When a white person kills a black man in the United States, the killer often faces no legal consequences.

In 1 in 6 of these killings, there is no criminal sanction, according to a new Marshall Project examination of 400,000 homicides committed by civilians between 1980 and 2014. That rate is far higher than ones for homicides involving other combinations of races.

In almost 17 percent of cases when a black man was killed by a non-Hispanic white civilian over the last three decades, the killing was categorized as justifiable, which is the term used when a police officer or a civilian kills someone committing a crime or in self-defense. Overall, police classify fewer than 2 percent of homicides committed by civilians as justifiable.

The disparity persists across different cities, ages, weapons and relationships between killer and victim.

To understand the gaps, The Marshall Project obtained dozens of data sets from the FBI and examined various combinations of killer and victim. Two types of “justifiable homicide” are noted: “felon killed by private citizen” or “felon killed by police officer.”

(In a bit of circular logic, the person killed is presumptively classified as a felon, since the homicide could be justified only if a life was threatened, which is a crime.)

The data were processed to standardize key variables and exclude more than 200,000 cases that lacked essential information or were homicides by police. The resulting data detail the circumstances of each death: any weapons used; information on the killer’s and victim’s race, age, ethnicity and sex; and how police investigators classify each type of killing (“brawl due to the influence of alcohol,” “sniper attack” or “lover’s triangle,” for example).

Little large-scale research has examined the role of race in “justifiable” homicides that do not involve the police. The data examined by The Marshall Project are more comprehensive and cover a longer time period than other research into the question, much of which has focused on controversial Stand Your Ground laws.

IN THE United States, the law of self-defense allows civilians to use deadly force in cases where they have a reasonable belief force is necessary to defend themselves or others. How that is construed varies from state to state, but the question often depends on what the killer believed when pulling the trigger.

“If there are factors — even if they’re stereotypes — that lead the defender to believe he’s in danger, that factors in, whether it’s a righteous cause or not,” said Mitch Vilos, a Utah defense lawyer, gun rights advocate and author of “Self-Defense Laws of All 50 States.”

Self-defense decisions by regular people, much like those involving police, are made quickly and with imperfect information. As a result, a homicide can be ruled self-defense when the killer faced no actual threat but had a reasonable belief he or she did.

That is where irrational fear can come into play. Police, prosecutors and juries may be apt to give killers the benefit of the doubt in situations when they were faced with someone who seemed “dangerous.”

“Tell me that it doesn’t factor in if the person is black when they’re approaching the suspect,” Vilos said. “It contributes to the decision to pull the trigger because of the fear associated with the stereotype.

“Right or wrong, that’s what’s happening, in my opinion.”

The vast majority of killings of whites are committed by other whites, contrary to some folk wisdom, and the overwhelming majority of killings of blacks is by other blacks.

But killings of black males by whites are more than eight times as likely as all others combined to be labeled justifiable, a racial disparity that is hard to explain based solely on the circumstances reported in the police data and one that has persisted for decades.

By comparison, when Hispanics killed black men, about 5.5 percent of cases were called justifiable. When non-Hispanic whites killed Hispanics, it was 3.1 percent. When blacks killed whites, the figure was just 0.8 percent. When black males were killed by other blacks, and when whites killed other whites, the figure was about 2 percent, the same as the overall rate.

Although the data examined ends shortly after the fatal police shooting of Michael Brown in Ferguson, Missouri, in 2014, reform efforts since then have primarily focused on police shootings. That leaves these civilian cases largely forgotten.

IT IS impossible to say to what extent the disparities — no matter how great — are due to racial prejudice by police, prosecutors or juries. One possible explanation for the differing rates could lie in the different circumstances of the killings, including where they happened.

“If, for instance, white-on-black homicides were mainly defensive shootings in a residence or business, and black-on-white shootings mainly occurred during the commission of a street crime, then the (racial) disparity would be warranted,” wrote researcher John Roman in a 2013 Urban Institute study of justifiable homicides.

Although the FBI’s Supplementary Homicide Report tracks more than 100 details about each killing, the location of the death is not recorded. In addition, some police agencies, indeed some states, choose not to share some or all information on killings.

Still, the disparities in how police classify these cases remain across widely different circumstances and causes of death. Whether the killer and victim were married, lovers, neighbors or complete strangers, whether they were shot, stabbed or beaten, the trend holds. The killings of black men by whites were two to 10 times as likely to be called justifiable.

Even after adjusting for the ages of the killer and victim, their relationship and the weapon used, the likelihood of a white-on-black-male case being called justifiable was still 4.7 times higher than in other cases.

Beyond police consideration of each case, there are three additional stages at which a killer can establish self-defense: A prosecutor can decide not to pursue charges, a grand jury can decline to indict, and a jury can find a killer not guilty.

There is no national standard for how a prosecutor’s office decides whether to prosecute a case.

In Miami, for instance, an individual assistant state attorney can determine whether to take a homicide case to a grand jury or, with the approval of a supervisor, to decline to prosecute. In Phoenix, however, each case goes through a detailed review by a panel of prosecuting attorneys before a decision is made, according to officials there.

The FBI data may not reflect those determinations: Prosecutors are not generally required to collect and report their data. Despite these flaws, the data are the closest thing available to a comprehensive, nationwide accounting of homicides.

DESPITE LOCAL customs and policies for how prosecutors approach self-defense cases, in the FBI’s data almost every law enforcement agency, from Oakland, California, to Philadelphia, showed racial disparities in how they categorized killings as justified. The phenomenon is not peculiar to any state or region.

In all cases and at each stage, the system depends on the reports and decisions of police investigators, which provide the record upon which conclusions are drawn. Melba Pearson, a former prosecutor, said that in her experience, personal preconceptions affect how those cases are policed and whether they are prosecuted.

“The reality is every human being comes to the table with biases,” said Pearson, former president of the National Association of Black Prosecutors. “That’s human nature.”

She called that a larger problem than the smaller number of “straight-up racists.”

Personal experiences and cultural background mean even well-meaning prosecutors can succumb to stereotypes, said Pearson, who spent 14 years as a prosecutor in Miami and now works for the American Civil Liberties Union of Florida.

She said this problem, part of what some call “implicit bias,” can be overcome by training and policies to help achieve an objective analysis.

Cyrus Vance Jr., the Manhattan district attorney, adopted one such policy, an aggressive training program after a 2014 report by the Vera Institute of Justice found racial disparities in the outcome of prosecutorial decisions.

When announcing the report, with which his office cooperated, Vance said he was “committed to implementing preventative strategies to reduce any unintended racial and ethnic disparities.”

The office requires every prosecutor and staffer to take multiple classes on recognizing “implicit bias” and combating its results.

Without looking at prosecutors’ and police’s unconscious prejudices, Pearson said, racially biased outcomes are inevitable.

“Ignorance is not bliss,” she said.

———

METHODOLOGY

This article was produced in collaboration with The Marshall Project, a nonprofit news organization that focuses on criminal justice issues.

The Marshall Project requested data from the FBI’s Supplementary Homicide Report and received records detailing close to 600,000 homicides from 1980 to 2014. The analysis focused on 400,000 cases where information about the killer was available.

The police data do not necessarily reflect the final outcome of a case, which is determined by prosecutors and the court system and does not always get reported back to the FBI. For instance, Trayvon Martin’s death is listed simply as “other,” not as “justifiable,” though his killer was later tried and acquitted. That could mean that “justifiable” cases of all kinds are underreported.

The data entered by police are incomplete; not all police departments participate and not all states collect the same data. In some cases, particular states did not report data on justifiable homicides. The Marshall Project separately collected data from Florida, which has not provided homicide data to the FBI since the mid-1990s.

In another example, New York state failed to collect and report data on any justified homicides from 2007 through 2014. However, despite the gaps and problems, when jurisdictions with inconsistent data were removed from the equation, the racial disparities remained.

© 2017 The New York Times Company