Lies, Damned Lies, Statistics, and Race
Photo by ev on Unsplash
In the wake of the murder of George Floyd while in the custody of the Minneapolis police, many African-Americans and other people of color have spoken and written about their experiences with law enforcement, including racial profiling, abuse, and the fear that any interaction might result in their death. Such testimony of improper police behavior toward blacks is by no means a recent development, from the era of slavery, to Jim Crow laws, to cases of police brutality that have been in the media spotlight since at least the 1960s.
Some officials and commentators, mainly on the right, however, have argued that there is, in fact, no proof of disproportionate treatment across racial lines by the police. On Sunday, Attorney General William Barr and acting Homeland Security Secretary Chad Wolf both said they do not believe there is systemic racism in policing in the US. As another example, on June 3, the National Review published an article by Andrew C. McCarthy entitled “The ‘Institutional Racism’ Canard.” In the article, McCarthy argues from statistics that there is just no support for the existence of institutional racism by the police.
While the personal statements of so many African-Americans provides compelling testimony, they are anecdotal evidence. What, then, does the actual data show?
Unfortunately, much of the information we have is incomplete, and it can be very easy to use aggregated data to prove any point one wants, by choosing the inputs or manipulating the analysis. As the saying goes, in a sentence popularized by Mark Twain (who incorrectly attributed it to the British prime minister Benjamin Disraeli), "There are three kinds of lies: lies, damned lies, and statistics."
That being said, we can evaluate the arguments and come to some reasonable conclusions about what we can learn from such data as is available.
Many of those who argue that systemic racial discrimination does not exist in policing note that about 50% of people killed by police officers are white, while only about 25% are black. As McCarthy puts it in his article, “whites are nearly twice as likely as blacks to be shot to death by police.”
This statement is nonsense. The probability of a white person being killed by police would equal the number actually so killed divided by the entire population of whites in the country. Blacks make up about 13% of the US population, so the fact that 25% of police killings are of blacks is substantially out of proportion to their population. Non-Hispanic whites, on the other hand, constitute about 60.7% of the population, while only making up 50% of those killed by the police.
In dismissing these numbers, McCarthy asserts that using total population as the denominator is misleading (or, in his words, “ridiculous”). Instead, he suggests that we should look to the number of crimes committed by people of different races. He notes that, “While African Americans are involved in two times more police shootings than their percentage of the population would seem to warrant, they commit 53 percent of murders and 60 percent of robberies — well over four times their percentage of the population.” As a result, he concludes, the number of blacks killed by police is actually proportionately less than the number of whites.
McCarthy here is cherry-picking statistics, choosing two categories that are the most extreme as an example. If we are to consider nationwide police killings in proportion to the number of crimes committed, we should look at crime in the aggregate, not by individual categories. According to the FBI’s 2018 Crime in the United States report, African-Americans were the subject of 27.4% of all criminal arrests, 37.4% of those arrested for violent crimes, and 30.1% of those arrested for property crimes.
These percentages are relatively close to the percentage of African-Americans who are killed by the police. So, one might argue that this shows there is no racial disparity. However, it is important to dig deeper into the numbers to see whether they would really support such a conclusion.
First, we need to take care to recognize that these are just national averages. A jurisdictional analysis by Mapping Police Violence reveals that the amount of violent crime in a given city does not correlate with the number of police killings in that city. By way of example, they note that Buffalo, New York, has a population of about 260,000, 50% of whom are black, and which has a crime rate of 12 per 1,000 persons; yet between 2013 and 2016, the Buffalo police didn’t kill a single person. On the other hand, Orlando, Florida, with a population of about 255,000, 42% of whom are black, has a lower violent crime rate of 9 per 1,000 persons; but between 2013 and 2016, the Orlando police killed 13 people.
If the number of killings by police in a jurisdiction is not in fact in any way correlated with the aggregate violent crime rate, then comparing the ratio of such numbers for various demographic groups — as McCarthy would have us do — does not provide any meaningful analysis, particularly if performed on the basis of a national average. Simply put, the denominator he proposes (the number of violent crimes committed) is not relevant to measure the extent of the numerator (the number of police killings). Considering this on a jurisdictional level also reveals a consistency in the over-representation of black victims: in every single state which had at least one African-American victim of a police killing, the percentage of black victims exceeded the proportion of blacks in the state’s population.
Further, like all statistics, the accuracy of the result is only as good as the data input. Had the killing of George Floyd not been caught on video, it would have likely been reported as a death while in police custody from natural or other causes, and not a homicide. In one recent study that compared published reports of police killings with an official governmental database, fairly strong evidence was found that the number of law-enforcement-related deaths is indeed under-reported (and in many cases, by a factor of 2), though race was not found to be a major factor. However, this analysis was performed on the basis of publicly available information. A case like Mr. Floyd’s would not have been in the press had there been no video, and so would have gone entirely unreported. Given that often the only information available comes from the police themselves, the data must be considered at substantial risk for unreliability.
At this point, the best we can likely say about statistics on the number and proportion of African-Americans killed by police is that they are inconclusive, though there is good reason to suspect that the data does not reflect the full number of black victims. And on a state-by-state basis as well as in the aggregate nationwide, blacks are consistently killed by police at a ratio greater than their percentage of the population.
In any event, limiting the analysis to just the number of individuals killed by police misses the larger context. While the death of George Floyd has brought the topic of police killings to the fore, the testimony of people of color covers the much broader area of racial profiling and abuse by the police. In this area, the evidence of bias in the statistics is much more definitive.
Many studies have been conducted that provide strong evidence of racial profiling by police – that is, treatment by police of an individual of one race worse than would be the case had the same circumstances arisen in connection with a person of a different race.
The Stanford Open Policing Project, for example, has shown that police officers make traffic stops of black motorists at a higher rate than white motorists, conduct searches of the vehicles of black motorists at a rate higher than for white motorists, and require less suspicion to conduct a search of black motorists than white motorists.
In 2015, California passed the Racial Profiling and Identity Act, which requires law enforcement agencies within the state to report certain data concerning each stop their peace officers made. Among other things, the most recent data shows that blacks are stopped by police more frequently than their percentage of the state’s population, are much more frequently stopped for “reasonable suspicion” than any other race, and during a stop police searched blacks almost three times as frequently as they did whites, even though, when searched, whites had the highest likelihood of being found to have contraband or other evidence.
An article in the Washington Post has aggregated many of the studies analyzing racial profiling and provided links.
Despite all of the analysis to date, there is room for more analysis and better data. It is worth noting that objections from the police themselves are a source of some of the paucity of solid information. Prior to its enactment, the California Racial Profiling and Identify Act, referred to above, was strongly resisted by California law enforcement agencies as an unnecessary burden on law enforcement officers’ time.
And although anecdotes may not constitute proof, they provide insight into the relationship between the police and the black community. Incidents of racial profiling have been reported by Senator Tim Scott, the only African-American Republican senator, and by the astrophysicist Neil deGrasse Tyson, among many others. And in a 2019 Pew Research Center survey, 59% of black men and 31% of black women reported that they had been unfairly stopped by the police.
Some conservative commentators, McCarthy included, fall back on an argument that the number of blacks killed by police is small compared to the number killed in other circumstances, particularly by what they refer to as “black-on-black” crime.
As simple victim-blaming, this argument is morally bankrupt. Innumerable black organizations are working every day to reduce violence in their communities. The suggestion that the “real” problem is being ignored by blacks is both absurd and offensive. In any event, the fact that such other violence exists does not in any way excuse racial profiling or a single death of an African-American at the hands of the police.
The police motto is to “serve and protect.” By failing to treat Americans of all races equally, they are doing neither.