Black Lives Matter: racism at unexpected places

Recently, in the wake of the Black Lives Matter protests, I stumbled upon a graph about interracial violent crime incidents in the US in 2018. That graph was spread by conservative, right-wing and racist people. The original source, containing the underlying data of the graph, was the US Department of Justice. The data were surprising to me, and demonstrate how complex the issue of racism is. I am not an expert on this topic, I am not a criminologist or sociologist and haven’t read many scientific studies on this topic. Nevertheless, I think this topic is a good exercise in how views can be influenced by briefly looking at some studies, statistics and graphs. This example of interracial violence shows that there are plenty of hypotheses and surprising conclusions that one could draw from only a small graph. It is also a good exercise in rational, critical thinking.

As I will show, police violence and violent crime statistics do reveal racism at several places, but not at the places that I or many other people would expect. This implies that the causes of racism could be different or more complex than what one would naively guess. At least, the situation is more complex than I would have guessed.  

To be clear, I focus on statistical racism: the shares of crime rates by ethnicity compared to what one would expect based on population shares. For example, 60% of the US population are non-Hispanic whites, 13% are black and 17% are Hispanic. Hence, without statistical racism, one would expect 13% of victims and 13% of offenders of crimes to be black, and 1,7% (i.e. 13% times 13%) of crimes to be with black offenders on black victims. Statistical racism could be explained by, but should not be confused with e.g. ideological racism, institutional racism or systemic racism. If people have racist beliefs (ideological racism), that could influence the crime statistics and generate statistical racism, but statistical racism is possible even if people do not have explicit racist ideologies or implicit (unconscious) racist attitudes.

Let’s move to the graph that was spread by the racist people. That graph selected only the interracial violent crime rates of three ethnic groups: whites, blacks and Hispanics. Here I present the full graph, that also shows the intraracial crime rates of those three ethnic groups. The left, blue bars represent the expected number of crimes if there was no statistical racism. The right, orange bars are the real crime rates.

Some findings.

• There are more interracial crimes by black offenders on white victims (‘black on white’) than the reverse (‘white on black’), and these crime rates are far from what one would expect based on population shares. For example the crimes by white offenders on black victims is 7 times lower than black on white, and 10 times lower than what one would expect. This was the point made by the right-wing, conservative and racist people spreading the graph. One could conclude that, when it comes to interracial violence, there is no sign of white supremacy or antiblack racism by white people. White and Hispanic people are the least dangerous people for other ethnic groups.
• Blacks are statistically speaking more violent than whites or Hispanics. The total number of crimes by black offenders (the sum of the three left orange bars) is almost twice as much as what one would expect based on population shares (the three left blue bars). Whites and Hispanics are 10% less violent than what one would expect. The higher crime rates by black people could be due to misreporting (e.g. when people are more likely to report a crime when the offender is black) or a racist bias (e.g. police officers misreporting the ethnicity of the crime suspects). However, mere misreporting or bias could not explain the discrepancy between crime rates for different ethnic group victims. Looking at the graph: why would the racist misreporting be so much larger when the victims are black (look at the discrepancy between expected and real ‘black on black’ and ‘white on black’ rates), compared to when the victims are white? The ratio of real versus expected crime numbers for white on white is almost the same as for black on white, whereas those ratios are extremely different for white on black and black on black. Hence, something more than mere misreporting must be going on. I don’t expect that correcting for misreporting would significantly change the conclusions we can draw.
• Intraracial violence (within the same ethnic group) is higher than interracial violence (between groups), and this is true for all ethnic groups. For example crimes by black offenders on black victims are higher than what one would expect. There could be several reasons for this, such as the formation of ethnic communities. For example, black people tend to live in neighborhoods with many other black people, hence encountering more black than white people in their daily lives, and hence encountering more black offenders and victims.
• Black intraracial violence (‘black on black’) is almost 5 times higher than what one would expect, whereas white intraracial violence is only 10% higher. I don’t know why this is the case, but based on my economics knowledge, I can formulate some hypotheses. For example the black community can have relatively high levels of intragroup competition. The labor force of a minority group can be relatively more homogeneous, for example comprising of low-skilled workers, which means jobs are less complementary, which means more intragroup competition on the labor market, which means lower wages and incomes, which means dire economic situations, which means more resort to intragroup violent crimes against people considered as competitors.  
• The total crimes on minority groups (black people and Hispanics) is 17% to 19% lower than what one would expect, whereas the total crimes on whites is 9% higher than expected. This means minority groups are relatively safer against crimes. Even with a very high black on black crime rate, black people are less likely to be victims of crimes compared to a situation without statistical racism. (This conclusion may no longer be true if crimes against blacks happen to be underreported in the data. Perhaps the number of white on black crimes in the data is underestimated, but then one would expect that the number of black on black crimes is also an underestimate, which means in reality the intraracial black crime rate is even higher than the already very high level presented in the data.)

Now we can look at a measure of statistical anti-black racism. I define this as a ratio of two ratios. The first ratio is the real number of crimes with black victims to the real total number of crimes (i.e. summed over all victims). The second ratio is the expected number of crimes with black victims to the expected total number of crimes (expected based on population shares). When this ratio of ratios is larger than 1, there is statistical racism. This statistical anti-black racism measure corrects for e.g. the facts that some ethnic groups are larger than others or more violent than others. Here we see really surprising results.

• The statistical anti-black racism by the whole population is 0,75, i.e. lower than 1. Hence, the whole population does not have an anti-black racism bias when it comes to violent crimes. This can also be compared with the statistical anti-white racism by the whole population, which equals 1,15. This value is higher than one, so in general there is rather an anti-white bias. Whites are relatively worse-off than blacks when it comes to statistical racism in violent crimes.
• The statistical anti-black racism by whites (i.e. considering white offenders) is even much lower: 0,17 (consistent with a higher than 1 anti-white racism by white people of 1,33).
• Most surprisingly, the statistical anti-black racism by blacks is 2,7. This is extremely high. It is also higher than the statistical anti-white racism by whites (1,33). This means that the abovementioned ethnic community effect (blacks living among blacks, whites living among whites) cannot explain this anti-black racism by black people. When it comes to violent crimes, the real racism against black people comes from black people. I suspect that there is a socio-economic cause underlying these high levels of statistical anti-black racism by blacks and black intraracial violence. The lower socio-economic status of black people could partially explain the statistical racism, and could itself be the result of structural racism (e.g. structural racism on the labor market), but below I present another interesting hypothesis that is not related to racism (or only very indirectly).

The Black Lives Matter movement also focuses on police killings. Here the data and studies are very clear: of the people shot to death by police officers in the US in 2019, 31% were black (totaling 235 black people killed), 48% were white, 21% were Hispanic (neglecting the other ethnic groups for simplicity). 31% is off course higher than the population share of blacks (13%), but also higher than the share of violent crimes by blacks (25%, considering only crimes by blacks, whites and Hispanics). Even if black people commit more violent crimes, the police force is even more violent against blacks than that.

The statistical anti-black racism of the police force cannot be explained by the ethnic composition of the police force: roughly 13% of police officers are black, which equals the population share of blacks in the US. So the next question is whether white police officers are more anti-black racist than black officers. I could not find convincing data for a higher anti-black racism among white officers. One study says that “white officers appear to be no more likely to use lethal force against minorities than nonwhite officers”. Another study and aftermath discussion indicates that… the matter is very complicated.

At this moment, I personally don’t think that white officers are significantly more racist. The statistical racism of the police force is not at the level of the individual officers. It is more hidden and structural. One hypothesis (that I didn’t check) that could explain the statistical anti-black racism, is that the police patrols more in black neighborhoods, for example because they expect higher crime rates there. Or the police can target more black than white suspects. Also (and perhaps especially so) black police officers are send to the black communities. That means police officers (including black officers) are more likely being confronted with black offenders and black suspects on the streets.

The good news is: general crime rates, and rates of police killings are declining. People and police officers, be they black or white or Hispanic, become less violent. There are some interesting hypotheses that could explain not only this declining trend in violence and crimes, but also the higher crime rate by black people, why their decline is lagging behind the decline of the crime rate by white people. One of my favorite hypotheses is the lead-crime hypothesis (also discussed here and here). The idea is that black people live at places with higher levels of lead pollution. Lead uptake in the body by young children can cause a decrease in IQ and learning abilities (hence a decrease in socio-economic status later in life) and a decrease in impulse control. This makes black people more vulnerable to violent crimes. Lead pollution is decreasing, so we see a decreasing crime rate by black people.

The lead-crime hypothesis is an interesting example of why we might need to look for far deeper, less trivial causes of violence and crimes than the usual antiracism rhetoric and ‘easy’ explanations (e.g. about white supremacy or white privilege; these are notions that could be useful in other contexts, but not in the context of violent crimes). If true, hypotheses like the lead-crime hypothesis offer much clearer and effective solutions to decrease crime rates and hence police killings.

What other solutions could be effective? A possible explanation for the high anti-black killing rate by police officers, is the higher police patrol rate in black communities. Perhaps we should send police forces more to white neighborhoods instead of black neighborhoods? However, this might increase the crime rate in black neighborhoods to even higher levels than they already are. (I am not familiar with the scientific literature, so I don’t know whether this is true.) A better option might be to have more police officers. When there are only a few officers, they have to work long shifts, hence they become more tired, more easily frustrated and less able to deal with conflictual situations in non-violent ways. Campaign Zero proposes more measures to reduce police violence.

Perhaps most importantly, in terms of effectiveness: we need a criminal justice reform, especially in the US. Chloe Cockburn and the Open Philanthropy make some interesting recommendations (see also here and here for discussions about effective charities to reduce systemic racial injustice and police violence).