QPP Extra – Radiolab Got It Wrong

Earlier this year, Radiolab produced a two-part examination of police shootings in Florida, Shots Fired. They got it wrong.

I’m a real fan of the WNYC show, Radiolab – have been for years. So I was really excited that Radiolab took on police shootings. I was excited because I really thought they would dig into the numbers.

Use our RSS Feed, or use iTunes, Google Play, PlayerFM, or anywhere you get podcasts.

I’ve spent the past three years studying this; I’ve written a book on it, along with articles in the Post, Times, USA Today, National Review, RealClearPolicy, and other outlets. I’ve spoken on the topic on NPR, and shows like The Takeaway. But Shots Fired, based on some amazing reporting work by Ben Montgomery at the Tampa Bay Times, seems to reduce the shootings to just one common element: race.

I’m not saying police aren’t racist. There’s racism in policing, as in everything else. The problem is that the data on it is very hard to find.

But here’s what I am saying, based on a substantial corpus of significant evidence: race is not predictive of who will be killed by police, racism is not the driver of police use of deadly force, and Montgomery’s conclusions about race and police shootings – like those of reporters from the Washington Post, Guardian, ProPublica, and others – are wrong.

Getting this right is important because, as people consider police reform, using the wrong data will lead to policies that address the wrong things. That will mean we blew our chance. If, for example, we focus our attention on police use of deadly force, when the problems are really in police use of non-deadly force, we will feel good for a while that we made changes, but nothing will be fixed.

Along with a group, I studied police killings. Our methodology was published, and our data is open and is available to review. Our data is very similar to that of Mr. Montgomery’s – I do wish he would release the whole dataset, as opposed to just his course-grained analysis tools.

The case was laid out in their story that a disproportionate number of black people were shot by police – 40% of those shot were black as compared to Florida’s 17% African American population.

Then Jad said it: “If you’re a black person in Florida, you are four times as likely to be shot by police than if you’re white.”

So, is Radiolab really saying that, while 60% of the people shot by Florida police were armed with a gun or knife, and 48 percent were on drugs or alcohol, or mentally ill, and while half those people had pointed a gun at an officer, it is a “police racism” issue that “too many” of the people killed were black?

That, I maintain, doesn’t sound right, because it isn’t.

Behavior is almost always the reason police shoot.

In the podcast, I mention work by Professor Justin Nix and his colleagues that using population as the benchmark “does not account for each groups’ representation in a variety of more relevant measures, including police–civilian interactions and crime.” And Michigan State University researcher Joseph Cesario who said, “if you’re black in Florida you are four times more likely to be shot by the police than if you’re white” – were true, than an officer would be as likely to shoot the cashier selling him a cup of coffee as he would to shoot a citizen with an outstanding warrant he has just pulled over. That doesn’t make sense, and the reason is that skin color has been vastly overestimated as the driver for police killings just as behavior has been vastly underestimated.

Here are links to some of the other statistics I give:

The rates for cardiovascular disease were 373.8 per 100,000 for white males and 492.5 for black males; but we wouldn’t call heart disease, “racist”.

Black people, who constitute about 13 percent of Americans, are 1.4 percent of doctors, 38 percent of barbers, and 16 percent of cooks. They account for 14 percent of pedestrian fatalities and 74.4 percent of NBA players but just 8 percent of NPR newsroom employees. None of these are in line with their demographic representation. Is the NBA racist for having three in four players as black males? In fact, it has been hailed by The Institute for Diversity and Ethics in Sport for its “commitment to and record for racial and gender hiring practices.” Is racism behind NPR not hiring “enough” black people? Of course not.

We don’t live in the Census.

I also didn’t have a chance to mention two great pieces by Peter:  if race is driving deadly force, why are deadly force uses lower in states with more black people? And, over the past two years, homicide increased 31 percent in America’s 52 largest cities, but police shootings nationally are flat for the last three years.

As I wrote in the National Review, the best predictors of crime are broken families, living in a bad neighborhood, young mothers, and other risk factors known since the 1960s: a lack of education, nutrition, after-school activities, music, art, and other programs that create opportunity.

As I wrote with Peter in the Washington Post, behavior and context is important.

I also mentioned the 2016 NBER Working Paper Series from the National Bureau of Economic Research, entitled “An Empirical Analysis Of Racial Differences In Police Use Of Force,” Harvard Economics Professor Roland G. Fryer, Jr found that blacks and Hispanics are more than fifty percent more likely to experience some form of force in interactions with police. The study was criticized because Fryer tried to draw national conclusions from data on one agency – Houston. However, what wasn’t criticized was his data analysis. And if we look at his findings, they strongly suggest that non-deadly force is skewed against black people.

And Fryer cleaned his data. For example: “The results obtained using these data are informative and, in some cases, startling. Using data on NYC’s Stop and Frisk program, we demonstrate that on non-lethal uses of force – putting hands on civilians (which includes slapping or grabbing) or pushing individuals into a wall or onto the ground, there are large racial differences. In the raw data, blacks and Hispanics are more than fifty percent more likely to have an interaction with police which involves any use of force. Accounting for baseline demographics such as age and gender, encounter characteristics such as whether individuals supplied identification or whether the interaction occurred in a high- or low crime area, or civilian behaviors does little to alter the race coefficient.”

So Fryer controlled not just by race, but by behavior, and still came up with the numbers showing a bias against non-whites.

Context is everything. Race alone is fool’s gold. I don’t think that Mr. Montgomery cooked the books. I think that he is a great reporter, but he is not an expert in law enforcement procedure, policy, the laws governing use of force, or demographics. I think he should have stopped with his reporting – the gathering of the data – and brought in real statisticians and law enforcement experts and civil rights leadership to do the analysis, which is not as easy as sorting by race.

And I believe the reason we keep hearing these numbers – four-times-more-likely; 21-times-more-likely (Pro Publica) is that they support a narrative that explains the racial discrimination that the black community and progressive whites feel sound right.

You can’t make changes based on flawed analysis.  The real problem is finding the right data, and the right analytical techniques, to identify which officers are behaving poorly. Identifying “bad cops” is impossible when you don’t know how to identify good ones.

There are some other references: an article I ran in the Washington Post last year about analysis of traffic stops my group performed using data from a Dallas-area city whose stops did not fit any pattern of racial profiling.

Or my Post piece on the low hanging fruit suggested by my research – in which I point out that more than half of the unarmed people killed by police suffered from mental-health issues, drug intoxication, physical disability, or some combination of them. That’s something public-health policies can address head-on.

Or my New York Times piece on why protests MUST be allowed to continue.

Good analysis of good data creates the possibility to change.


As I mentioned in the afterword, after he listened to it, Peter reminded me of his blogpost from last year, Throwing Stones From Glass Houses. Peter was speaking in his post about the fact that the American Society of News Editors figured out that The Washington Post is 31 percent “minority” (and 14 percent black) in a city that is 60 percent minority, and 51 percent black. The New York Times is 19 percent “minority” (and 8 percent black) in a city that is 65 percent minority and 25 percent black.

Leave us some comments on the blog. Thanks for listening.