previously called “Oakland Unite”) has produced only one report on its efficacy in its 20-year history.” A reader alerted us that this is incorrect—there are multiple studies on DVP effectiveness. Accordingly, the paragraph was revised in the body of the article.

For context and completeness, we also describe here the results of these studies and why they were commissioned.

Measure Z requires an annual independent performance evaluation to assess if “violence prevention/intervention programs and strategies are progressing toward the desired outcomes”. Accordingly, the city commissioned a consulting firm, Mathematica, to conduct a four year evaluation study from 2016 to 2020. They produced a series of reports on the outcomes achieved by the Department of Violence Prevention, including the comprehensive report on Life Coaching originally noted in this article, and a summary report on the cumulative findings for multiple intervention programs.

Only one of these reports—the Life Coaching report—was published by the DVP on its website as of the time of this article. And there have been no new evaluations published after 2020 on the DVP website, nor is this author aware of any further studies of program outcomes, despite the requirements for such studies annually.

In these reports, four DVP programs were evaluated from 2016-2020. Two programs were studied for impact on violent crime using matched-comparison groups to control for confounding factors. Both the un-served comparison groups and the served intervention groups were evaluated at 6, 12, 30 months for arrests, convictions, probation, school enrollment, school behavior, and truancy. Sample size was 3001 individuals through the full 30 month study.

In summary, the findings were:

• Life Coaching reduced violent crime arrests by 1-3% at 6 months, but had no impact on the rate of arrests, convictions, gun offenses, or probation after 12 months or 30 months. Those participants who were also participating in Ceasefire (operated by OPD in partnership with DVP) did show a statistically significant reduction in convictions, gun offenses, and probation, but not arrests.

• Employment and Education Support Services reduced violent crime arrests by 6% at 6 months, but had no impact on arrests, convictions, nor victimization after 12 months. The program was not evaluated at 30 months. The program improved enrollment in school by 13%, but youth participants who were enrolled in school had high rates of chronic absence and violent school incidents.

• Commercially Sexually Exploited Youth Intervention was assessed only for program implementation, but was not evaluated for outcomes.

• Shooting and Homicide Response Program participants, who were recent victims of gun violence, avoided re-injury. But it is unknown if that lack of injury was the result of intervention. The program did not include un-served comparison groups which would be needed to make such a determination.

In addition to the Mathematica studies, one report was published in 2013 by a different firm, Resource Development Associates. It was not a controlled study. It tracked participants in the violence prevention programs during the 2012-2013 fiscal year. The analysis suggested a reduction in arrests among DVP program participants. But the results were subject to selection bias because they did not have comparison groups. As noted in one of the Mathematica reports, the lack of comparison groups leads to erroneous results:

”Few studies of similar models, such as intensive case management, have found that the programs led to reductions in arrest rates. For instance, in an analysis of multiple studies of intensive case management programs for people with mental illness, the only studies that found a decrease in arrest rates did not include comparison groups (Loveland and Boyle 2007).”

Thus, the more recent and rigorously controlled study conducted by Mathematica over a 4-year period serves as the most definitive assessment of Oakland’s violence prevention programs to date. As noted in both the original and updated version of this article, the study showed that the DVP programs did not lead to a reduction in violent crime one year or more after intervention.