The counted: Inside the search for the real number of police killings in the US
If a police officer fatally shoots an unarmed citizen in the United States – and it was happening on average more than twice a day even before the killing of Michael Brown seven months ago in Ferguson, Missouri – most people find out about it not from law enforcement but from the 24-hour news cycle.
The best counts America currently has of killings by police are the work of activists and journalists – online databases like Fatal Encounters , Facebook compilations like Killed By Police , or Operation Ghetto Storm , which estimates that one African American is killed by police, security guards or vigilantes at least “every 28 hours”.
Citizen activists keep the best national counts. But there are corners of the country where the police track use-of-force more closely than any outside activist could. For example if an officer intentionally fires a gun in Montgomery County, Maryland – even if no one is hit or hurt – the police department posts a detailed description of the circumstances on its home page, usually within 24 hours. It is policy.
But transparent police departments are by no means the norm: the United States has no uniform count of people killed by police officers. The problem of missing data stems from more than just police obstructionism or oversight. A national infrastructure for data collection has never been built. Instead a confusing mosaic of city, state and county reporting leaves too many cracks for data to fall through, without imposing consequences for local police failing to report when they kill those they are sworn to protect.
“The reality is that there is not a good national data – or even a regional, state or local data on officer-involved shootings,” Laurie Robinson, a professor of criminology at George Mason University and co-chair of Barack Obama’s task force on community policing, told the Guardian.
In step with Obama’s call for better data in policing, and a public outcry for information about police use-of-force in general, policing leaders are increasingly admitting that carelessness, defensiveness – or, in the worst cases, indifference – does play a role in the failure of the country to track officer-related deaths.
A justice department investigation of the FBI’s published statistics has already revealed the worst from a data standpoint: more than half the people killed by local and state law enforcement officers in the US went uncounted in the country’s most authoritative crime statistics every year, for almost a decade.
But now, a Guardian review of how that local patchwork gets put together – including conversations with dozens of current and former police officers, policing organizations, academics and reformers – shows that non-reporting by police is only part of the problem. As many have noted before, the sheer number of local law enforcement agencies – “we’ve got 18,000”, Obama himself said at the White House earlier this month – presents a significant data standardization challenge. It is supplemented by questions at every turn over what a national database of people killed by police would look like, and how it should be built.
In talking about the complexities of the problem, longtime observers and innovative number-crunchers share a quiet hope that a revolution in US crime statistics is closer than ever before, as anger and mistrust turn to reform and technology.
“What we will have – hopefully in the next three years – is a robust database in the major cities of all police shootings,” David Klinger a University of Missouri criminologist, said. “Then we can say, ‘You know what, federal government, here’s what’s doable, now pass a law to fund this.’”
A good-cop theory tests a national number – with ‘heroic’ transparency
The Montgomery County police department is authorized to employ 1,265 sworn officers and as one of the biggest departments in the country, it serves a suburban Washington DC community of more than a million people with a large nonwhite population.
In some ways, Montgomery County is like a lot of police departments across the United States, but in one respect the department is unusual: it is committed to communicating about its work, even when that work takes an especially difficult turn.
Over the seven years of 2007-2013, Montgomery County had 29 incidents in which officers discharged their firearms. The department averaged 488 reports of use-of-force by officers for those years, out of about 216,000 annual calls for service. The department registered nine officer-involved deaths, and counted four “other in-custody deaths,” with all male decedents. Six of those who died were black, four were white, two were Hispanic and one was Asian.
It is department policy to make all this information public. In addition to updates on its web site, which is mobile-compatible, the Montgomery County police department issues regular press advisories; posts to Facebook and Twitter; and publishes weekly crime summaries that are compiled into yearly reports, giving the community an accurate picture of local crime and inviting analysts to draw comparisons with other jurisdictions.
“Transparency is really important,” said assistant chief Luther Reynolds. “The idea is that as a public servant, as a public agency, we have a constituency, and they have needs. And part of that need is data.”
Not all police chiefs are so committed to transparency though. D Brian Burghart, editor of the Reno News & Review, runs Fatal Encounters , a growing online portal built in part on more than 2,000 requests for public records from local and state law enforcement agencies. In Burghart’s extensive experience, police typically are not eager to supply numbers for officer-involved deaths.
“They put you off. They build obstacles,” Burghart said. “In fact they will violate public records law to prevent you from getting this information. You often have to remind them what the law is.”
Any particular department’s disposition to transparency is important because police are not required to report to the federal government when they kill someone. As the FBI emphasizes, reporting is voluntary. The autonomy of local police in this regard is thoroughly American.
“Unlike countries like the UK, where there is a centralized justice system, we have a federal system,” said Robinson, of Obama’s task force. “Law enforcement, and the justice system in fact, is traditionally and historically something handled by state and local government.”
It remains completely unclear what proportion of local law enforcement agencies tell the federal government when they kill someone, something that is a rare occurrence in most departments but an annual recurrence in larger ones.
To report deaths to the FBI, police must first conceive of the death as an offense, which is not a given. Then, officers must fill out paperwork in the uniform crime reporting system to tally the death, and then fill out extra paperwork to minimally describe it. This additonal paperwork, known as a supplementary homicide report , is filed in 89% of homicide cases overall, according to a December 2013 study by the bureau of justice statistics (BJS).
In many cases, however, the extra paperwork is incomplete as submitted, with a particular tendency to omit basic information about who committed the homicide in question, including whether the killer wore a badge. In 2011, 31% of supplementary homicide reports omitted the offender’s sex, age and race. When the victim was a black male, basic identifying data on the offender was omitted more often, 39.9% of the time. The form itself is not designed to record, specifically, law enforcement involvement in a death. That information must be written in, inside a small box prompting: “Relationship of victim to offender (husband, wife, son, father, acquaintance, neighbor, stranger, etc).”
Jim Bueermann, a 33-year veteran and former chief of the Redlands, California, police department, and president of the Police Foundation, a Washington-based non-profit, said “good cops are heroic in nature” but can fail to live up to the “unique transparency and accountability responsibilities” of policing in the US.
“There are some jurisdictions that don’t report a justifiable homicide by one of their officers because they don’t believe they’re crimes,” Bueermann said. “Or they just choose not to do this. When you are very diligent at collecting data, it may appear as if you have a much greater problem than your colleagues, your counterparts in other agencies, and that’s something that puts police leaders at risk.”
Even with perfect police compliance, assembling a database with basic details about each incident – how, who, by whom – would require a data revolution. The basic narrative details of each incident are still confined, largely, to local police records, using idiosyncratic method for recording, storing and reporting data. Technology varies widely, from pen-and-paper to 100% digital. Software varies. Human resources vary. And if a new national protocol is ever finally implemented, even the fanciest digital interface currently in use might not be compatible with it.
“Cops, like everyone else, particularly in the offices, are overworked,” said Burghart. “And so when you ask somebody: ‘Oh, I need 14 years of data from you,’ and they keep half their data in files in a room somewhere in boxes, and the other half is on a computer that doesn’t associate, say, officer-involved shootings with vehicular homicides, or Tasers – you’re asking them a lot.”
Police are dependably good at reporting certain numbers. Each year, the FBI publishes a report about law enforcement officers killed or assaulted, known as Leoka. The Leoka reports count assaults on officers – there were 49,851 incidents in 2013 – but more notably devote hundreds of words in each case to describing in detail the circumstances of every “officer feloniously killed.” Excluding accidents, 27 law enforcement officers were killed in the line of duty in 2013, according to FBI figures.
Police work can be fatal for police, too.
A state-by-state ‘super-count’ gone bad – and a search for solutions
Angie Baker works in the Oklahoma state bureau of investigation. She is part of a national network of statisticians producing the US government’s best count yet of people killed by police, under a program known as the arrest-related deaths (ARD) count. For the program, the bureau of justice statistics – nested deep within the Department of Justice –activated a kind of sleeper network of criminologists across the country .
In Oklahoma, Baker started her count with media reports, in part because Oklahoma has 481 law enforcement agencies, too many to survey simply. Baker also consulted medical examiners’ records. She and her colleagues tallied 51 arrest-related deaths by homicide, out of 64 total arrest-related deaths, in the state for 2003-2009. Baker said the data from her state was accurate: “In my opinion, in Oklahoma, I think that it was,” she said. “We just stay really close to the ground, and try to always know what’s going on in our state.”
Arizona, which only has 120 law enforcement agencies, started out by mailing forms to all the agencies. Then the state coordinators – two people working part-time – placed calls to agencies that had failed to reply. Then they searched medical records and compared the data. BJS tallied an average of almost 21 arrest-related deaths by homicide for the state each year. “I think data collection being led locally gives us an advantage,” said Phillip Stevenson, who ran the program. “We talk to every single police agency.”
Multiple state coordinators interviewed by the Guardian offered a reasoned explanation for why the numbers from his or her state were sound. But at the end of the 11 years the program existed, BJS concluded the opposite, without finding flaw in the work of any one state-reporting coordinator.
The count was shut down , because the state-by-state data wasn’t good enough – and there was still a dark number the statisticians were missing.
How should the United States go about building a national database of people who die in interactions with law enforcement? The problem with top-down counts, like the bureau of justice statistics count, is that the tally starts out a long way from the data. The problem with bottom-up counts, like the FBI collection that begins with police records, is a severe attrition rate for data as it climbs toward the national light, as deaths go unreported or unrelayed from one level to another.
“Does that start locally?” asks John Firman, director of research at the International Association of Chiefs of Police. “Does it then go up to the state level? Does it at the end get to the Feds? Or does it go the other way, do the Feds drive it, with national data? That’s the philosophical and actually practical question that everybody’s got to ask themselves.”
Klinger, the Missouri criminologist, has been meeting with the Los Angeles police department to develop a new model for police reporting that would track not just fatal encounters, but all intentional discharges of firearms by police.
“We can’t just look at officer-involved shootings if what we’re trying to understand is the dynamic of police-citizen violence,” Klinger said. “And that’s what we need to get at. We need to get at police-citizen violence. And having a count of dead bodies simply doesn’t give us that much information.”
To gain insight into the use of lethal force by police, Klinger said, the public also needs data on violent encounters that turned out not to be fatal – because most police shootings aren’t. Most intentional firearms discharges by police result in either total misses or woundings.
A study of LAPD shootings from 1980-1988 found total misses 72% of the time. In New York City from 1974-2008, 31.4% of people shot byNYPD officers died, according to a 2011 study by Klinger. In Los Angeles over the same years, the rate of fatalities was 41.8%.
“We’ve already got the instrument squared away – it’s a matter of putting together a code book, getting people trained up,” Klinger said. “And then what they want to do, what LAPD wants to do, is they want to go back to 1993, and look at every single police shooting and enter it into the database.”
Bueermann, the police foundation president, said one sure way to get police to improve reporting would be to use federal money as a carrot. “I think ultimately a system that says if you receive any federal money at all in any form, then you’re required to report this, is probably going to be more effective than one that is voluntary,” he said.
Another way forward, according to experts who have studied the issue, would be for states to pass laws requiring police agencies to report deaths. Twenty-five states already require participation in uniform crime reporting (although data collected in some of those states is never actually transmitted to the national level). Technology offers a potential solution, too, with the possibility of data being culled directly from raw camera footage or other on-the-spot documentation.
Or another, better-funded attempt by the bureau of justice statistics could work. A BJS report last month found encouraging signs that the ARD program might one day capture the universe of law enforcement homicides. The program moved from capturing an estimated 42% of all such homicides, at best, in 2003, to capturing an estimated 69%, at best, in 2011. The statisticians seem eager to keep trying.
“It wouldn’t be too much for them, if they got the resources to do it,” said James P Lynch, a former BJS director and current chair of the department of criminology and criminal justice at the University of Maryland. “They’ve learned so much from their mistakes, that I think that they’re in a much better spot now than they were three or four years ago to do this kind of thing.”
guardian.co.uk © Guardian News and Media 2015