A Note to My Calculation in “Are African–Americans Disproportionately Victimized by Police?”

• Only one truly plausible critique has so far been presented to me so far of the the argument in my essay, “Are African–Americans Disproportionately Victimized by Police?”, where I explain that:

(1) The violent crime rate presents us with an effective way of estimating how frequently police are having justified encounters with individuals from varying racial demographic groups. If we want to ask whether police treat individuals who belong to some racial demographic groups whom they encounter differently from others, it simply doesn’t matter in the slightest what racial percentage of the general population those individuals represent—for exactly the same reason that it does not matter in the slightest for the purpose of this analysis that 22% of the world population is black whereas only 5% of the world population is white. What matters is how many members of these racial demographic groups police are encountering on a regular basis. And the answer to the question, “Who is committing the violent crime?” tells us not only how often police are likely to be encountering members of different racial demographic groups, but how often they are justified to by the fact that addressing violent crime is the primary job and purpose of police. 

(2) We are, thankfully, not reliant on police arrest data alone to determine how frequently members of various racial demographic groups are committing violent crimes. In addition to that data, we have a federal collection of eye–witness data from both victims and third–party witnesses stretching back for decades in the form of the National Crime Victimization Survey (NCVS) collected and available online at the National Archive of Criminal Justice Data (NACJD)—and this data reveals that victims and witnesses have recorded a higher percentage of black perpetrators committing violent crimes than are found in police arrest data consistently for years. It therefore turns out that the disparity in arrest rates understates how disproportionately black perpetrators are responsible for violent crimes. Even though African–Americans are just 13% of the population, approximately a full 50% of the United States’ murders are committed by perpetrators who are black. And since most of these are committed by younger black males, who make up less than 6.5% of the population, the disparity is even worse than it seems. Whether this is, as liberals would have us believe, solely because of the influence of poverty or not, the reason why this disparity exists is simply irrelevant to the analysis itself. At least part of the explanation, in any case, is neither particularly “conservative” nor “liberal” in its implications—most violent crimes are committed by males between the ages of 18–25, and the African–American population in the United States skews much younger than others do, which means males in the crime–prone age demographic make up a larger percentage of the African–American population than they do of other racial demographic groups.

(3) When we compare the per capita rate of violent offenses committed against the per capita rate at which police  shoot and kill suspects of differing demographic groups (whether these shootings are justified or unjustified), not only does the disparity between African–Americans’ representation of the population and their representation in such shooting deaths disappear, but in fact it turns out that the trend reverses to the point that per justified encounter with police, African–Americans are in fact less likely to be shot by a police either during an attempt to commit a crime or while being legitimately suspected of one (because a suspect is known to be black). In other words, African–Americans commit a substantially larger number of violent crimes, and have a substantially higher number of valid encounters with police, before any one African–American ever ends up shot by police—while more white and even Hispanic suspects will be shot before they have committed nearly as many crimes or had nearly as many justified encounters with policemen approaching them during, or questioning them about violent crimes.

(4) Not only does that hold true in the extant national numbers, but it turns out that that finding actually also confirms what we have found in the most thorough experimental study data collected so far: police hesitate more before shooting black suspects in experimental trials which actually control rigorously for individual behavior by ensuring that the only difference between the virtual “suspects” these police have to make a decision to either fire or not fire at is race (unlike the real world, where this behavior absolutely inevitably differs on average).

Now, there is in fact some validity to the critique that I chose a questionable data–set for my argument in point (3), when I selected data regarding how many members of different racial groups are shot and killed by police, even if the data I selected for point (2) regarding how often members of different racial groups commit violent crime was correct and accurate. It is trueI chose numbers at this stage of the argument that vastly underestimate the number of police shootings that take place each year. However, this still does not ultimately weaken my final argument—it strengthens it. I should have clarified up front in the original essay what my reasons were for choosing to use the CDC’s data on what it calls “death by legal intervention for this stage of the argument even though it clearly underestimates the total number of shootings each year.

The reason I chose this particular data–set is: (a) because I knew that my argument was strong enough that any existing data–set that can be plugged into it will produce the same bottom–line conclusion; quite simply, by no estimate anyone has made is the statistically disproportionate rate at which African–Americans are shot by police anywhere close to the disproportionate rate at which African–Americans are responsible for violent crime; and (b) because what matters for this stage of the argument is not the raw numbers, but rather the relative percentages of black, white, and other individuals found in its data—and the CDC data actually identifies a relatively large disproportion towards black “deaths by legal intervention” even compared to more accurate data.

In short, whereas the study of CDC data I used for my original argument found that blacks were 34% of deaths by “legal intervention”, the more accurate and recent data only places this relative number at 21%. (Update: I’m confused by what is going on in this data now. They claim that 21% of their 1600 victims are black whereas 32% were white, but this would mean 336 blacks versus 512 whites (to which we’d also have to add Hispanics), and that isn’t what I count when I count the entries they include under race myself; I count 390 blacks in 2013 alone, and 597 whites in that same year. To be as cautious as absolutely possible, I have removed the previous numbers—which found that whites+Hispanics were 175% as likely to be shot in a given encounter with police as blacks—and replaced them with my own count, even though I’m not sure why the two differ, and it could be that their summary is accurate and that my count is not. In any case, also bear in mind that this new estimate—which still does support my argument, anyway—is biased downwards by the inclusion of Hispanics, who have a higher per capita crime rate than whites.)

(Old paragraph: Substituting more accurate data, then, will actually lessen the relative percentage of black compared to white “deaths by legal intervention”, even if it increases both absolute numbers. And it thereby renders the point actually made in my conclusion even more solid—the data I originally chose artificially strengthened the liberal narrative that I was arguing against; not my own conclusions. For the original essay, I checked this against more thorough data before ever publishing what I had written; but during my early investigations into the question, I had typed out what happens to the CDC data in that calculation simply for my own benefit—and once I saw that the same conclusion is reached no matter what data we use, I went ahead and published what I had already had conveniently typed out anyway. It was a mistake to do this without explaining what happened thoroughly.)

To rectify this mistake now, I’m going to take the highest unofficial estimates of how many police shootings, whether justified or unjustified, take place each year that exists anywhere and re–run the original analysis I made in the third step of that argument to show that the same conclusion is still so thoroughly established that it would take an epidemic of black individuals shot by police invisible to all anecdotal, media, or official reports that exist anywhere in order to reverse the trend I’ve identified. In other words, an actual conspiracy theory.

The highest estimate for those numbers that exists anywhere, the data informally collected at the Killed By Police database, brings it to about 1,100 per year, or 2.6 police shootings each day. To get the years’ end tally of how many of these individuals belonged to which racial category, I pulled compiled data from an analysis here.

Note that everything that follows will use 2013–2014 numbers, and recall that in order to fit this data into my analysis, we’re going to have to lump Hispanics killed by police in with “whites” killed by police. Why? Because until some time in the middle of 2015, the federal data on crime offending classified Hispanic perpetrators as white. If we compare the rate of “white” victims of police shootings to the rate of crimes committed by “whites+Hispanics”, then, our numbers will compare apples and oranges; to make the comparison meaningful, our only choice is to compare the rate of “white+Hispanic” victims of police shootings with the rate of “white+Hispanic” violent crimes. Again, as I demonstrate in the original essay, this actually once again turns out to strengthen my analysis in the end, because it inflates the actual white per capita rate of crime. In the meantime, I’ll use the phrase “non–black” to refer to this combined “white+Hispanic” number. So while the report identifies 597 white victims and 390 black victims in its tally, we must add the 251 Hispanic to the white tally to obtain the appropriate numbers for the purposes of this calculation:390 black to 848 “non–black” persons shot by police.

2013 black deaths: 390 (30% of total)
2013 white deaths: 597 (50% of total)
2013 Hispanic deaths: 251 (19.7% of total)
2013 “white” deaths: 848 (67% of total)
2013 total known race: 1271

And with that, we can finally plug these numbers back in to my original calculation:

390 black deaths out of a black population of 37,685,848 equals 0.0000010349% of that total population shot by police, or 1.03 deaths per 100,000 black individuals.

848 non–black deaths out of a non–black population of 271,059,650 equals 0.00000031284% of that total population shot by police, or 0.3128 deaths per 100,000 “non–black” individuals.

Using the official arrest data for 2013 collected by the Bureau of Justice Statistics (which we have established underestimates the rate at which black perpetrators are disproportionately responsible for violent crime, according to victim and witness reports spanning across decades), the 2013 violent crime rate per 100,000 people for black individuals is 465.7, while the 2013 violent crime rate per 100,000 people for “non–black” individuals is 122.7.

Thus, as before, dividing the second number by the first number gives us the rate of police shootings per violent crime committed, and therefore of police shootings per police encounter with a member of that racial demographic group justified by violent crimes committed by that group.

The “black” rate is therefore 465.7 divided by 1.035, or 449.95.

The “non–black” rate is 122.7 divided by .3128, or 392.26.

In other words, black individuals will commit about 450 violent crimes before any one black individual will end up shot by police—whereas “non–black” individuals will commit about 392 violent crimes before any one “non–black” individual will end up shot by police. In other words, police will encounter 15% more violent or legitimately suspected black individuals before shooting one—and conversely encounter 13% fewer violent or legitimately suspected “non–black” individuals before shooting one. The “non–black” suspects end up getting shot faster.

Reducing those numbers down to make them more comprehensible, the proportions are as if police shot one out of every seven black individuals they encounter in the line of duty, but one out of every six “non–black” individuals (6:7 is the smallest possible 15% increase, so this makes the situation of blacks and “non–blacks” with respect to each other easier to visualize even though police shootings obviously do not happen at this frequency.) Once violence is accounted for, the numbers are in blacks’ favor—not against it.

To arrive at these same numbers by a different means, the likelihood that a black individual committing a crime will end up shot by police is 1/450, or 0.0022, while the likelihood that a “non–black” individual committing a crime will end up shot by police is 1/694, or 0.0025. The likelihood that a “non–black” individual legitimately suspected of committing a crime will end up shot by police is therefore an additional 0.0003 larger than the likelihood that a black individual legitimately suspected of committing a crime will—and since that 0.0003 increase is 15% of the original baseline value of 0.0022 for the likelihood that a black individual legitimately suspected of committing a crime will be shot by police, this again means that non–black individuals who encounter police are 15% more likely to be shot during the course of that interaction.

Black individuals end up shot by police out of proportion to their population rate, then, because black individuals commit violent crimes out of all proportion to their population rate, and because police therefore end up in a disproportionate number of encounters with black individuals for perfectly justified reasons; not because police are more likely to shoot any given black individual they come into contact with due to subliminal racism—because in fact, once again, the data once properly controlled for actually demonstrates exactly the opposite.

On a related note,

Another set of data I’ve presented has been critiqued on very similar grounds—and once again, while there is validity to this critique, correcting it only serves to strengthen my case. In the fourth entry to the “Is Dylann Roof “White Like Me?”” series, I wrote: “It’s worth making a comparison of the relative rates of police brutality and black–on–white violence in the United States to try to put things in perspective. According to the FBI, there were an average of 14,545 murders per year across the years of  2011–2013. which comes out to an average just shy of 40 murders per day. Since African–Americans commit approximately half of those, and pick white victims about 1/5th of the time, that means there are about four black–on–white murders every day in the United States. White perpetrators commit the other half of murders in the United States, but only choose black victims about 2.4% of the time—which means there is slightly less than one white–on–black murder in the United States every two days.

According to data that does not take statistics reported by police departments for granted, but in fact calls them into question, based on data from the early months of 2015, police kill approximately 2.6 subjects per day—approximately half of which are black, which brings the number down to 1.3 police shootings of black suspects per day. Of this number, it is unclear how many are justified or unjustified. According to the FBI, in 2013 police were attacked by someone carrying a weapon roughly 10,000 times—2,200 of those times with a firearm. If police kill 2.6 suspects per day every day for a year, that’s still only about 1000 total killings at the end of the year. Some liberal readers may point to gaps in the data (call it the “racism of the gaps” strategy) and insist on disagreeing, but if police are killing suspects far less frequently than they’re being attacked by them, it seems safe to me to bet that the vast majority of those killings are probably justified.

However, even if we assume that every single one of them was unjustifiedcombining the number of police shootings of black suspects per day (1.3) with the number of white murders of black victims per day (0.48) would still give us a smaller number (~1.8) than the number of black murders of white victims every day in the United States (~4). More than twice as many black murderers are choosing white victims as the number of white murderers choosing black victims and the number of police shooting black suspects (justified or not) combined. (Meanwhile, there are 16 black murders of black victims every single day across the United States—more than eight times the number of white civilian murders and justified or unjustified police shootings of black victims combined.)”

(I continued: “However, both of these statistics really still need to be taken account of in terms of the wider context that murder only accounts for 0.6% of the deaths in the United States in general. While there are approximately 40 murders, 4 of which are black–on–white, on a typical day in the United States, on the same day 90 Americans will die in car crashes, 110 will commit suicide, 120 will overdose on drugs, 256 will die in accidental falls or other accidents, 1580 will die of cancer and more than 1600 will die of heart attacks. If Roof is concerned about “saving the white race,” then Burger King, cigarettes, drunk driving, wobbly ladders and clinical depression are far more formidable foes than black criminals. But what goes for Roof’s underlying logic goes for “#blacklivesmatter,” too. Tim Wise is right that it’s only a tiny fraction of the black population who commits an act of violence in any given year—the only problem with that is the hypocritical inconsistency we can well know to expect should anyone say the same about racist attacks against black Americans, whether committed by civilians or police, which even combined are still yet only half the size of the fraction of black citizens committing acts of violence Wise himself has just called “tiny.” Whatever goes for the relative insignificance of disproportionate black–on–white violence goes at least twice as much for both white–on–black and police–on–black violence combined. And it goes even more so for hysteria about mass shootings, which make up only 0.2% of that 0.6% of deaths in America.”)

In fact, these numbers overestimated what percentage of police shootings are of black suspects—once again in order to artificially strengthen the claims I am presenting arguments against. Where I granted the assumption that 50% of the 1,100 yearly (or 2.6 daily) police shootings are of black suspects, in fact, as detailed above, the actual range falls somewhere between 21–34% (and closer to the lower of those numbers in more accurate, and recent, sets of data)—rather than 550 black suspects shot by police each year, we’re actually talking about fewer than 150.

Likewise, when I quoted these paragraphs on Reddit, the top criticism claimed that my implications must be incorrect because I’ve portrayed the situation as if only black and white individuals commit murder. But once again, I arrived at these numbers by starting from the established knowledge that black perpetrators commit half of the national total and then lumping all the rest of the murders in as “white” in order once again to artificially strengthen the case I was critiquing. Not only are there apparently 72% fewer total shootings of black suspects than I assumed in this calculation—instead of 4 police shootings of black suspects every three days, the most damning data suggests an average of only 2 police shootings of black suspects every five days—but every murder committed by an East or South Asian or member of any other non–white, non–black ethnic group subtracts from the white number of murders of either white or black victims which I assume, not from the black number. My simplification of the numbers drastically strengthens the left–wing case which I have attacked; they do not distort the facts in my favor—and even while granting the left–wing case this many assumptions, including that every single one of those 1.3 daily police shootings (which the Killed By Police dataset actually places at 0.4) was unjustified, with not one of them ever justified by self–defense at all—the numbers still result in the conclusion that “More than twice as many black murderers are choosing white victims as the number of white murderers choosing black victims and the number of police shooting black suspects (justified or not) combined.”

In truth, the shorthand inaccuracies in my argument actually understate the strength of its conclusion—the reality of the situation will be even more strikingly skewed in this direction than this. 


Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: