As soon as news began to spread about the horrific and utterly senseless murder of an Australian-born baseball player in Oklahoma at the hands of three young black males (or at least, that’s how they were first described in the media), I knew what would be coming; specifically, into my e-mail inbox and all around the internet. And even though one of the three implicated in the killing is actually white, and so the supposed racial angle of the crime is incredibly convoluted, if indeed race had anything to do with it at all, I had no illusions that this would somehow change the reaction from certain quarters. Whenever something like this happens, white supremacist groups make sure to blanket the web with vitriol in which they seek to remind white folks how endangered we are at the hands of black people. Knowing that many whites are anxious about the changing demographics of the nation — which portend that around 2043 we will no longer be the absolute majority of persons in the U.S — playing upon fears of criminal victimization by people of color is seen as a great way to win new recruits to the white nationalist cause.
And sure enough, it didn’t take long. As soon as I turned on my computer yesterday morning, there it was: an e-mail from an angry and rather blatantly racist white male, spouting all the same nonsense as I’ve come to expect in moments like this. More savvy than most though, he made sure not to rely on data and arguments from neo-Nazis, but rather from well-known black conservative Walter Williams, who just about a week ago penned a column in which he parroted the line of the white racists, inveighing against the supposed epidemic of black-on-white crime. For Williams, it wasn’t the first time he had done this. Several years back, and taking his cues and information straight from the work of white supremacist and separatist Jared Taylor, he had made the same arguments.
As Williams noted, and as many a white nationalist has over the years, 83 percent of the time when violent crime is interracial, the perp is black and the victim is white. So although most criminal violence may be intra-racial (that is to say, black-on-black and white-on-white), when the offender and victim differ racially, it is typically the case that someone black is doing the crime and someone white is suffering the consequences. As the sender of yesterday morning’s e-mail put it, what do I have to say about that?
The answer, of course, is quite a bit, actually. Williams’s claim — which he makes upon the basis of Justice Department data from the National Crime Victimization Survey — suggests an incredible imbalance, almost as if blacks were deliberately targeting white people in a veritable orgy of hate! And yet, upon closer inspection the number isn’t nearly as meaningful as Williams, or the white supremacists who first began circulating it many years ago, take it to be.
Let’s look at the data, so as to better understand the logical fallacy in Williams’s position.
So Just How Dangerous Are Black Folks? The Trouble With Right-Wing Data Analysis
First, according to the 2008 data Williams relied on (see Table 42), which is the most recent comprehensive data published by the Justice Department on violent crime victimizations, that year there were approximately 3.6 million violent crimes involving a single-offender. These crimes include assault, both simple and aggravated (mostly simple), robbery, and sexual assault or rape. Of those 3.6 million violent crimes, whites committed 2.1 million of them (58.4%) while blacks committed about 830,000 (22.8%) About 442,000 (or 12%) involved perpetrators whose race was not known to the victim. Thus, for crimes where the race of the perpetrator was known to the victims (about 3.2 million crimes), whites would have committed about 66 percent (two-thirds) and blacks about 26 percent (one-fourth). If we assume the same rough racial distribution for crimes where the perpetrator’s race was not known as for crimes where the offender race was known (a reasonable guess), this would mean that whites committed an additional 300,000 crimes, roughly, while blacks would have committed an additional 115,000. In all, this would mean that in 2008, whites committed roughly 2.4 million single offender violent crimes, while blacks would have committed around 950,000.
Since there were roughly 31 million African Americans, age 12 and over — and thus eligible for consideration in crime data — in 2008 (this is noted in the same DOJ tables Williams and I are both referencing here), at most, this would mean that for every 1000 blacks in the population there were 30 criminal offenders, and thus, at most, 3 percent of blacks committed a violent crime in 2008 (meaning, importantly, that at least 97 percent did not). And since there were approximately 205 million whites, 12 and over, in the population that year, this would mean that for every 1000 whites in the population there were 12 criminal offenders, and thus, at most, about 1.2 percent of whites committed a violent crime in 2008.*
Of course, in truth, and for both groups, the numbers are quite a bit smaller than this. After all, the 30 offenders per 1000 persons (the 3 percent rate of offending for blacks) are not 30 different people. In other words, to say that 3 percent of blacks commit a violent crime each year would assume that each offender committed only one crime, such that the number of offenses equalled the number of offenders, but that isn’t the case. There are, as we all know, many offenders who commit multiple offenses each year. As such, the number of offenders would be quite a bit smaller than the number of offenses.Criminologists have estimated, for instance, that 70 percent of criminal offenses are committed by just 7 percent of the total offenders, meaning that there is a small hardcore group of seriously predatory criminals out there doing most of the crime. This would mean that 93 percent of all offenders commit just 30 percent of the crimes. So this would mean that of the 950,000 violent crimes committed by blacks in 2008, 70 percent of them (or 665,000) would have been committed by just 7 percent of all black offenders, while 285,000, roughly, would have been committed by the other 93 percent of offenders. If we assume that the 93 percent who weren’t the major repeat offenders only committed one crime each (likely a conservative estimate, but one which errs on the side of the right-wing argument by maximizing the potential numbers of black offenders), this would mean that, at most, the 285,000 offenses actually equate to 93 percent of the offenders. If 285,000 represents 93 percent of all black offenders, then the remaining 7 percent of offenders above that number would come to only an additional 20,000 or so offenders — major hardcore criminals who commit about 665,000 crimes each year. This would mean that at most there might be a little more than 300,000 individual black violent offenders each year. As a percentage of the 12 and over black population in 2008, this would represent only about 1 percent of all blacks who will commit a violent crime in a given year, versus 99 percent who will not. Even if we just restricted the analysis to black males — and even if all these crimes were committed by males, which they were not — it would mean that no more than 2 percent of black males would commit a violent crime in a given year. This alone is an important point to keep in mind, as it suggests that having a generalized fear of black folks, or black men, is the height of statistical irrationality.
But…But…What About the Black-on-White Crime Data? How the Right Distorts Interracial Crime Figures
As for interracial crime data, this is where right-wing statistical stupidity just becomes downright manipulative. On the one hand, yes, according to the tables, there were indeed far more black-on-white violent crimes (B-W) than white-on-black violent crimes (W-B) in 2008: about 429,000 in the first case and only about 91,000 in the other. Meaning that out of about 520,000 single-offender interracial violent crimes that year, 82.5 percent were black-on-white (B-W) while only 17.5 percent were white-on-black (W-B). My goodness! Perhaps Williams has a point.
Well, no, not really. And for someone who is supposedly a reputable scholar, his illogic is particularly glaring. Before explaining why, let us just note at the outset that there is no reason to believe most of these attacks (either the B-W or the W-B) were racially motivated. This data does not refer to hate-related violence, or violence where there has been a discernible racial motivation (as Pat Buchananrecently claimed on Fox), and indeed, the biggest racial disproportion for violent crime, numerically — and year after year — is in the area of robbery, the motivation for which is pecuniary (financial), rather than bigotry.
But more than that, even the basic claim that B-W crime is wildly disproportionate and indicative of some kind of race-based targeting by black offenders against white victims is utter nonsense. Here’s why.
First, relative rates of interracial offending are directly influenced by an important factor, which goes ignored by Williams: namely, the relative rates of criminal offending in general for blacks as opposed to whites. In other words, if the black criminal offending rate in general is 2.5 times higher than the white offending rate, which it was in 2008 (30 per 1000 blacks/12 per 1000 whites), we should logically expect far more black-on-white crimes than white-on-black crimes, simply as a matter of random chance, and having nothing to do with racial targeting.
Imagine, for instance, a hypothetical community where there are 6.7 times more whites than blacks (as was true in the U.S. in 2008) and where black offending rates in general are 2.5 times higher than for whites. And let’s say that in our hypothetical community, there are 20,000 blacks. Multiplied by 6.7 this would give us 134,000 whites for a total population of 154,000. If, as the real world data suggests, the black violent offending rate is 30 violent crimes per 1000 black persons (or 3 percent), we would expect to see about 600 violent crimes committed by blacks in this community. And if, as the real world data suggests, the white violent offending rate is about 12 violent crimes per 1000 white persons (or 1.2 percent), we would expect about 1608 violent crimes committed by whites in this community. Given relative population sizes (and thus, availability to be victims of one another), we could expect 87 percent of the victims of black violent crime in this restricted white-black sample to be white, since 87 percent of this restricted range, black-white universe, is white; likewise, we’d expect about 13 percent of white crime victims to be black, since 13 percent of this restricted range, black-white universe, is black. This would mean that we could predict about 522 black-on-white violent crimes in a given year, as opposed to only 209 white-on-black violent crimes. In other words, just based on relative population sizes (and thus, victim availability), and general offending rate differences (and thus, perpetrator availability), we should expect 2.5 times more B-W violent crimes than the reverse. In this scenario, about 71 percent of all interracial crimes would be B-W as opposed to W-B.
Aha! Walter Williams might say: See, I was right! Random chance would only predict that 71 percent of interracial violence was black-on-white, but in truth, 83 percent of such violence was. And random chance would predict 2.5 times more B-W crime than W-B crime, but in actuality the ratio is 4.7 to 1. So there must be racial targeting!
The Importance of Differential Encounter Rates
Well, again, no. And again, for reasons that any reasonably competent social scientist should be able to discern; foremost among them, the fact that whites and blacks don’t actually encounter each other randomly throughout the population, as if we were evenly distributed throughout the communities where we live. We still tend to live in relative isolation from one another, such that whites do not actually encounter blacks at a rate commensurate with their overall share of the population, and blacks do not encounter whites at a rate commensurate with our share, either. Because of ongoing residential and spatial isolation, the rates of interracial encounter are far less, for both groups, than this. According to the best estimates we have (which are admittedly a bit dated but have not likely changed dramatically, given ongoing patterns of white/black housing isolation from one another), only about three percent of persons encountered by whites are black, on average, while about 57 percent of persons encountered by blacks are white, on average — in both cases far less than population percentages alone would predict. So now we would need to adjust for encounter rates, in order to determine the amount of interracial violent crime that should be expected given random chance and even without any racial targeting.
In our hypothetical above, the 600 black offenders would have a 57 percent chance of victimizing a white person, given encounter rates, while our 1608 white offenders would have only a 3 percent chance of victimizing a black person, also because of encounter rates. This would mean that in our hypothetical community we should expect 342 black-on-white violent crimes, compared to only 48 white-on-black violent crimes, for a ratio of 7.1 to 1 — a ratio that is 1.5 times higher than the actual real world ratio of 4.7 to 1. And in this hypothetical community, 88 percent of all interracial crimes would be black-on-white, which is actually higher than the 83 percent of such crimes in the real world.
In other words, given general offending rates which are indeed higher for blacks than whites (but which still indicate that the overwhelming majority of blacks are not violent criminal offenders), and given the likelihood of interracial encounters between whites and blacks (and which are especially rare for whites encountering blacks), the rates of black-on-white interracial crime are either pretty much exactly what would be expected without any racial targeting whatsoever, or they are even lower than random chance would predict.
Even if we assumed that a reduction in hyper-segregation had increased the likelihood of interracial encounters since the above-referenced interracial encounter rates were estimated, this would not change the fundamental argument here, or the conclusions we should draw concerning the supposed disproportionality of B-W crime. So, for instance, even if we assumed that whites were now 2/3 more likely to encounter blacks than before, due perhaps to the growth of the black middle class, and thus, more blacks moving into previously white spaces (which is likely a very high estimate), this would mean that still, only about 5 percent of those encountered by whites would be black. And, since blacks moving to white areas as they move into the middle class would also result in a dramatic increase in the percentage of whites encountered by those blacks, this would mean that not only the 3 percent W-B encounter rate but also the 57 percent B-W encounter rate would grow. Let’s say that it only grew by 10 percent, such that blacks today were only 10 percent more likely than before to encounter whites, which would likely be a conservative estimate. This would mean that rather than 57 percent, now about 63 percent of persons encountered by blacks would be white. Although the encounter rate differences would have shrunk to be sure — from a 19:1 ratio down to a 12.6 to 1 ratio, the resulting statistical logic would remain operative. Under this scenario, in our hypothetical community, we would expect the 600 black offenders to have a 63 percent chance of victimizing a white person, while our white offenders would have a 5 percent chance of victimizing a black person. Thus, we could expect 378 B-W violent crimes, compared to 80 W-B crimes, for a ratio of 4.7 to 1, which is exactly the ratio of B-W to W-B crime in the real world for 2008. And in this hypothetical, 82.5 percent of all interracial crime would be B-W crime, which, once again, exactly mirrors the real-world data referenced by Williams. In other words, and without any racial targeting whatsoever, the interracial crime data about which Williams and various others on the right make such a big deal, would look exactly as it actually looks. Simply put, when it comes to the idea that blacks are targeting whites as some kind of race-based pattern, there is no there, there.
Some additional things are also worth noting. First, in 2008, according to the very same table Williams uses for his claim, black and white victims of violent crime were equally likely to have been interracially victimized by members of the other race group. In other words, from the victim perspective, the odds of interracial victimization are the same. From the offender perspective, the odds of black-on-white victimization are greater, but this is entirely a function of population demographics, encounter rates and general offending rates. Once these are controlled for, there is simply no truth to the suggestion that black criminals are targeting whites as victims.
And there is one more thing. When we compare interracial offending rates to Census data in 2008 for persons 12 and over (which is also provided in the DOJ report), we discover that at most, 1.4 percent of blacks, 12 and over, criminally victimized a white person that year, and this is assuming that each of the 429,000 B-W violent crimes were committed by unique perpetrators, which is unlikely. If we assume the 70/7 rule as noted above — which left us with about 300,000 individual black offenders rather than 950,000 in 2008 (or only about 1 percent of the 12 and over black population, as opposed to 3 percent before the 70/7 adjustment) — and even if we assumed that three-fourths of black offenders would victimize a white person in a given year (a ridiculously high estimate), this would mean perhaps 225,000 blacks who in a given year would victimize a white person. As such, this would mean that the percentage of the black population victimizing whites in a given year would be no more than 0.7 percent of the black population 12 and over.
But even this estimate is high, since 92 percent of all Hispanics are included in the white category (of both crime victims and perps) in DOJ data, and so Hispanics victimized by blacks will appear as “white victims” (see note below). Since Latinos are more likely to live near blacks than non-Hispanic whites are, given economic stratification and levels of urbanicity for Hispanics and blacks as opposed to “real whites,” we can safely assume that a significant number of white victims in the data are actually persons of color (Latino and Latina). But even if we assumed that all of the B-W violent crimes were committed against non-Hispanic whites (who numbered about 173 million people age 12 and over that year) this would mean that at most, perhaps 0.25 percent (2.5 tenths of one percent) of all whites were violently victimized by blacks that year.
Interracial Homicide Data: or How to Totally Destroy a Right-Wing Meme
As for homicide — obviously the most serious of all violent crimes — the supposed interracial imbalance, and the supposed B-W disproportionality is even less impressive. According to FBI data (which is where the racial perp/victim data is tallied for homicide, as opposed to the NCVS tables considered by Williams above), for those crimes where the race of the perp and victim are known, in 2010, there were 447 B-W murders and 218 W-B murders: in neither case a particularly substantial number. Indeed, whites were 6.2 times more likely to be murdered by another white person than by a black person and blacks were about 11 times more likely to be murdered by another black person than by a white person. Although both of these numbers (the 447 and 218) are no doubt lower than the true numbers that year for both directions of interracial homicide (since they only represent murders where the race of the perp is known, and for many homicides that information is not clear), relative ratios would not change much even if that information were available for all homicides. As such we can use these numbers to determine whether the rates of B-W homicide are truly so high, relative to W-B homicide as to suggest some kind of racial targeting of whites by blacks. The key in making this determination is similar to the one employed above for all violent crime: namely, the relative homicide offending rates, generally, for blacks versus whites, since this will directly effect the predicted rates of interracial homicides in each direction.
In 2010, of homicides where the race of the offender was known (10,870) blacks committed 53 percent of all homicides (or a total of 5770), while whites committed 48 percent (or 4849). If we extrapolate those percentages to the crimes where the race of the offender was not known, this would mean that of the 4224 homicides with a “race unknown” offender, 2239 would have been committed by blacks, while 2028 would have been committed by whites. In all, this means that in 2010, blacks committed 8009 homicides, while whites committed 6877 homicides. As a percentage of the black population, in 2010, 8009 homicides represent 0.02 percent (2/100ths of 1 percent) of all blacks that year (roughly 39 million) who will kill someone. As a percentage of the white population in 2010 (223.5 million), 6877 homicides would represent 0.003 (3/1000ths of 1 percent) of all whites that year who killed someone. This means that blacks commit homicide, generally, at a rate that is 6.7 times greater than the rate for whites. Now, what does this mean for estimated rates and numbers of B-W homicide as opposed to its W-B counterpart? Let’s see.
Even if we assumed a random and perfectly mixed white and black population — such that whites and blacks encountered each other at rates relative to their population percentages — the much higher black homicide offending rates alone would predict that there should be 6.7 times more B-W murders than W-B murders. But in fact, as we saw, there were only about twice as many B-W murders as W-B murders. And when we consider the above-mentioned data on relative rates of interracial encounter, the numbers are even more striking. Even if we assume that 5 percent of all persons encountered by whites are black (an increase of 2/3 from prior and clearly documented data), and that only 63 percent of persons encountered by blacks are white (an increase of only 10 percent in the same period), we would expect 12 times more B-W homicides than W-B homicides in a given year. In a community of 3 million people, for instance, in which whites outnumbered blacks 6.7 to 1 (as is the case in the real world) there would be roughly 2,610,000 whites and 390,000 blacks. If 0.02 percent of blacks committed a murder, this would mean that we could expect 78 black homicides that year, and 63 percent of these (or 49 homicides in all) would involve white victims. If 0.003 percent of whites committed a murder that year, this would predict a similar number, roughly 78 murders committed by whites, of which only 4 would involve black victims. In other words, given relative rates of homicide offending along with relative rates of interracial encounter, we could expect 12.25 times more B-W homicides than W-B homicides in any given year. But in fact, in 2010, B-W homicide was only twice as numerically prevalent as the opposite. In other words, B-W homicide is roughly 1/6 as common as random chance would predict.
And given the relative population percentages of whites and blacks, blacks are actually more likely to be interracially murdered by a white person than vice-versa. After all, as for homicides where the race of the offender is known, 447 B-W murders as a share of the white community is 2/10,000ths of 1 percent (0.0002) of all whites killed by blacks, which is 1 in every 500,000 white people who will be killed by a black person in a given year; meanwhile, 218 W-B homicides as a share of the black community is 5.5/10,000ths of 1 percent (0.00055). So although interracial homicide is incredibly rare in either direction, any given black person is more than 2.75 times as likely as any given white person to be interracially murdered, with roughly 1 in every 180,000 black persons being killed by a white person in a given year.
How anyone could fully examine this data carefully, either for violent crime generally or for homicide in particular, and conclude that there was a black-on-white crime spree underway is beyond the scope of the rational mind to comprehend. But apparently such claims are the stuff of professional scholarship to the likes of Walter Williams, which says a lot about the pathetically low quality of scholarship demanded of right-wing economists, or your garden-variety white nationalist on the internet. That such claims are taken seriously attests to the propaganda value of racist argumentation, and suggests how much work we still have to do to derail this counterfactual narrative before it does even more damage to race relations in America.
*It is true, of course, that some of the “white” offenders — and victims, it should be noted — in DOJ data are actually persons of color, namely Hispanics. Although Hispanic victims are treated separate from white victims in FBI Hate Crime data, they are not broken out of the DOJ data discussed here. Because Hispanics are an ethnic group rather than race, they are assigned to the various primary racial groups in almost all government data, and not broken out separately in any given data table, unless noted as such (as with the hate crime data collected by the FBI). In the case of crime data from the DOJ, about 92 percent of Hispanics are classified, racially, as white. You can verify that Hispanics are rolled in with whites and blacks (mostly whites) as victims of violent crime, by adding the numbers of victimizations of those labeled white (in Table 5 of the 2008 data), those labeled black (also Table 5), those labeled “other,” which means Asians and Native Americans, principally (also Table 5), those who are listed as being of “mixed race” or more than one race (Table 5) and then those who are Hispanic (in Table 7). The sum of these figures will amount to more (5.42 million) than the total number of violent crime victims listed in table 5, (4.856 million) indicating that Hispanics are, in effect, already being counted in the existing racial totals. As indicated in other government data collections, about 92 percent of Hispanics are classified as white, which is likely to obtain in regard to crime data as well. This means that of the Hispanic victims tallied in Table 7, about 92 percent of them would also be found in the white column of victims in Table 5, thereby artificially inflating the number of “real white” victims.
As for offenders, although white supremacists have long argued that Hispanics commit violent crime at a disproportionate rate — somewhere between whites and blacks — the most comprehensive analyses of Hispanic crime suggest this claim is utterly false. Recent analyses done by a prominent far-right conservative and even, in one case, a source long revered by white nationalists, suggest that Hispanic violent crime rates are essentially no different from those of non-Hispanic whites. As such, the 1.2 percent rate, per capita, of criminal offending for whites (including Hispanics), would apply to both Hispanic whites and non-Hispanic whites. This means that with Hispanics representing 34.5 million persons age 12 and over in 2008, and 92 percent of these being classified as white, racially, there would have been 31.7 million Hispanic whites that year. If they offend at a rate comparable to non-Hispanic whites (12 per 1000 persons, or 1.2 percent), this would translate to 380,954 violent crimes by Hispanic whites in 2008, most of which would come from the existing white totals, but many of which would also have been located in the “don’t know/not available” category, as with certain white and black crimes (noted above).
In all, because Hispanic victimization rates (Table 7) are similar to the overall white victimization rates (seen in Table 5), and because the recent research suggests Hispanic offending rates are also comparable to non-Hispanic white rates (when it comes to violent crime), doing a full methodological extraction of Hispanics from the white totals (both for victims and perps) would result in roughly proportional reductions in the white column for both. The effect on interracial crime figures would be that, a) the numbers of W-B violent crimes would drop (since some of those whites would have been Latino), and b) the number of B-W crimes would also drop (since some of those whites would also have been Latino). If anything, we would expect the effect on white “victims” of interracial crime to be larger than the effect on white perps. Since the black offending rate is disproportionate (roughly 2.5 times the white offending rate), it would stand to reason that there would be more blacks offending against Latinos, proportionately, than the opposite, and for the same reasons as discussed above, in terms of the effects of encounter rates and general offending rates. So while there would be fewer “real” white-on-black crimes too, the reduction in the number of “real” white victims would likely be greater, thereby further reducing the actual interracial crime gap between B-W and W-B crime. In other words, if the data were truly available (and they are not) to estimate the Hispanic/non-Hispanic ethnicity of whites who victimize or are victimized by blacks, and we were to perform the extractions necessary to leave only “real” whites in both categories, the effect would be to make Williams’s argument (and that of the white nationalists from whom he initially got this line of attack) even weaker.