Audio By Carbonatix
At 2:10 a.m. on December 19, Denver police officer Timothy Campbell was standing in the middle of the street in a west Denver neighborhood, his gun pointed at a man.
The patrolman had been driving north on Irving Street when he’d passed a 1997 Saturn that seemed suspicious. When Campbell made a U-turn, the Saturn quickly sped down a side street and pulled into a driveway. As the officer drove up, a man — he looked to be in his early thirties, Hispanic, wearing a light, baggy jacket — jumped out of the car and ran. Campbell followed him on foot, through back yards and over fences. The man reached the 3200 block of West Ada Place, where he slipped on a patch of ice. He got up and continued down the street, falling twice more. By now Campbell had closed the gap, and when the man got up again, the two were facing each other, less than ten feet apart. Campbell had his service pistol drawn: a .45-caliber semi-automatic Glock.
The man reached into his pants pocket, put his hand behind his back, then started moving his hand forward. Campbell saw the glint of something metallic. He fired two rounds, paused, then fired four more. The man fell onto a pile of dirty snow.
When paramedics arrived just after 2:15 a.m., they found 33-year-old Jason T. Gomez, hit in the shoulder, stomach and legs, mortally wounded. Near his left hand, they spotted a white Bic lighter with a silver rim.
A lighter on the pavement where there should have been a gun — that sight can make even the most hard-boiled law-and-order types queasy. And the image of a dying, unarmed man, a minority shot by a cop, can rip open a city’s carefully patched-together image. When news broke that Gomez had been pronounced dead at Denver Health, readers began leaving online comments comparing Gomez’s lighter to the soda can that Frank Lobato reportedly was holding when he was shot and killed in his home by a Denver officer in 2004. Or the kitchen knife that Paul Childs had in his hand when the mentally disabled teen was shot and killed by cops the year before. The posters reached back nearly a decade, to the death of Mexican immigrant Ismael Mena, shot by SWAT officers in a botched drug raid.
“[Gomez] was not a perfect person, but [he] did not deserve to have an entire clip of bullets emptied into him for pulling out a lighter,” said one.
“Again, Denver cops are exterminating Blacks and Mexicans,” wrote another.
Long before Campbell faced off against Gomez on that icy street, though, the Denver Police Department had started taking a long, hard look at what role race played in officer-involved shootings. To do so, it was using an unlikely tool: a rudimentary video simulation developed by psychologists at the University of Colorado. Over the past half-dozen years, that simple computer game has allowed researchers to not only measure the influence that cultural bias has on police decisions, but to make some surprising discoveries regarding how the human mind forms and acts upon racial prejudice.
In 2002, Tracie Keesee spotted a small article in the Rocky Mountain News about a CU study demonstrating that participants playing a virtual-simulation scenario were quicker to fire at black male figures than at whites. This interested Keesee, who was not only a University of Denver graduate student working toward a degree in criminal justice, but also a lieutenant in the DPD with deep roots in the city’s African-American community.
“I thought it was really relevant to large police organizations — the use of deadly force and how it impacts people of color, specifically African-Americans,” says Keesee, who’s now a district commander considered a strong candidate to become the city’s first female and first black police chief. “Whenever you read the newspaper, whether it be New York or Chicago or Denver, it continues to be a very prevalent question.”
Over the years, law-enforcement officials have used hundreds of jargon-filled euphemisms to avoid the query at the heart of so many police-shooting controversies: Are cops more trigger-happy when aiming guns at minorities? Since the 1970s, sociologists and political scientists have consistently found that minority suspects in the United States face lethal force from police officers at a disproportionate rate. According to 2001 figures from the Department of Justice, black suspects were five times more likely to be shot and killed by officers than white suspects. But that same study also showed that the chances of a police officer getting shot by a black man were about five times higher than by a white man. And how much could these findings be attributed to the fact that minorities are much more likely to face economic deprivation and populate disadvantaged, high-crime areas — and thus have a greater probability of contentious encounters with police?
For social psychologists at the CU Stereotyping and Prejudice lab (CUSP), the 1999 death of Amadou Diallo — an African immigrant shot nineteen times by New York City cops when he reached for his wallet rather than a gun — seemed an ideal starting point for a study of racial bias. Joshua Correll, a graduate student at the time, followed the subsequent investigation of the officers and the allegations that race might have played a role in the shooting. “And that seemed interesting and plausible, but it was hard to understand how much of a role race actually played, because we didn’t know what would’ve happened if Diallo had been white,” says Correll, now a professor at the University of Chicago.
So Correll and Bernadette Park, a psychologist who’s been at CU since the mid-’80s, developed a video shooting game that involved black and white male targets holding either guns or innocuous objects such as cell phones or soda cans. They told study participants — found in college classrooms and along the 16th Street Mall — to shoot only when the simulated characters were armed. The data revealed that players forced to make split-second decisions were prone to shoot images of unarmed black men and on average were quicker to shoot at black men holding guns than white men who were armed.
“[Participants] set a more lenient criterion to shoot for African-Americans than for whites,” says Park, adding that this tendency was seen not just among Caucasian players, but also among players identifying as black or Hispanic. In 2002, she and four colleagues published a report on their research in the Journal of Personality and Social Psychology. A year later, University of Washington psychologist Anthony Greenwald did his own study, putting college students in the role of plainclothes police officers in a computer simulation where potential targets of different races appeared from behind dumpsters as fellow officers, citizens or gun-wielding criminals. Players had greater difficulty distinguishing weapons from harmless objects when they were in the hands of blacks rather than whites, resulting in more wrongful shootings of black targets.
The studies established shooter bias as a broad cultural predisposition, but researchers had no way of knowing if police officers would test the same way. So CUSP contacted the Chicago and Los Angeles police departments, but they refused to participate in further studies. “Basically, we had a really hard time finding anyone who was willing to even say ‘I’ll open the doors,'” Park remembers. The academics finally teamed up with the Police Executive Research Forum, a national law-enforcement policy group, and won a grant to extend the shooter-bias tests to cops. But the police group got nervous and backed out at the last minute, Park says, effectively closing the door on the CUSP project.
Then Park got an unexpected call from Keesee.
“She said she was a graduate student and wanted to ask me some questions about the study,” recalls Park. “I called her back and we chatted a little bit about it, and about ten minutes into the conversation, she said she was actually a police lieutenant.”
Park mentioned the conversation to her colleagues: “And they said, ‘Did you ask her if we could have access to the Denver police?’ And I said, ‘No, she just wanted to know about the work.’ And they said, ‘Call her back after the meeting!'”
Park did, and Keesee was open to discussing the idea. But as a then-thirteen-year veteran of the DPD, she knew that gaining authorization to test the city’s cops for racist shooting tendencies would not be easy. For starters, the departmental bureaucracy would have to overcome its natural instinct to close ranks to outsiders, a stance that becomes particularly useful whenever race is at issue. “A lot of large organizations would not allow outside researchers to come in and look at hot-button topics such as race bias and the use of deadly force,” Keesee acknowledges.
So she decided to take the idea right to the top: Chief Gerald Whitman, who agreed to let the CUSP researchers have a sit-down with department brass. Then Whitman had to weigh whether getting useful data on police-shooting decisions was worth the possible damage that negative results might have on the department’s already tenuous image within the black and Hispanic communities. After all, fallout from the Mena shooting had ultimately forced the resignation of the previous chief, Tom Sanchez, in 2000. But ultimately, Whitman agreed to the DPD’s participation.
“I wanted to see what we could learn from this survey,” says Whitman. “It was something that would improve officer safety and citizen safety and make us better police.”
“It took a lot of courage for the chief to agree to let them come in and do some research,” adds Keesee.
Keesee, who is continuing her education by working toward a Ph.D. in intercultural communications at DU, is currently the commander of District 3, which covers Capitol Hill south to the Denver Tech Center. It’s the district in which she was born and raised — her mother was a nurse, her father in the military — and where she started her career as a police officer in 1989. At the time, she was a single parent with a young daughter. She’d considered becoming a lawyer, but opted for law enforcement instead. “It was between joining the Denver Police Department and joining the Houston Police Department,” she says. “This being my home, I decided to stay here.”
But Keesee’s district includes neighborhoods with some of the lowest minority populations in the city. So when it came time to find officers to volunteer for the study, Park and Keesee spread their recruitment efforts across four police districts, eventually getting shooter data from 124 local officers. To compare the cops with community members, they staked out Division of Motor Vehicle offices in the same districts and asked citizens to do the shooter simulation. For a national sample of cops, Park and other researchers traveled to officer-training seminars across the country.
The results of their study were published in the Journal of Personality and Social Psychology last July, in a paper titled “The Thin Blue Line: Police Officers and Racial Bias in the Decision to Shoot.” As in the earlier CUSP study, community members displayed a greater willingness to shoot an African-American target than a white one. But while police officers still displayed discrepancies in speed of response depending on race, they vastly outperformed civilians in accuracy — meaning that cops did not make the ultimate decision to shoot based on a target’s skin color.
For Correll, who’d initially hypothesized that officers would show the same biases as civilians, the results were very surprising. “Police officers are people,” he says. “They have the same basic mental processes that we do, and they should be governed by the same types of things. So in a culture that pretty regularly associates black people with danger and criminality, we might expect [police] to look just like everyone else.”
Instead, this study indicated that police officers’ training not only affected whether they chose to fire at a target, but made them less likely to shoot on the basis of race.
The DPD and the Greater Metro Denver Ministerial Alliance held a press conference to announce the results. The New York Times picked up the story, as did National Public Radio, but to the disappointment of both the researchers and the DPD, no local outlets covered it. After all the times the department had been hammered by the media for controversial police shootings, the one time it had new information shedding light on a complex subject, no one paid attention.
“It made you stop and think about the long-held assumption that we all have,” Keesee says. “This was very important from a law-enforcement prospective. A significant step was made to answer a question that communities of color have had for a very long time.”
Keesee did receive inquiries about the study from police departments in several other cities. And national law-enforcement organizations heralded the results as further evidence that cop-shooting decisions are not driven by race bias, that the widely held belief that police are “trigger-happy” for certain ethnicities doesn’t stand up.
If anyone is trigger-happy for minorities, it’s society at large.
I have come to the “shooter study” test site in the psychology building on CU’s Boulder campus to find out if I’m a racist.
I’m welcomed by a graduate student whose job is to watch other students as they simulate shooting people of different races. The data is then compared to the performances of Denver citizens and police officers, as well as officers from around the country. This is not the only CUSP study in progress; five faculty social-psychologists and half a dozen graduate students work here, researching everything from how blacks with greater Afro-centric facial features are disproportionately represented in prison populations to how racial prejudice can be measured in brain waves.
“Go ahead and have a seat,” the grad student says, gesturing toward dozens of desks, a small computer on each, lined along the walls. I’m given a consent form that says I know this is an anonymous, voluntary study. On the desk in front of the computer screen is a little box with three buttons. The green one is labeled “shoot,” the red one “don’t shoot.” There’s also a yellow button that I’m supposed to press when I want to start. I do, and instructions pop up on the screen that tell me my task is to shoot anybody holding a gun. I have less than a second to make a choice, and if I don’t, I lose points. The point system goes like this:
+5 points for deciding not to shoot at an unarmed man
+10 points for shooting an armed man
-20 points for shooting an unarmed man
-40 points for being shot
There are two versions to this game — a fast one and a super-fast one. I’m on the super-fast one, which calls for quicker decisions and has a greater likelihood of revealing racial bias. I’m given a practice round, and the game flips through a series of background images showing locations in Denver: a light-rail shelter behind Union Station, a sidewalk in Civic Center Park, an alleyway near downtown, a spot on Capitol Hill. The first man who pops up is a white guy crouching with a gun. I press “shoot.” As the game continues, the images come faster. I find myself accidentally shooting white men with Coke cans, black guys with cell phones. But if I wait too long on an image, I lose my chance and points. So I choose “don’t shoot” prematurely a few times and get shot, losing still more points.
When I start the real game, I make correct choices on the first three images but bungle the next two. There’s no gunfire; the only sound is the periodic clack of the button as I try to decide on my course of action as quickly as possible. When I shoot another black guy with a cell phone, I wonder if I’ve just revealed some hidden racial bias. So when I see the image of another black male, I overcompensate by hitting the red button too quickly — and end up getting shot myself.
After about two minutes, the game is over. While I’m not confident of my accuracy score, I think I’m pretty safe from scoring as a racist, since I probably shot as many unarmed white males as blacks. The truth of the simulation, however, is not necessarily measured by the overall score, but in milliseconds. How long did it take me to decide if I viewed a person as a threat — and with which race was I most often correct?
The results of my test are determined by this formula: f(1,361)+239.37,p<.001. it showed that i was percent more likely to mistakenly shoot an unarmed black than white took me a third longer decide man not threat. and this computer game.>
