Interracial violence is rare in the United States
It just constantly gets shoved in our faces.
Interracial violence is a difficult thing to write about, for the exact same reason that people get so upset by it in the first place. It conjures up the dreaded specter of intercommunal violence — massed attacks of one group on another. Race riots. The Tulsa Massacre. The Gujarat riots. The pogroms of old Russia. The Rwandan genocide. And so on. When we read a news story about one of them killing one of us, it raises the gnawing fear — articulated or not — that the civilized world of laws and police and the courts is just a veneer, a thin wrapping that will soon be ripped off to expose the monstrous primal world beneath, an anarchy where our only protection is the law of kin and clan. A world of them against us, in the streets, to the death.
This ever-present fear presents an especially difficult challenge for a very diverse society like the United States. All societies have episodes of individual violence, and in a diverse society, it’s a statistical certainty that some of those episodes will be interracial or inter-religious. Because of the lurking fear of intercommunal violence, interracial murders, assaults, rapes, and so on tend to get broadcast and politically weaponized in a way that violence between people of the same race doesn’t. The more diverse society becomes, the more such incidents there will be to feed our fears, whether they’re random or systematic.
One way people try to deal with this is to look at the aggregate statistics. Are “they” killing more of “us”, or are “we” killing more of “them”? But aggregate statistics are very subtle things. There are issues of data quality and completeness. There’s the question of which measure to use. There are all kinds of contributing factors that simple summary statistics fail to take into account. So the debates about these numbers can often be as bitter and fearful and full of assumptions as the debates about individual incidents of violence.
That fact was showcased this week, in the wake of an incident where a Black man, Jordan Neely, was choked to death on the New York subway by a White man (with some assistance from two other men of varying races). There were protests about the killing, but on Twitter, some protested the media’s focus on the incident in lieu of other incidents of Black-on-White violence. A chart from the right-wing account “End Wokeness” got an especially large amount of attention:
Statistics PhD student Kareem Carr criticized the chart in a thread. Among other criticisms, he pointed out that the vast majority of violent crime is same-race:
But because same-race violence doesn’t create the same fears that interracial violence does, I fear that Carr is swimming against the tide on this one.
Anyway, Carr’s thread adds some important points to the discussion, but I feel like it’s still easy for readers to walk away from the exchange with all their priors — and all their fears — intact. In fact, I’ve looked at the interracial violence numbers myself, and I feel like I can summarize what we know about interracial violence pretty effectively. My basic findings are:
Interracial violence is uncommon in America, and interracial murder is very rare.
White-on-Black violence is much rarer than Black-on-White violence, though the gap is much smaller for murders than for non-fatal violence.
Black people are more at risk of being murdered by White people than vice versa, while the risk for non-fatal interracial violence is about the same.
How can these last two things both be true at the same time? The answer is that the Black population is simply much smaller than the White population, so the number of attacks isn’t the same as the risk of being attacked.
So anyway, let’s dig into the data.
Murder
There are two main data sources on interracial violence in the U.S.: the FBI’s Uniform Crime Reports, which collects reports from law enforcement agencies, and the Census Bureau’s National Crime Victimization Survey, which asks people whether they were victims of a crime. Obviously only the first of these covers murder, because you can’t really ask people whether they’ve been murdered or not. So let’s look at the UCR. Data on interracial homicides is contained in Table 6.
Now, there are a couple of things to note about this data. First, it’s for 2019, which is the most recent year available, but was also before the big murder spike that began in 2020. Second, this data breaks out Hispanic/non-Hispanic, but it doesn’t break out White Hispanics from Black Hispanics, so when we calculate the numbers for White and Black murderers and victims using this table, we’re going to have to include Hispanics in both of those.
Third, and most importantly, this table doesn’t include all murders; it only includes those for which:
an arrest was made, and
there was only one murder and one victim.
There were 16,425 reported murders in the U.S. in 2019, but 6,578 that fit the two criteria above. So the numbers we’re looking at represent only about 40% of the murders in the country. Which means in order to draw general conclusions, we’re going to need to make some assumptions about the other 60%. But anyway, let’s see what the data shows:
So we can see that out of 6,578 murders, 566 were Black-on-White murders, and 246 were White-on-Black murders. If that holds for all of the murders in the country, then Black-on-White murder is about 2.3 times as common as White-on-Black murder. Whether or not the true number is lower or higher depends on whether there are more White-on-Black or Black-on-White murders that go unreported. If you think murders with a Black victim are less likely to get reported, for example, then the true number will be less than 2.3.
But anyway, this doesn’t tell us the answer to another question we’d like to know: How likely are you to get murdered by someone of another race? Here, we have to divide the total murder numbers by population. In 2019, the White population of the U.S. (alone or in combination with other races) was estimated at 258.65 million, while the Black population (alone or in combination) was 48.22 million. So those are the numbers we have to divide by.
Dividing by those numbers, we get:
Black-on-White murders as a percentage of White population = 566/258646488 = 0.22 per 100,000 White people murdered by Black people
White-on-Black murders as a percentage of Black population = 246/48221139 = 0.51 per 100,000 Black people murdered by White people
The first thing to note is that these are both very small numbers. They’re comparable to the overall murder rates in South Korea or Japan. Now, remember that this is only 40% of the murders in the U.S. But even if we scale the numbers up by a factor of 2.5, that’s 0.55 per 100,000 for Black-on-White murder and 1.28 per 100,000 for White-on-Black murder.
Interracial murder is rare in the United States. If you’re a White person in the U.S., you’re probably less likely to get murdered by a Black person than a South Korean person is to get murdered by another South Korean. And if we assume that the murders that don’t make into Table 6 — because there was no arrest made or because there were multiple victims or multiple offenders — are more likely to have Black victims, or more likely to be same-race in general, then the true Black-on-White murder risk is even lower than 0.55 per 100,000 — maybe even as low as the overall murder rate in Japan.
In fact, if you take all the interracial murders of every kind in Table 6, and assume that all the murders not in Table 6 are equally likely to be interracial, and divide by the total U.S. population, you’ll find that the total interracial murder rate in the U.S. is a bit under 1 per 100,000 — less than the overall murder rate in Denmark. Of course this risk depends on where you live, what you do, etc. But the basic point stands: Interracial murder is just a rare thing in the United States.
The next thing to note is while no group is in particularly great danger from interracial murder in the U.S., Black people are at greater risk of being murdered by White people in America than vice versa. The reason this is true, despite the fact that Black-on-White murders are more numerous than White-on-Black, is that there are simply a lot more White people than Black people in America to begin with — about 5 times as many. And remember, this is not even counting police killings, which aren’t included in Table 6, and which disproportionately kill Black people.
Let me sum it all up in a chart:
In other words, looking at these numbers, it’s easy to see why both White and Black Americans might feel under siege from reports of interracial murder. For White people, there are a greater number of such reports to trigger their availability heuristic and make them think Black people are out to get them. And for Black people, each report might serve as a reminder that they’re a minority, and would be massively outnumbered in any actual intercommunal violence.
It would help, of course, if Americans could realize just how rare any sort of interracial murder is — to know that they’re as safe from interracial murder as Danish people are from murder in general. But putting threats in perspective has never exactly been Americans’ strong suit.
Non-fatal violence
Now let’s talk about non-fatal violence, which is what the Twitter debate was about. Here we go to the National Crime Victimization Survey for 2019. There’s a table very similar to the UCR’s Table 6:
There are two big differences here. First, the NCVS separates Hispanic from White and Black. So now, when we talk about White, we’re no longer including White Hispanics, and when we talk about Black, we’re no longer including Black Hispanics. The second difference is that this table includes 84% of all reported crimes (a lot better than the UCR’s 40%!).
It’s also important to remember that the NCVS is survey data — it comes from asking people whether they were victims of a crime, instead of reporting arrests by the police. You can decide for yourself how accurate you think people’s memories are, or how truthful they are on surveys.
But anyway, we can see somewhat similar patterns here to the FBI’s murder numbers. Black-on-White violence is a lot more common than White-on-Black violence — about 5.2 times as common. But because there are a lot more White people than Black people, the risk of violence is a lot more even. The Census Bureau tells us that in 2019 there were 203,768,170 non-Hispanic White Americans and 44,419,767 non-Hispanic Black Americans. So repeating the exercise from the section above,
Black-on-White nonfatal violent incidents as a percentage of White population = 472570/203768170 = 232 per 100,000
White-on-Black nonfatal violent incidents as a percentage of Black population = 89980/44419767 = 203 per 100,000
Here, unlike in the case of murder, White people are at very slightly higher risk from violence at the hands of Black people than vice versa. But the numbers are pretty similar!
I can’t compare these numbers to other countries, because I’m not sure that other countries’ surveys are similar. But again, as with murder, we can see the risk of same-race violence is generally higher than the risk of interracial violence. Here’s a table that sums it up very well:
For White people, 38.1% of violent crime comes at the hands of non-White people — almost exactly the same as the proportion of non-White people in the population. About half of that is at the hands of Black people.
But for Black people, only 30% of violent crime comes at the hands of non-Black people, even though non-Black people are 87.1% of the population! (Only for Hispanics does the pattern not hold; more than half of the violence they experience comes at the hands of non-Hispanic people.)
In other words, Black attacks on White people are a relatively small percent of the violence White people experience. And White attacks on Black people are a relatively small percent of the violence Black people experience. We pay a vastly inordinate amount of attention to these incidents of interracial violence, relative to how frequently they actually occur.
Ultimately, that attention worries me even more than the violence itself. We don’t yet know what kind of societies will be stable in the age of social media. We would like — or at least, I would like — diverse societies like the United States to be stable. And intercommunal violence seems like one of the most plausible failure modes for a diverse society. If Americans pay too much heed to the attention-seeking social media shouters constantly trying to use every random act of interracial violence to drum up fears of race war, those fears could become a self-fulfilling prophecy, and our country could come apart at the seams. Even if it doesn’t spark actual riots and pogroms, it could lead to the breakdown of social cooperation between people of different racial groups, and paralyze the government via racial polarization of politics.
In other words, in order to succeed as a diverse society, we have to find some way to remind people that interracial violence is actually rare, and that the incidents that get shoved in their face on social media are being selected in order to fan their outrage and fears. A diverse country will always have isolated cases of violence that happen to occur between people of different races. That’s not a race war. And if we keep our cool, we can make sure it never turns into one.
Just seems to make sense that violence requires proximity and most people's social circles (and neighborhoods) are more racially homogenous than expected by random distribution.
More illuminating data (to me) would be rates of violence across class lines. Even the Cash App founder's murder was by someone in his social circle.
So who shoves it in our faces and why?
Must be somebody shoving and must be for s reason.