Racism in statistics

Time and time again, the media prey upon the statistical ignorance of the masses to lead them to a factoid conclusion.

The newest case comes from The Guardian:

Black Americans killed by police twice as likely to be unarmed as white people

The Guardian should be commended for their effort to document how many Americans are killed by police each year, since the FBI and police agencies throughout the nation are not doing it properly.

However, they’re using their data to spread misleading and exaggerated assumptions about what that data shows, instead of getting to the bottom of why the numbers are what they are.

The Guardian writes that 32% of black Americans killed so far in 2015 were unarmed, compared to 25% of Hispanic/Latino and 15% of Caucasians. The reader is to infer by these numbers that American police are more racist towards blacks and Latinos because a greater share of them are shot unarmed.

Such assumptions rely on a long list of premises that we know aren’t true, statistical variables that need to be controlled before presenting numbers in context with reality.

For instance, we have to take into account the following facts about black and Latino Americans compared to white Americans, on average, due to a variety of reasons (including historical and structural racism): they’re poorer, involved in (violent) crime more often, and live in more dangerous areas.

In more dangerous areas, you’re more likely to find poorly trained police, due to lack of resources, lower wages and them being a less attractive place to work.

A poorly trained police officer is more likely to use deadly force.

Not only that, but The Guardian provides absolutely no context as to what happened in all these cases. We aren’t told how each of the unarmed victims were reacting to the police at the time they were shot, what situation the police officer was dealing with, if the officer had reason to believe the unarmed person was a danger to them, themselves or others — all of which is part of a police officer’s gauge to use force.

Michael Brown is a perfect example.

The Department of Justice report exonerated police officer Darren Wilson for any racist intent or wrongdoing when he shot Brown, for several reasons.

Brown and his friend had just robbed a convenience store, meaning he had a motive to act suspiciously when confronted by a police officer. When that confrontation started, over a simple traffic safety issue, Brown acted violently towards Wilson — in an attempt to resist an impending arrest.

Wilson feared for his life because the larger and stronger Brown tried to steal his gun, caused it to fire inside the police car, and then turned around and ran at Wilson when he told Brown to stand down.

The DOJ report is quite clear on how meaningless the label “unarmed” can be in such situations:

When the shootings are viewed, as they must be, in light of all the surrounding circumstances and what Wilson knew at the time, as established by the credible physical evidence and eyewitness testimony, it was not unreasonable for Wilson to fire on Brown until he stopped moving forward and was clearly subdued. Although, with hindsight, we know that Brown was not armed with a gun or other weapon, this fact does not render Wilson’s use of deadly force objectively unreasonable. Again, the key question is whether Brown could reasonably have been perceived to pose a deadly threat to Wilson at the time he shot him regardless of whether Brown was armed. Sufficient credible evidence supports Wilson’s claim that he reasonably perceived Brown to be posing a deadly threat. First, Wilson did not know that Brown was not armed at the time he shot him, and had reason to suspect that he might be when Brown reached into the waistband of his pants as he advanced toward Wilson. See Loch v. City of Litchfield, 689 F.3d 961, 966 (8th Cir. 2012) (holding that “[e]ven if a suspect is ultimately ‘found to be unarmed, a police officer can still employ deadly force if objectively reasonable.’”) (quoting Billingsley v. City of Omaha, 277 F.3d 990, 995 (8th Cir. 2002)); Reese v. Anderson, 926 F.2d 494, 501 (5th Cir. 1991) (“Also irrelevant is the fact that [the suspect] was actually unarmed. [The officer] did not and could not have known this.”); Smith v. Freland, 954 F.2d 343, 347 (noting that “unarmed” does not mean “harmless) (6th Cir. 1992). While Brown did not use a gun on Wilson at the SUV, his aggressive actions would have given Wilson reason to at least question whether he might be armed, as would his subsequent forward advance and reach toward his waistband. This is especially so in light of the rapidly-evolving nature of the incident. Wilson did not have time to determine whether Brown had a gun and was not required to risk being shot himself in order to make a more definitive assessment.

When blacks and Latinos on average are more involved in crime than whites, that will inevitably lead to more situations with the police where they have a motive to escape or harm the police — even when unarmed.

Without assessing how aggressive or threatening each of these unarmed people were being at the time they were shot, presenting raw percentages by race is meaningless as to ascertaining the reasons for why it is happening.

The differing attitudes police among these groups is also not factored into these raw numbers. In some ways, it’s reasonable for blacks and Latinos to be more wary or distrusting of the police, which leads to them acting differently when confronted by police — especially white police officers.

Blacks and Latinos are also, on average, less educated than whites (due to poverty, etc.) — does this play a role in how they think through their options when confronted by a police officer? Is a black or Latino criminal more likely to try and run away or attack a police officer, even while unarmed, than white criminals?

I don’t know the answers to these questions, but we can’t just assume that these groups act exactly the same in these situations, especially given all the socioeconomic differences between the groups.

A similar report by the Washington Post points out that, among those shot and killed by police, most of them have something in common:

Although race was a dividing line, those who died by police gunfire often had much in common. Most were poor and had a history of run-ins with law enforcement over mostly small-time crimes, sometimes because they were emotionally troubled.

Along these lines, The Guardian does the same exact thing when they present percentages of each race killed by police compared to their portion of the population.

37% of the US population are non-white minorities, yet they make up 47% of those killed by the police, according to the newspaper’s numbers.

No further context is given to these numbers.

Nearly 50% of all murders in the United States are committed by blacks, though they only constitute 13% of the population. Blacks are arrested at a rate of 2.6 times their share of the population. For murder and robbery, blacks are arrested at a rate of 6 and 8 times their share of the population respectively.

According to the newspaper’s numbers, blacks are 29% of those killed by police, or a little more than twice their share of the population.

In other words, the black murder rate is almost four times their share of the population, yet they’re killed by police less than three times their representation. Compared to their arrest rates for murder and robbery, it’s even more lopsided.

Why would anyone expect blacks to be murdered much less often by police than they murder others?

Here’s a statistic you’re unlikely to ever see written by The Guardian: black Americans kill other Americans almost twice as often as police officers kill them, according to their share of the population.

These journalists are not dumb, they know about these facts — they just don’t want to present them because they ruin the inference they want you to make.

It’s imperative that American authorities find ways to decrease the amount of deaths caused by the police. Newspapers such as The Guardian are doing this issue a great disservice by cherry-picking numbers and misleading the public into believing false or exaggerated causes of it.

In order to fix a problem, one must understand why the problem is happening, not set off on a wild goose chase based on a factually deficient reading of statistics.