Okay, so we are finally coming to the end of this series. Today we are going to discuss Flaherty and Sethi’s Homicide in Black and White (pdf):
African-Americans are six times as likely as white Americans to die at the hands of a murderer, and roughly seven times as likely to murder someone. Young black men are fifteen times as likely to be murdered as young white men. This disparity is historic and pervasive, and cannot be accounted for by individual characteristics. … We argue that any satisfactory explanation must take into account the fact that murder can have a preemptive motive: people sometimes kill simply to avoid being killed. As a result, disputes can escalate dramatically in environments (endogenously) perceived to be dangerous, resulting in self-fulfilling expectations of violence for particular dyadic interactions, and significant racial disparities in rates of murder and victimization. … Differences in the manner in which the criminal justice system treats murders with victims from different groups, and differences across groups in involvement in street vice, may be sufficient to explain the size and pattern of the racial disparity.
Well, I must give this one credit for offering up a new hypothesis: black people murder each other so often because they believe that other black people are murderous and are trying to avoid being the victim by killing the other guy first.
This makes sense in a Hobbesian, Napoleon Chagnon studying Yanomamo tribesmen kind of way. (If only there existed some kind of state-run institution that had historically cut discouraged homicide by punishing murderers so that people could go about their business in the heart of American cities without fear of Amazonian jungle-style violence.)
The fact that whites do not go out of their way to preemptively murder blacks to the same degree that blacks murder other blacks suggests that whites don’t think blacks are as violent as other blacks do. This is a curious implication, all things considered, but not unreasonable. Aggressively “paranoid” behavior/belief that one’s neighbors are likely to be violent toward you is probably heritable, itself a result of having ancestors whose paranoia about their neighbors enabled them to survive in a hostile, homicide-ridden environment.
Flaherty and Sethi continue:
The magnitude of the difference in murder and victimization rates far exceeds any difference in characteristics that appear to predispose people to kill and be killed: being poor, being a highschool dropout, living in a dense urban environment, or being raised in a single-parent household, for instance. Blacks are about 2.75 times as likely as whites to be poor, 2.2 times as likely to be high-school dropouts, 2.9 times as likely to live in a large city, and 2.7 times as likely to grow up in a single parent household–all ratios that are far below the observed ratios for murder victimization and offending.
Well, that’s interesting.
Moreover, the racial homicide gap is long-lasting and widespread, and is much greater in cities and among young men than in other places or among other age-gender groups. In rural areas, there is no racial disparity in murder. The homicide gap is also much larger than the racial disparity in aggravated assault–in some ways the crime closest to murder–and there is no racial disparity in aggravated assault among young men.
Source: Color of Crime
Okay, so, our authors are clearly lying: blacks commit aggravated assault at 7x the rate of whites, which is not that different from their rate of 8.5x murder rate. (And, in general, the Color of Crime report demonstrates that incarceration rates reflect actual offending rates.)
Unfortunately, the authors don’t cite any evidence to back up their claim that in rural areas, blacks and whites have equal homicide rates, but as mentioned before, this seems a little problematic, given that the entire continent of Africa has fairly low population density and still has fairly high homicide rates. And why would rural environments make white people more likely to murder each other and black people less likely? Why would cities cool whites’ murderous tempers while inflaming blacks’? Why aren’t any of the world’s most violent cities located in India or China?
Even if we are just looking at selection effects–murderous whites like living in the countryside; murderous blacks like living in cities–we still have to wonder Why?
The Wikipedia page on Statistical Correlations of Criminal Behavior doesn’t mention neighborhood density.
The Atlantic notes:
There’s a big increase in crime as density rises from rural to urban, because crime thrives on anonymity–you don’t rob your neighbors, not necessarily because you like them, but because the likelihood of being identified is very high. In an urban environment, random assaults like Matt’s are much easier to get away with.
While researching this post, I came across some very interesting thoughts on Density and Human Behavior:
Researchers in the social sciences have long tried to explain the effects of urbanization on the human animal. Of special interest has been the observed rates of crime and deviant behavior found in cities. In the United States city crime rates are higher than suburban rates, which in turn are higher than rural rates. …
Two major theories have developed to explain the effects of density on human behavior. Wirth’s (1938) is the most common with his famous statement that size, density and heterogenity explain the effects of urban life on the human animal. The experiments done by Milgram (1970) suggest that when people are confronted with a large number of strangers in everyday life, they tend to withdraw and take less interest in the community in order to protect themselves from overload. Wolfgang (1970), among others, suggests that urban withdrawal and anomie resulting from density explains higher urban crime rates.
Animal studies made famous by Calhoun (1962) show that crowding in the animal world results in what he calls the behavioral sink. Normal behavior and reproductive habits fail. Aggressive behavior increases when density passes a certain point as animals compete for resources. In the experience of the reviewer, those who deny any possible connection between any human behavior simply say that humans are not animals so there can be nothing learned from animal experiments. However, human animals do seem to exhibit much lower fertility rates in cities than is true in rural areas. …
Using data from the Toronto Mental Health and Stress study (Turner and Wheaton 1992), Regoeczi looks at crowding in housing using the measure of persons per room. …
There is an optimal relationship between crowding and withdrawal. The optimal point is 1.18 persons per room. This relationship holds even when the control variables are introduced. “The threshold for aggression is identical to that for withdrawal: 1.18 persons per room. After this point, the deleterious effect of density begins to take off and increased crowding leads to more aggressive responses among individuals.”
In other words, it’s probably safe to conclude that Cabrini Green and Pruitt Igoe, despite the good intentions behind them, were inhuman hellscapes that would drive any sane person to despair.
On the other hand, Manila, Philippines, is (according to Wikipedia) the densest city in the world, at 41,500 people per square kilometer, (or 107,500 people per square mile,) and it isn’t even ranked in the top 50 most murderous cities.
So while density may cause anomie, despair, plunging birth rates, and even anime consumption, it is clearly not the only ingredient involved in making some people murder their neighbors at higher rates than other people.
The best explanation I’ve come across for why our authors might have found closer to equal crime (not murder, haven’t found that) rates in rural areas than in urban areas comes, again, from La Griffe du Lion’s Politics, Imprisonment, and Race:
We all know that African Americans are imprisoned disproportionately to their numbers in the general population. According to the last decennial census a black man was 7.4 times more likely than his white counterpart to be incarcerated. In the language I’ll use today, we would say that the disparity or incarceration ratio was 7.4. State-by-state, the figures varied widely from 3.1 to 29.3. But contrary to expectation, the highest disparity ratios turned up mostly in politically progressive states, while the smallest ratios were mostly found in conservative states. Though the numbers change a bit from year to year, this racial-political pattern of imprisonment endures. One of the questions I will answer today is, why?
La Griffe’s answer is that more conservative (read, rural) states criminalize more behavior and so put more people in prison. Liberal states are more likely to put only the worst criminals behind bars, resulting in even more disproportionate imprisonment of blacks.
But returning to our PDF:
We begin with a baseline model in which race is the only visible characteristic, and the distribution of unobserved characteristics may differ across groups. In this setting, we explore two possible mechanisms through which significant racial disparities in homicide rates can arise. First, suppose that the costs of committing murder are contingent on the identity of the victim, with murders less likely to be solved and less aggressively prosecuted if the victims are black. … But this means
that blacks face greater danger in all their interactions, and are more likely therefore to kill preemptively. Anticipating this, whites will be more likely to kill preemptively in interactions with blacks than in interactions with other whites.
While it appears to be true that people who murder blacks (primarily other blacks) receive lesser sentences than people who commit similar crimes against whites*,** (which makes the disparities between black and white prison populations all the more concerning,) I don’t think this gets us to “blacks are vastly more likely to murder each other than whites are to murder blacks.” If it’s perceived as “open season” on killing blacks, then blacks and whites will kill blacks. And if blacks are killing back in self-defense (or perceived self-defense,) then they’re going to kill other blacks and whites.
*Note that there is an even greater disparity in sentencing between people who kill men and people who kill women, but no one suggests that this disparity is driving male-on-male violence.
**Note also that this does not imply that “society thinks black lives matter less than white lives,” as these lesser sentences maybe a result of black juries being more sympathetic toward black criminals than white juries toward white criminals.
Back to the PDF:
The second mechanism is based on costs of murder being contingent on the identity of the offender rather than the victim. Systematically lower incomes and higher rates of unemployment among blacks make the penalties for attempted murder or manslaughter lower for blacks relative to their outside options.
This is an interesting theory, but it’s kind of destroyed by the fact that high-income blacks still have higher homicide rates than lower-income whites.
So, even before we get to the paper’s math (and Flaherty and Sethi certainly do a lot of math,) I have a lot of disagreements with the paper’s basic assumptions. I don’t dispute the authors’ basic Hobbesian sense that your chances of being punished for a crime or your fear of being murdered by someone else can influence behavior, and I agree that black communities would probably be better served by locking away more criminals so that innocent black people can live in peace instead of fear, but I also think they are ignoring some pretty big behavioral correlations (ie, the behavior of people in Africa; the behavior of people in other dense places,) and have failed to ask themselves why they think people got the notion that crime rates are high in black communities in the first place.
In other words, they’re not entirely wrong, but they’re missing some big pieces of the puzzle.