I want to thank LetterWiki for hosting our exchange, and you for your letters. As you know, I've followed you for some time, and I know that whatever disagreements we may have, we share a common respect for honest conversation. And the current protests over the horrific murder of George Floyd demand nothing less.

I want to start by challenging the title of this conversation: Systemic Racism in Police Shootings. Here's why:

George Floyd wasn't shot.

Eric Garner wasn't shot.

Freddie Gray wasn't shot.

The #BlackLivesMatter slogan, "I Can't Breathe," isn't about anyone getting shot.

And Rodney King — whose brutal beating by four white police officers was one of the first to be videotaped and shown to the world — wasn't shot, or even killed.

In fact, a good amount of the data that you cited on police shootings in your three-part letter would likely exclude some of the most prominent cases that the #BlackLivesMatter protests are driven by — including that of George Floyd.

This is why I would argue that this needs to be a broader conversation. We can talk about police shootings as well, but that would be like trimming a twig off a poisonous tree while leaving the roots intact.

We will get to the data in a bit. But I want to ensure that we first define our terms and set up the context of this conversation.

Since we're talking about systemic racism, I want us to be clear about what it means. In your letter, you wrote, "BLM is an organization that came into existence after Trayvon Martin was fatally shot in Sanford, Florida in 2013..." This is not entirely accurate. The About section of the #BlackLivesMatter website states that "#BlackLivesMatter was founded in 2013 in response to the acquittal of Trayvon Martin’s murderer." [emphasis mine]

Same goes for the 1992 Los Angeles riots. The Rodney King beating was captured on video and disseminated to the media in March, 1991. The riots began in April, 1992. They weren't sparked by the beating. They were sparked by the acquittal of the officers charged with it.

This is a key distinction — and it's at the heart of what systemic racism really means. It's not just about shootings. It's not just about police brutality, either. It's about accountability. It's about justice. It's about those in power getting away with bad things that those without power wouldn't. To paraphrase Stokely Carmichael, “If you want to lynch me, that’s your problem. If you have the power to lynch me, that’s my problem. It’s not a question of attitude. It’s a question of power.”

This is about the fact that the vast majority of police officers who shoot and kill are never convicted. This was true long before the #BlackLivesMatter movement. But most of us gave these officers the benefit of the doubt. We assumed that those who pledge to serve and protect us were doing exactly that.

But something changed with Rodney King: the brutality was recorded on video and broadcast to the world. The very day the cops were acquitted, Tom Bradley, the city's mayor at the time, captured this new reality in his words: "The jury’s verdict will never blind the world to what we saw on the videotape."

Now, in an age where everyone has a video camera in their pocket, we know better.

The police said George Floyd was resisting arrest. The video showed he wasn't.

The police said Martin Gugino — the 75-year-old Buffalo protester pushed to the ground and ignored by them as he bled out — "tripped and fell." The video showed that this is not what happened.

The police said an Australian news cameraman "may have fallen." The video showed he was struck hard with a shield.

The police do lie. And they have reason to.

This is a key point to consider when contextualizing the research and data around police violence. Remember, in much of this research, the data is compiled from police reports, often written by the officers involved in the incidents.

If you were researching deaths due to medical errors or negligence, would you rely solely the official reports written by the accused doctors or hospitals?

Let's look at an example. The paper I see cited most by those who say there are no racial disparities in police shootings is the Ronald Fryer paper from 2017.

Fryer's research found that blacks and Hispanics were more than 50% more likely than whites to experience force in interactions with the police, and these disparities were not fully explained by correcting for factors like increased crime rates in minority communities. However, when it came to officer-involved shootings (this is the part cited most by systemic racism-skeptics) there were no racial disparities seen.

The study was widely criticized, primarily because nearly all of the data on police shootings comes from police reports and accounts written by the involved officers themselves.

This also applies to the FBI stats (which you cited). Even when video contradicts the officer's report, it is the officer's report that is sent to the FBI. Moreover, as Radley Balko at the Washington Post has pointed out, the FBI only tracked "justifiable homicides" until a few years ago. How much of a difference does this make?

A huge one. Writes Balko:

According to 2014 report in the Wall Street Journal, over a seven-year period ending in 2011, 41 police officers were criminally charged for an on-duty shooting. Over the same period, the FBI reported 2,711 justifiable homicides by law enforcement. That means that in at least 98.5 percent of police shootings, the officer was cleared of criminal wrongdoing. Given that we now know that the FBI figures have been vastly underreporting police shootings, it seems safe to say that the clearance rate is over 99 percent.

He continues:

According to The Post’s data for 2015, in “74 percent of all fatal police shootings, the individuals had already fired shots, brandished a gun or attacked a person with a weapon or their bare hands” — what the project has deemed an “attack in progress.” This has led many to conclude that 74 percent of police shootings are justified. The Post’s figure draws on media reports in addition to police reports, so it’s significantly more nuanced and targeted than the FBI data. It’s also already a much lower number than the percentage of shootings deemed justified by judges and prosecutors. Note that the Post doesn’t state the other 26 percent were illegal, just that they weren’t clearly justifiable, so it isn’t a perfect comparison. But this still means there are significant questions about one in four fatal police shootings. Before, all we had to go on were prosecution statistics, which cleared cops in more than 99 percent of shootings. More information has raised more questions. I don’t know what an acceptable number would be, but one in four seems awfully high.
But again, what #BlackLivesMatter is talking about is hardly restricted to shootings. And, as even Fryer's research demonstrates, there is clear evidence of racial bias when looking at force beyond just shootings — as experienced by George Floyd — and these incidents cannot be fully explained by "adding controls that account for important context and civilian behavior."

It is also important to understand what systemic bias does not mean. For one, it is expressly not about an army of racist cops looking to shoot or kill black people. As Balko writes, "[Systemic bias] means that we have systems and institutions that produce racially disparate outcomes, regardless of the intentions of the people who work within them." [emphasis mine]

In my book (thank you for quoting it, by the way!), I talk about how I'm frequently told not to blame religion for the bad actions of a few religious people. I disagree. If anything, I see it the other way around. The problem isn't Muslims in general — the vast majority are peaceful. The problem is the ideological system that intends to indoctrinate and train them to believe and internalize specific attitudes towards women, gays, and non-believers.

I find a similar dynamic in the "few bad apples" philosophy in policing. We repeatedly hear that the system is okay, and a few bad cops are misrepresenting it. But if we are going by the evidence — which you very eloquently made the case for — it's actually the other way around.

None of us can say for sure that Derek Chauvin was a racist. We can presume he was, but none of us have access to his mind. None of us can say for sure that he wouldn't have killed a white man the way he killed George Floyd. As you noted, white cops killed Tony Timpa, a white man, in a very similar way. The evidence gives us very limited information about these individual "bad apples."

But the evidence does tell us that men who look like George Floyd are much more likely to experience force in their interactions with police than men who look like Tony Timpa (more than 50% more likely, according to Fryer himself.)

Evidence from just last month — a huge study of 95 million traffic stops over a span of eight years — not only tells us (as other studies have), that black people are much more likely to be pulled over than white people; it also confirms racial bias in these stops by demonstrating how the disparity significantly lessens at nighttime, when the driver's race can't easily be identified. And when they are stopped, the evidence shows black drivers and passengers are much more likely to be searched compared to whites, even though whites are more likely to be carrying contraband.

The evidence shows that 1 in 1000 black men in the United States are likely to die at the hands of police. As you correctly pointed out, the probability of unarmed black Americans being shot by police is about 3.5 times that of analogous whites; but this particular study also shows there was "no relationship between county-level racial bias in police shootings and crime rates (even race-specific crime rates), meaning that the racial bias observed in police shootings in this data set is not explainable as a response to local-level crime rates."

The evidence shows that once arrested, black people are likelier to be charged, incarcerated, and receive longer sentences than whites for similar crimes. There are several studies that even show disparities in sentencing between light-skinned blacks, medium-skinned blacks, and dark-skinned blacks, who receive progressively longer sentences, respectively, compared with whites for similar crimes; this sheds light on how the problem may not just be about blacks committing more crimes, but directly related to their skin color.

Evidence from a 2011 study on bail shows that blacks received $7,000 higher bail than whites for violent crimes, $13,000 higher for drug crimes, and $10,000 higher for public order-related crimes; the value of blacks' lost freedom, says the study, "was less than two-thirds the value of whites' lost freedom."

I'll stop here out of respect for LetterWiki's word count limits.

No, not all of the evidence above is about police shootings (although some is).

But neither were the deaths of George Floyd, Eric Garner, Freddie Gray, Rodney King, and many others.

It also wasn't "officer shootings of unarmed suspects" that "sparked the creation of #BlackLivesMatter," as you wrote. It was that these officers are frequently acquitted and get away with it.

One simply cannot isolate one rare aspect of policing — shootings — from the vast evidence for systemic bias because it (arguably) helps make one's point.

Will leave it there for now. Over to you.

Ali Rizvi

Ali Rizvi

There are many things we agree upon. For example, not all police shootings are justified. In a nation of 330 million people, 393 million guns and 375 million police-contacts a year, American police officers generally do excellent work, and the data supports that. But police officers, like all human beings, sometimes commit crimes.

I also agree that racism exists. I agree, as you stated recently, that America is the least racist nation on Earth. And while we may disagree on the utility of the current protests, we would agree on the importance of the Bill of Rights and the need for these principles to be defended and applied to all citizens.

The narrative repeated at the protests, in our media, and by BLM is that white police officers are murdering black Americans. I don’t want to belabor the point by citing source after source. There is no denying this fact. Let’s agree on the seriousness of that accusation and the stakes involved. Cities around the nation are considering abolishing police departments. For some, the slogan “Defund the Police,” is more nuanced, but for others, in Congress and in New York Times OpEds—they literally mean abolishing the police. That is an idiotic, dangerous idea.

Already the NYPD has disbanded their plain clothes division. And my own city has ended the School Resource Officers (SRO) program. Studies have shown SRO to reduce things like the sexual abuse of children, guns in school, and overall violence. These reforms will have an impact. The problem is that impact may end up being counted in lives lost.

As I wrote previously, police shootings did not account for “suspects killed by officers using other means, as happened in the case of George Floyd.” I didn’t focus on the topic of police shootings to avoid talking about other uses of force. I began with police shootings because it is the primary accusation one hears repeatedly leveled against police.

Also, I’ve frequently witnessed people who argue that systemic racism is apparent in police shootings, apply a motte-and-bailey strategy, switching back and forth between more defensible claims to get mileage for less defensible ones. I will address all uses of force, but for clarity, we must discuss each dataset carefully.

There are dozens of studies (I will post a list for those interested later) that corroborate what I’ve argued. However, I didn’t want to pepper you with more data points than you could possibly address in your 1000 words per page letter.

The statistic you note: “the probability of unarmed black Americans being shot by police is about 3.5 times that of analogous whites,” is frequently cited. What is not cited is the rate of offending. My point, which you have yet to address, is simple—stating that black Americans are 3.5 times more likely to get shot by police than white Americans in no way proves systemic racism any more than saying 95% of all human beings shot by police are male proves systemic sexism. To show bias we have to control variables like the rates of offending, and when we do, that 10% gap (13% of the population but 23% of those shot by police) is wholly explained.

Next, you mention Rodney King, and list him lower in your letter as someone killed by police. Rodney King was not killed by police. He won a 3.8 million lawsuit, and later drowned in a pool while intoxicated. We should remember that the LA riots, which followed the acquittal of the officers in the King case, resulted in 63 deaths, 2,383 injuries, and material losses in the range one billion dollars. While I have seen many people in the media justify and rationalize burning and looting in our current riots, not one of them was ever at risk of having their own home or business destroyed.

You also mention how BLM came into existence, and discuss what’s meant by the term “systemic racism.” I referenced BLM’s creation after the Trayvon Martin case solely for chronology, not intent. I fully understand what you and others mean by “systemic racism,” but that definition only furthers my point. If the bias as it relates to police officers using force that results in the death of suspects really was built into the system, then that bias—systemic as it is—will be reflected in the data. It’s not. This is why a close and careful reading of the data is indispensable.

You then write:

“Now, in an age where everyone has a video camera in their pocket, we know better.”

The truth is, now, in age where everyone has a video camera in their pocket—what we know is what we see on television and what pops up in our social media feeds.

Anecdotal stories and individual cases are not proof of systemic problems or nationwide trends. They are, however, food that fuels “availability bias."

I previously I mentioned Tony Timpa. Most had never heard of him. He died in identical circumstances to George Floyd. In fact, officers held down Timpa for 14 minutes and joked as he went unconscious. The only difference between the cases is that Timpa was white. I could have just as easily named Alfred Redwine, Andrew Thomas, Loren Simpson, Dylan Noble, James Boyd, Brandon Stanley, Mary Hawkes, etc. How many people who’ve heard of Trayvon Martin and George Floyd have also heard of any of the names listed above?

We don’t decide if police officers are murdering black Americans at a higher rate based on what CNN, MSNBC, or Fox air on television. We don’t don’t get to decide reality based on what gets shared on social media platforms. We have to look at evidence.

In fairness, I am sure you were pointing out that prior to the ubiquity of cell phone cameras, many of these cases may have never been brought to light. I agree.

But the ones we see repeatedly on our cable news Networks tend to involve black victims, whereas the more frequent cases involving white victims rarely make it past the local news.

We are a Nation of 330 million people, 400 million guns, and more than 16,000 homicides annually. There will be another George Floyd or Tony Timpa. Given the fact that 63 people lost their lives in the LA Riots alone, more responsibility on the part of our national media is due.

Next you mention the Fryer study:

“the evidence does tell us that men who look like George Floyd are much more likely to experience force in their interactions with police than men who look like Tony Timpa (more than 50% more likely, according to Fryer himself.)”

And you cited this study to support your claim: https://scholar.harvard.edu/files/fryer/files/empirical_analysis_tables_figures.pdf

What you did not mention is that on page #2 of that same study it states:

“Put simply, if one assumes police simply stop whomever they want for no particular reason, there seem to be large racial differences. If one assumes they are trying to prevent violent crimes, then evidence for bias is exceedingly small.”

The “assuming they’re trying to prevent violent crimes” is the part that went unmentioned.

We can assume racism and bad faith on the part of these officers, which is what we do when we cite the “50% more likely aspect” without also mentioning the “assuming they’re trying to prevent violent crime” detail. The problem is that we’re back to post-hoc, faith-based reasoning, which you rightly argue against in your book. Or, we can dig into the study and its four datasets. Of these, two were related to shootings (they found that “conditional on a police interaction, there are no racial differences in officer-involved shootings on either the extensive or intensive margins” as you acknowledged), and two that were related to a “wide range of uses of force—from putting hands on civilians to striking them with a baton.” With data from NYC.

If we look at NYC during that time period, are there credible reasons to believe the NYC officers were trying to prevent violent crime?

Here’s what the data shows: in 2008, 62% of NYC homicide victims were black, and 52.3% of those arrested for homicide were black. In 2009, 57.6% of NYC homicide victims were black, and 58.5 of those arrested for homicide were black. In 2010, 67% of the NYC homicide victims were black, and 54.5% of those arrested for homicide were black. In 2013, 62.9% of NYC homicide victims were black, and 54.0% of those arrested for homicide were black. I could go on, but I’m sure you see the trend. These percentages remain consistent every year. And homicide isn’t the only violent crime where these patterns manifest. If we look at reported rapes in that same year, 2013, 40.8% of the reported victims were black, and 46% of those arrested for rape were black. African Americans represent 24% of the NYC population.

Now to victims. Unless someone believes there are a substantial number of dead bodies that are never found (not suspects but bodies), homicide victim numbers are consistent. Every year in NYC a demographic that makes up 24% of the total population also makes up more than 60% of the homicide victims. We also know that around 90% of all homicides, whether the victim is black, white, Hispanic, or Asian, are intraracial. This is not surprising, as more often than not people are murdered by someone they know.

Now back to our question. Were the NYC officers trying to prevent violent crime?

If a group that makes up 24% of the total population in NYC is also making up more than 60% of the homicide victims, year after year, what percentage of police attention should they receive? What percentage of police attention and interaction do the victims in that community deserve?

When we account for rates of offending, the idea that NYC officers were trying to prevent violent crime is a reasonable, cogent explanatory mechanism.

You then add:

“but this particular study also shows there was "no relationship between county-level racial bias in police shootings and crime rates (even race-specific crime rates), meaning that the racial bias observed in police shootings in this data set is not explainable as a response to local-level crime rates."”

However, the study you cite as evidence is an analysis of traffic stops in Rhode Island. I’ll assume that was just an error.

You also bring up other subjects I want to discuss. Religion as an example of something that does contain “systemic” problems. I agree. I also think the comparison to law enforcement institutions will be useful. A traffic study, which I read. I will also go into that in future replies.

At the end you add issues of sentencing, incarceration terms, and bail. While I understand you want to paint the complete picture of systemic racism, I was specifically talking about police shootings. We have now broadened that to use of force, and even as in the case of the traffic stops, policing in general. That’s fine. However, with sentencing you are diving into things that are outside the scope of a police officer’s control. These are worthwhile topics, but we have just 1000 words per page replies. Studies which have published criticisms and analysis take thousands of words each to unpack. Since protestors, and the entire might of America’s mainstream media have turned their anger on the men and women of law enforcement, let’s give them the focused attention this aspect deserves.

If we find no evidence of racial bias in police using deadly force, that should be significant. It should lead to important questions.

This is not data that can be ignored if we really want to "fix" policing.

Matt Thornton

Matt Thornton

Thanks again for a thoughtful response. Again, we will get to the data in a bit, but I think this is a good place to acknowledge where we do agree.

We both recognize that racism is real and is still a problem in the US. We both also agree that the US is one of the least racist countries on Earth. Having grown up in countries like Pakistan and Saudi Arabia, where racism is not only rampant but often state-sanctioned, I know that the racism in the US and Canada, both nations of immigrants, is not as prevalent as it is visible, which is a testament to our processes of justice and accountability.

But it's even more important to recognize how the US and Canada came to be among the least racist, misogynistic, and homophobic countries in the world.

It is precisely because of the kinds of protests you're seeing now.

Whether it was the suffrage movement, abolition of slavery, desegregation, civil rights, or LGBTQ rights — protests and public uprisings have been the key driver for the transformative changes that actually did make America great.

And these struggles were far bloodier in the past — riddled with violence, turmoil, and assassinations of key civil rights leaders — even if you don't count the Civil War.

While it seems like chaos when we're in the thick of it, with the fringe elements on both sides making headlines, the trendlines are already beginning to look positive. 74% of Americans generally support the protests, and in a stunning sign of progress, this includes a majority of Republicans (53%). NASCAR has gotten rid of Confederate flags. The NFL has reversed its position on players like Colin Kaepernick taking a knee during the national anthem, and now support them. Major corporations — from Wall Street to tech to retail — are openly showing solidarity. The television and film industry is re-evaluating how it tells stories about policing. Police departments are being overhauled and reformed.

Moreover, there are prominent voices that are able to separate substantive ideas from the fringe ones. Obama, Killer Mike, Atlanta Mayor Keisha Lance Bottoms, and even the family of George Floyd have denounced the vandalism and the looting. Mainstream liberals, including the Biden campaign and Bernie Sanders, have rejected the absurd calls to "abolish the police," while embracing powerful reform proposals that divert more funds to mental health professionals and social workers, to ease pressure on police from having to deal with complex issues like substance abuse and homelessness on their own, instead giving them a secondary role. These measures aren't just hypothetical — they have been tried and tested in places like Camden, NJ, which used to be one of the most violent cities in the US.

But let's get back to the topic of systemic racism. By definition, the term refers to racism in the entire system, not just police shootings (a particularly rare form of police intervention). In my previous letter, I specifically mentioned how police shootings data would exclude the cases that triggered the current protests (George Floyd wasn't shot), the 1992 LA riots (Rodney King wasn't shot), or even the birth of #BlackLivesMatter (Trayvon Martin wasn't shot by police).

Let's talk about the data point both of us noted: the probability of unarmed black Americans being shot by police is about 3.5 times that of analogous whites. You maintain that this does not account for the rate of offending. Actually, it does. The paper, a peer-reviewed analysis by Cody Ross, states clearly:

"There is no relationship between county-level racial bias in police shootings and crime rates (even race-specific crime rates), meaning that the racial bias observed in police shootings in this data set is not explainable as a response to local-level crime rates."
(I know I quoted this in the previous letter but posted the wrong link. I have corrected that now.)

Now, I understand that the Ronald Fryer paper comes to a different conclusion for shootings. For this reason, the Fryer paper is one of a handful of sources repeatedly cited by systemic racism skeptics. But, like everything else, data must also be looked at critically.

First, the Fryer paper, published at the National Bureau of Economic Research (NBER), is not peer-reviewed. At the outset, a disclaimer states:

"NBER working papers are circulated for discussion and comment purposes. They have not been peer-reviewed or been subject to the review by the NBER Board of Directors that accompanies official NBER publications."
The Ross paper is peer reviewed.

Second, studies like Fryer's have relied almost entirely on police reports written by the involved officers themselves for its shootings data. The Ross paper corrects for this bias in its methodology. From the abstract:

"In contrast to previous work that relied on the FBI’s Supplemental Homicide Reports that were constructed from self-reported cases of police-involved homicide, this data set is less likely to be biased by police reporting practices."
I can't stress enough how important this is when it comes to looking at putting data in context.

Consider this: according to the UN, the prevalence of rape in Sweden is 24 times higher than in India, and 135 times higher than in Saudi Arabia, which just happens to have one of the lowest rates of rape in the world. These are official stats.

Do you believe them? I don't. Why?

Most of us know that women in Sweden are much more likely to report rape without fear of stigma or retribution. This is not the case in India or Saudi Arabia, where women can be stigmatized, even punished, for coming forward. The authorities have reason to hide their dirty laundry — especially when those among them are the ones dirtying it.

"Wait!" you might protest. "The US is not India!" True. But the dynamics are similar. When you're trying to figure out what level of racial bias exists among law enforcement authorities who also have the power and authority to hide these incidents from the public eye, you’re just not going to have a clear picture of what’s happening. We’ve seen examples of video contradicting police reports. Anecdotal? Recall from my previous letter how the number of potentially justified shootings drops dramatically when media reports are sourced in addition to police reports.

Third, there is the problem of "collider bias," which plays into many of these studies. This is when researchers' attempts to control for a common effect of the exposure and outcome end up modifying the association between the exposure and outcome. For example, you might find a strong link between two diseases in a sample of hospitalized patients, but when you look in the general population, you find none, because both diseases separately lead to hospitalization. This is a significant problem in policing studies, where the sample you're studying consists of encounters that have already occurred, not all possible encounters. The very nature of police records introduces this bias, because they only include encounters that triggered reporting in the first place. According to an excellent recent paper from Princeton, police records end up masking racial bias as a result:

"These unobserved differences can lead analysts to understate anti-minority racial bias — or even produce the appearance of anti-white bias — in the use of force."
This means that the already overwhelming evidence for systemic racism in policing may still be an under-representation of the real situation. And speaking of closely and carefully looking at the data, Laura Bronner at FiveThirtyEight has done an extremely thorough job of explaining this mathematically for interested readers.

Finally, we now know — thanks to an explosive 2017 report from NYC's Health Department that only saw the light of day this month — that the city vastly underreported police-involved deaths for five straight years. NYC reported a total of 46 police-related deaths from 2010-2015, whereas the Health Department identified 105 — more than twice what was reported. Racial disparities were significant. No unarmed whites were killed, but 6 black and 5 Hispanic people killed were unarmed. Mara Gay's reporting likely triggered the release of the report, which was effectively buried for three years — a fact that might be more shocking than the report itself.

These kinds of problems also feed into predictive policing, used to forecast criminal activity and decide where to allocate police resources. When the tributaries feeding the river are already contaminated, the river can fast become a cesspool.

Data on police shootings must be contextualized and analyzed critically, keeping in mind all the limitations outlined above. When this is done, the studies demonstrating systemic racial disparities in policing show clear advantages over those demonstrating none. The furthest you can reasonably go as a skeptic is to say that stats on police shootings specifically are ambiguous or incomplete. I can respect that. But you cannot conclusively say that these disparities don't exist. You also can't say that the vast amount of evidence demonstrating systemic bias in the criminal justice system as a whole (outlined in the previous letter) is somehow completely dissociated from the narrow focus of police shootings.

You bring up the question of whether this evidence actually demonstrates racial bias at play — is it possible that these officers are trying to prevent violent crime? This is a fair question, and I have always maintained that the majority of police officers do their work with good intentions, in good faith. But this is not relevant to this discussion. To reiterate what I said in the previous letter, we're not talking about overtly racist cops. (If this is the threshold, we will never satisfy it short of doing fMRI scans on officers as they shoot.) We are talking about a criminal justice system that was built and established on overtly racist premises, during the Jim Crow era.

Research on the 990 fatal police shootings in 2015 showed that minority victims were significantly more likely than whites to not have been attacking the officer, and black victims were more than twice as likely as whites to have been unarmed. Another study shows that police perceive young black men as more physically threatening than young white men. These "biased formidability judgments" lead them to conclude that "they must therefore be controlled using more aggressive measures." These studies, and others like them, are especially designed to evaluate racial bias on the part of officers. While it may be implicit, the evidence shows — again, overwhelmingly — that it exists.

And yes, the same system trains and produces black officers alongside their white counterparts. Black officers are not immune from these biases. Bringing up the religion analogy again, we know that there are genuine feminists who don the hijab, and openly gay conservative men who vote for politicians that deny them their rights. When biases are systemic, the behavior of individuals becomes complex. It's important to remember that racist practices can be internalized by non-racist people. The focus, again, is on the system, not the individual.

(I hope I have satisfactorily shown here that research showing racial disparities does account for offending rates, but I also want to quickly address your analogy of sex differences in violent crime rates, because this is an entirely different ballgame. It is scientifically established that males are biologically more prone to aggression than females. Although environmental factors do play a small part, this propensity is largely explained by genetic, neurobiological, and hormonal differences between the sexes. Unless you're proposing that black people are somehow biologically or genetically more prone to aggression or violence, this analogy doesn't hold.)

In closing, I want to encourage readers to understand and critically evaluate the data showing systemic racism in the criminal justice system beyond just police shootings — not just accept it blindly. They cannot be seen in isolation. The movement is called #BlackLivesMatter — not #BlackDeathsMatter — for good reason.

Ali Rizvi

Ali Rizvi

When you say “public uprisings,” if you mean peaceful demonstrations, we may have some agreement. However, if you mean looting, burning, vandalizing, and violence then we disagree. Any child can destroy things. It takes an adult to build them. I don’t believe the Weather Underground helped the anti-war movement. And I don’t think Antifa does anything other than ceaselessly work for the re-election of Donald Trump.

You’ve continually veered away from the subject of police shootings, and the even broader topic of arrest related deaths, which I am addressing, to discuss all of “systemic racism.” This is a motte-and-bailey fallacy. With 1000 words per page, we barely have the space needed to accurately discuss police-shootings. While I don’t disagree that we will want to look at the totality of the accusation, it’s equally obvious that in order to find racism within the system we need to carefully examine each part of it separately. The reluctance to do so gives the appearance of “racism” being intentionally defined in a manner so vast, and so vague, as to render it impossible to locate. That’s counter-productive.

Police murdering black Americans is the most serious accusation. As such it deserves focus. Cultivating a sense of distrust and contempt for the police is dangerous. If that fear is based on a false premise, fueling it is also morally wrong.

Focusing on homicides rules out issues of significant underreporting. Your example of rape being vastly underreported in many nations is an important topic. But unless one wants to posit a significant number of unreported murders every year, it is irrelevant to homicide statistics in the United States. I mentioned that in 2013 62.3% of the murder victims in NYC were black. There are bodies and coroner reports to account for those victims. It’s not a guess.

That number may increase. For example, on June 1st it was reported that 85 people were shot and 18 killed in Chicago—in one weekend. A few days later the number dead from the same incidents rose to 24. Last weekend, in the same city, 52 were shot and 14 are reported dead—including a 20-month-old little boy. The weekend prior more than a 100 were shot, and those killed included a 3-year-old little boy named Mehki James. Those numbers will rise as more investigations are completed and more victims die in hospitals. None of this is surprising.

The NYC Health Department report, which you labeled “explosive,” found that over a period of five years there were 59 deaths (just under 12 a year) related to legal intervention, and not recorded accurately. These included accidents, deaths in jail, booking centers, detention, and medical facilities. Some of this will have little or nothing to do with the arresting officer, but, rather, as the report states, issues of “death coding.” Their conclusion was: “Interventions to improve quality cause of death coding, including training of medical examiners to ensure that they indicate police involvement on the DC when it occurred and training nosologists.” A nosologist is a medical coder. They do data entry.

The same report notes: “given higher stop/arrest rates among non-Hispanic blacks, native Americans, and Hispanics, there was no racial/ethnic disparity in the probability of death among persons who are stopped/arrested by police.”

In other words, when they controlled for rates of offending, there was no racial bias in the numbers.

These kinds of reports are helpful and important, but reporting on them needs to be done responsibly and accurately. While it’s completely reasonable to have conversations about better policing, what people are now doing is blaming law enforcement for social inequality. That’s both unreasonable and unfair. We owe it to the people who risk their lives protecting us to answer this specific charge: “Is systemic racism reflected in officer involved killings?” An affirmative response has become the mantra of many in the media, activist groups like BLM, and academicians.

You continue by citing the Ross paper: “the probability of unarmed black Americans being shot by police is about 3.5 times that of analogous whites. You maintain that this does not account for the rate of offending. Actually, it does.”

Any study that claimed black suspects were 3.5 times more likely to be shot than white suspects and also claimed it controlled for rates of offending, should set off alarm bells. Those findings run completely counter to all data, from all sources, media, government, victims’ surveys, the Washington Post, BJS, FBI, as well as peer-reviewed studies using rigorous methodologies. My contention was Ross’ study did not account for rates of offending. You state: “Actually, it does.” And then go on to quote it:

"There is no relationship between county-level racial bias in police shootings and crime rates (even race-specific crime rates), meaning that the racial bias observed in police shootings in this data set is not explainable as a response to local-level crime rates."

People familiar with the data might immediately wonder how one study could come up with findings so contrary to what the data tells us. The answer is, it didn’t.

Pay attention to the last two words in the part you cited from the abstract—crime rates. No one, including me, is claiming that knowing the crime rate in a given county will explain differences in the ratio of police shootings of black suspects versus white suspects. Why would it?

What I have been arguing, and I think I’ve made this clear, is that if you control for a given demographics’ rates of offending, “bias” disappears. And it does. To control for rates of offending, and therefore find bias in police shootings, Ross would have had to compare the rate of crime committed by blacks in that area to the rate of crime committed by whites and other races in that area. He didn’t.

When someone says black Americans are some percentage more likely to be shot by police than white Americans, without also mentioning the higher rates of offending, it’s half-true and wholly dishonest. Many in the media do this daily. When someone brings up black-on-black crime, those same media personalities will counter with “most crime is intraracial,” without ever mentioning that crime occurs at much higher rates in the black community—which is, of course, the point. Anyone who has studied the topic will be familiar with this sort of dishonesty. When someone points out that if you control for crime rates bias disappears, what they mean is a comparison of black versus white or Hispanic rates. Without the comparison, how else could one find bias? This abstract was worded in such a way as to imply it did make that comparison, when, in fact, it did not. If the facts related to police shootings really were on the side of those who claim racial bias, why not be forthright?

In fairness, Ali, I don’t believe you were being dishonest. The way the abstract is worded is misleading.

There are also other problems with this study. For his data, Ross used a blog called Deadspin, a crowd sourced website where anyone can add police shootings. While the goal is the elimination of bias, the site created a new problem as it relied on media reports and Google’s algorithm. Furthermore, your mention of collider-bias is interesting, and something for those who design studies to consider. It could also be at play in every study you’ve cited, including this one.

What level of bias exists among media leaders who have the power to put these incidents in front of the public-eye? Shootings that involve white officers and black suspects, such as the Michael Brown case, tend to make the national news. Shootings that involve a black officer or white suspects tend to remain local. Finally, at the time of Ross’ analyses, only half of the calendar days on Deadspin’s blog had been searched. What’s instructive is that the Washington Post, New York Times, and other large media outfits then went on to report a study whose findings were based solely on stories in the Washington Post, New York Times, and other large media outlets.

It’s not a secret that high profile police shootings of poor white suspects receive far less coverage than those involving black suspects. Study after study (see citations below) confirm that white officers are less, not more, likely to shoot black suspects versus white ones. Yet most in the media consistently state the opposite. And it isn’t just the amount of coverage that’s different. It is also the nature of it. When Alton Sterling, a man who was threatening someone with a gun, had previously resisted arrest and was found to have a gun, and who had a long record of convictions for things like child-molestation and battery (beating women), was killed, the coverage told a story of a “gentle-giant” who was guilty of little more than selling CD’s. The New York Times ran a piece titled: “When Black Lives Stop Mattering.” When Dylan Noble, a 19-year-old, unarmed, white-male, was killed during a traffic stop in Fresno, the New York Times story was titled: “Fresno Police Shooting shows Dylan Noble Ignoring Orders to Stop.” I could list case after case, but I lack the space, and I don’t think it’s necessary. The fact is that 80% of all people shot by police are white or Latino and they receive substantially less press coverage than their black counterparts.

Any honest look at the individual cases BLM mentions, and I strongly suggest readers take the time to do that, will find little to no evidence for racially motivated murders. Instead you’ll see cases like Michael Brown, where half-a-dozen black witnesses and forensic evidence confirm that Brown attacked officer Wilson and tried to take his gun. At what point do people become responsible for their actions? If Michael Brown had not attacked a police officer and had not tried to grab his gun, he’d be alive right now. And I think most Americans, black and white, understand that.

Once you actually look at the evidence, read past the abstracts, past the headlines, past the current moral panic, you see that George Floyd’s death isn’t an indication of any kind of trend or bias; to the contrary, George Floyd’s death stands out because in a nation of 330 million people, and 375 million police contacts with the public every year, it’s actually an aberration. It is incredibly rare.

Ali, you cited (I have not) the Fryer report. Let’s compare the Ross study to Fryer’s paper. His dataset had detailed and complete coverage of 500 police shootings from 2000–2015, which he also compared to a random sample from the same time period in which lethal force could have been justified but wasn't used. Fryer’s conclusion was that blacks, compared to whites, were actually 25 percent less likely to be shot by police in such encounters. That conclusion matches all the data we have, going back decades.

Finally, your comment: “Unless you're proposing that black people are somehow biologically or genetically more prone to aggression or violence, this analogy doesn't hold,” was disingenuous. The analogy was about what I’ve been arguing all along—rates of offending—and I am quite sure you know that. I noticed you tweeted an apology for this comment. I appreciate that. We have a valuable opportunity to show that two people with different views can still have a constructive dialogue.

When you control for rates of offending – something that must be done to answer the question, is systemic racism reflected in officer involved killings? – racial variance in those shooting numbers is wholly explained.

This data does not address sentencing or incarceration terms. I’ve not claimed it does. Those issues are also out of the hands of police officers.

To analyze those topics we would need to apply the same careful approach, controlling for factors that may help explain differences. If we don’t, if instead we assume that any distinction in the numbers or discrepancy in outcome must be racism, then not only have we failed to answer the question, we have not even tried. The Creationists fallacy of the “God of the gaps,” while convincing to some, has never reflected a serious attempt at a truthful answer. Assuming any distinction in the numbers to be racism, without bothering to control for other factors like rates of police contact, is no more than a “racism of the gaps” fallacy.

The narrative that police officers are looking to kill black Americans is a pernicious lie. In cities across the country, regressive anti-policing policies are being rushed in. Because homicides within the black community occur at 3.9 times the national average, the people who will suffer most from these changes won’t be upper-middle-class, urban-elites who foolishly push them through. It will be poor, black Americans who live in the kinds of areas that 3-year-old Mehki James was murdered in just a few days ago. Black Americans consistently poll higher than whites in wanting an increased police presence. Cutting police funding isn’t going to solve our violence problem, it’s going to increase it.

We can want to save black lives and not support BLM. The latter is, as Peter Boghossian and James Lindsay have repeatedly pointed out, a rather obvious Trojan Horse for Critical Race Theory and an ideology that’s antithetical to liberalism and the civil rights movement. We can acknowledge that few people care more about black lives than the police officers who risk their own to help protect them. And we can take the time to read past the abstracts, examine the data, and then tell the truth. If we want to save lives, good intentions will never be enough. We also have to be right.

As this is my last reply, I want to thank Clyde and Dayne, and everyone at Letter, for hosting our conversation. It’s important. And Ali, my praise for your book at the start of our conversation was sincere. You are a talented writer and brave human. Thank you for taking the time to engage with me on this topic.

Citations and studies:

The findings in these studies are more complex than the brief descriptions I have space for. I also want to note that throughout this debate I’ve not cited any studies, Fryer’s or otherwise, to make my argument. I believe strongly that a comparison of rates of offending alone provides enough of a reasonable, cogent, explanatory-mechanism so as to rule out systemic-racial-bias in police shootings.

- The Washington Post police shooting database

- A 2019 study in Proceedings of the National Academy of Sciences found it is a racial group’s rate of violent crime that determines police shootings, not the race of the officer. The more frequently officers encounter violent suspects from any given racial group, the greater the chance that members of that racial group will be shot by police. In fact, if there is a bias in police shootings after crime rates are taken into account, it is against white civilians.

- A 2019 study in Public Administration Review found white officers no more likely to use lethal force against minorities than nonwhite officers.

- A 2018 study in Social Psychological and Personality Science found that adjusting for rates of offending, we find no systematic evidence of anti-black disparities in fatal shootings, and fatal shootings of unarmed citizens,.

- A 2016 Crime Prevention Research Center study found the probability of an unarmed black suspect being killed by a white police officer is not significantly greater than the probability of a black suspect being killed by a black police officer.

- A 2016 Criminology and Public Policy study found officers were slower to shoot armed black suspects than armed white suspects, and less likely to shoot unarmed black suspects than unarmed white suspects. To quote: “These findings challenge the assumption that implicit racial bias affects police behavior in deadly encounters with Black suspects.”

- A 2015 Criminology and Public Policy study found that the racial composition of neighborhoods nor their level of economic disadvantage directly increase the frequency of police shootings, whereas levels of violent crime do.

- This 2007 University of Chicago study compared police with community members in a simulation to find bias in the decision to shoot. Community members set the decision criterion lower for blacks than for white suspects (indicating bias). Police did not. *(This might be of interest to those wanting to replace professional law-enforcement.)

- This 2016 study in the British Medical Journal studied 10,000 stops/arrests where there were injuries resulting from legal intervention by US law enforcement. They found that rates did not differ significantly between racial/ethnic groups.

- This 2020 Justice Quarterly study states: “The key finding was that black suspects were no more or less likely to have weapons drawn against them than other suspects.”

- This 2014 study in Social and Personality Psychology Compass compared a group of undergrads to police to see if they would be more likely to shoot a black suspect. The undergrads were, the officers were not. *(Again, this might be of interest to those wanting to replace professional law-enforcement.)

- This 2020 study from Race and Justice found black homicide suspects are not more likely than similarly situated White homicide suspects to be arrested.

- Here is the 2018 Harvard professor Ronald Fryer’s paper found no bias in officer related shootings.

And Fryer’s follow up paper, where he found blacks are 27.4% less likely to be shot at by police relative to non-black, non-Hispanics.

Matt Thornton

Matt Thornton

Thanks for your letter. In this final letter, I want to both summarize my position in this exchange and respond to some of the points in your last response.

You write:

"You’ve continually veered away from the subject of police shootings, and the even broader topic of arrest related deaths, which I am addressing, to discuss all of “systemic racism.”

I have not "veered away" from the subject. What I have done is contextualize police shootings within the larger epidemic of systemic racism. I have explained why this must be done. Among other reasons:

1. Many of the most prominent high-profile cases being discussed — George Floyd, Trayvon Martin, Rodney King, Freddie Gray, Eric Garner — are not cases of police shootings;

2. Police shootings are rare (as you also pointed out) and the data on them is not only ambiguous, but comes with many serious limitations that I outlined in detail in my previous response;

3. Police shootings simply cannot be divorced from the larger problem of systemic racism in the criminal justice system, where the data for racial disparities is unequivocal — whether it's encounters, physical force, sentencing, or bail — as I've shown via links to many studies in my previous responses.

In this final response, I want to expand more on 2 and 3.

We've talked at length about the Fryer and Ross studies. You referenced the 2019 Johnson and Cesario PNAS study as a primary source on how racial bias disappears when one adjusts for rates of offending. This is one of the most widely cited studies among systemic racism-skeptics like Heather Mac Donald.

As you likely know, shortly after you sent your last letter, this study was officially retracted by its authors.

The retraction has caused outrage among some conservatives, including Mac Donald herself, who alleged that the authors and the National Academy of Sciences succumbed to political mob pressure. This is a superficial and objectively faulty assumption. Here's why.

The Johnson paper had been widely criticized well before the current protests. Criticism from Mummolo and Knox in January prompted the authors to issue a correction around one of their central claims. A second criticism from Schimmack and Carlsson showed a vastly different conclusion from the same data when age and suicidality status were considered. In a more detailed version of this criticism from September 2019, Schimmack wrote:

"[T]his article should not be used to support empirical claims about police shootings. Ideally, the article would be retracted. At a minimum, PNAS should publish a notice of concern."

The authors have now retracted the study. And they insist, correctly so, that the retraction wasn't due to politics, but because they were "careless when describing the inferences that could be made from our data," and that the data cannot "be used to support the position that the probability of being shot by police did not differ between Black and White Americans."

Moreover, an article by Princeton's Mummolo — explaining why Johnson and Cesario's claims of no racial bias in policing "rest on flawed science" — has now been signed by an astounding 871 academics and researchers to date. The signatories come from a variety of disciplines, from criminal justice and economics to medicine, engineering, and statistics. Specifically, all 871 academics agree that the study “violates a central axiom of data analysis": Bayes' theorem.

Yet, despite all of the above, Heather Mac Donald — who is not a scientist, social scientist, or statistician — continues to believe that the retraction was because of her.
To step outside all of this for a moment, this is pathognomonic of a central problem that we face today.

On the one hand, we lament the lack of "honest conversations."

On the other, we continue to insist that this study — which (i) triggered multiple published criticisms from academic researchers, (ii) is deemed scientifically and mathematically flawed by at least 871 academics from a wide array of disciplines, and (iii) was first corrected, and then retracted in full by the authors themselves — is somehow still legitimate because of some kind of "mob" conspiracy?

Honest question: which side here is really engaging in mob-style groupthink?

Aside from a thorough debunking of the data analysis — which should be sufficient in itself — what is the threshold for a legitimate refutation that Mac Donald would accept? 1,000 signatures? 10,000?

Now, the authors did have concerns about their conclusions being misrepresented and misused, as academics often do. Roland Fryer, author of the other widely cited (though non peer-reviewed) study on police shootings which I wrote about at length in my last response, echoed similar concerns about his work in a recent WSJ op-ed:

"To my dismay, this work has been widely misrepresented and misused by people on both sides of the ideological aisle. It has been wrongly cited as evidence that there is no racism in policing, that football players have no right to kneel during the national anthem, and that the police should shoot black people more often."

There is a strange contention in the discourse today that academics and researchers are bowing to the mob, while activists and ideologues are "telling it like it is." I believe it's the other way around.

Now, let's set aside the retraction and look at the point itself around "rates of offending." The argument is that because black people commit disproportionately more crimes, it is reasonable for us to expect that they will be shot by police at disproportionate rates. "If you control for a given demographics’ rates of offending," you write, "'bias' disappears."
The problem with this position is that it ignores a key part of the story: police encounters. The majority of killings by police don't happen when cops are trying to stop a murder. They happen during routine police work like traffic stops (most common), arrests for non-violent crimes, or responses to domestic disturbances.

This is why it's absolutely key to understand thatwe simply cannot divorce police shootings from the larger epidemic of systemic racism.

I'll explain.

Let's suppose you're right about race not being a factor in the likelihood of being shot by police. In other words, once an encounter has been initiated between police and a suspect, the likelihood of the suspect being shot would not be based on the racial demographic they belong to.

Let’s also acknowledge the obvious fact that every shooting begins with an encounter.
Herein lies the problem: the data on police encounters unequivocally shows systemic racial bias:

We know via innumerable studies that blacks are more likely to be stopped by police than whites.

Once stopped, they are more likely to be searched than whites.

Once searched, they are less likely to be carrying contraband than whites.

Once arrested, they are more likely to be charged, incarcerated, and receive longer sentences for analogous offenses than whites.

Blacks are more likely to be subjected to physical force by police, even when correcting for offense rates.

They are less likely to resist arrest, and more likely to be unarmed.

Young black men are more likely to be seen as threatening by police than young white men.

We know that black people reported as compliant in police reports were 21% more likely to suffer police aggression than compliant whites. That’s right. According to Roland Fryer's data, compliance benefits whites significantly more than blacks.

The point here is simple. You could isolate police shootings and find no racial disparities, as some studies have done. But most fatalities begin as encounters, where blacks are disproportionately disadvantaged. It logically follows that if systemic racism plays a role in encounters — which it does — then it also plays a role in the subsequent shootings that result from them.

As physicist Neil deGrasse Tyson wrote:

"[I]f your demographic gets stopped ten times more than others, then your demographic will die at ten times the rate."

And from the Mummolo article signed by 871 academics:

“To begin to measure racial bias in police violence, careful researchers must ask how often officers use force against minority civilians out of all police encounters, versus white civilians, then adjust for relevant differences between minority and white encounters. Comprehensive records of lethal force — the numerator in the force-per-encounter ratio — only recently became available. ...The denominator — how often racial groups encounter police — is largely unknown.

Rather than analyzing shootings as a fraction of all encounters, [the PNAS study] analyzed only shootings. This elementary error — only examining cases where events of interest occur — is called ‘selection on the dependent variable,’ and is one of the first mistakes social scientists are warned about during academic training.”

This illustrates what has been at the heart of my argument. If we really want to evaluate systemic racism in police shootings, we shouldn't be looking at racial disparities among those who were shot; we should be looking at racial disparities in police encounters — which ultimately select who gets shot.

We cannot talk about systemic racism in police shootings without the systemic part. The context isn't just important; it’s wholly inextricable from what we're discussing.
In closing, I want to point out some places I hope we can agree.

Like you, I support peaceful protests and oppose violence, looting, and vandalism. Now, I also understand that someone who has lost a son, brother, father, or husband to police brutality may not have the privilege of mental and emotional clarity that we might. I am sympathetic to that. But violence is counterproductive. I was moved to see George Floyd's family denounce it, as did Obama, Mayor Keisha Bottoms, Killer Mike, and others. And I have spoken to #BlackLivesMatter activists who are frustrated with Antifa, a group that Noam Chomsky has correctly described as "a gift to the far-right."

But I also disagree vehemently with those who want to make violence and looting the face of BLM. This is now the largest protest in American history, and the vast majority of protesters have been peaceful. Yes, every worthwhile movement has its share of opportunists and assholes. I passionately support the atheist and secular movements in the Muslim world, both of which have their share of odious characters as well. That doesn't take away from my support for the movements.

I think we can align on this: what happened to George Floyd at the hands of police — and what happened to Tony Timpa at the hands of police — should never happen to anyone. There should be no number of incidents like this that a civilized society should let pass just because they're "rare."

Systemic racism disproportionately targets black people, but police brutality targets everyone. It is in the interest of all of us — white, brown, or black — to fight it. #BlackLivesMatter is admittedly decentralized, with some subgroups pushing agendas I don't agree with. This has been true for many transformative movements in history. But I support the broader changes they align on: dramatic police reform, criminal justice reform, investment in public education, social work, and mental health resources, and more. This benefits all of us. And #BlackLivesMatter is leading the fight to make it happen.

Matt, I deeply appreciate this exchange. I've always preferred written conversations because they allow us to think, research, and reflect. This has certainly been the case here. We may disagree, and we weren't able to address every point the other made, but I really believe we have helped our readers navigate this subject in a way few others do. Thank you.

Clyde, Dayne, Iona, and everyone else at LetterWiki, thank you for hosting this. Your incredible work is ever more crucial in these crazy times.

For more, I'd like to refer readers to the excellent work of Radley Balko at WaPo. Thanks also to Habib Fanny, Will Comerford, Fay Rahman, Maryam Namazie, and Armin Navabi for their engagement and feedback.

Ali Rizvi

Ali Rizvi