Alternative sources of data on police homicides

There were empty seats at the table on Thursday for young males of color who have been shot by police, most recently Tamar Rice, a 12-year-old boy who was carrying a toy gun. His case resonates powerfully in Santa Rosa where, last year, 13-year-old Andy Lopez was shot for the same reason. He is memorialized in this moveable mural currently on display at the Peace and Justice Center around the corner from Luann’s studio:

Our son is an Airsoft enthusiast, just like  Tamar Rice and Andy Lopez were. Unlike them he is white. In circumstances like theirs, would that have made the crucial difference? We think so. But when you look for data to confirm or reject that intuition, it’s thin and unreliable.

Criminal justice experts note that, while the federal government and national research groups keep scads of data and statistics— on topics ranging from how many people were victims of unprovoked shark attacks (53 in 2013) to the number of hogs and pigs living on farms in the U.S. (upwards of 64,000,000 according to 2010 numbers) — there is no reliable national data on how many people are shot by police officers each year.

– How many police shootings a year? No one knows, Washington Post, 09/08/2014

The one available and widely-reported statistic is that, in recent years, there have been about 400 justifiable police homicides annually. In Nobody Knows How Many Americans The Police Kill Each Year, FiveThirtyEight’s Reuben Fischer-Baum reviews several sources for that number, and concludes that while it’s a reasonable baseline, the real number is likely higher.

Fischer-Baum’s article, and others he cites, draw on a couple of key Bureau of Justice reports. They are not hard to find:

https://www.google.com/?q=site:bjs.gov+”Justifiable+Homicide+by+Police

One report, Policing and Homicide, 1976-98: Justifiable Homicide by Police, Police Officers Murdered by Felons [1999], says this about race:

A growing percentage of felons killed by police are white, and a declining percentage are black (figure 4).

Race of felons killed
1978 50% White 49% Black
1988 59% White 39% Black
1998 62% White 35% Black

Felons justifiably killed by police represent a tiny fraction of the total population. Of the 183 million whites in 1998, police killed 225; of the 27 million blacks, police killed 127. While the rate (per million population) at which blacks were killed by police in 1998 was about 4 times that of whites (the figure below and figure 5), the difference used to be much wider: the black rate in 1978 was 8 times the white rate.

A more recent report, Homicide Trends in the United States, 1980-2008, is one key source for the widely-cited number of 400 justifiable police homicides per year:

Interestingly, the gap between justifiable police homicides and justifiable citizen homicides has widened in recent decades.

Table 14 addresses race:

It’s a complicated comparison involving the races of shooters and shootees when the shooters are civilians and also when they are police. In the latter case, the trend noted in the earlier report — “a declining percentage [of ‘felons’ killed by police] are black” — has reversed. Combining this table with the previous one, we get:

% of blacks killed by police
1978 49%
1988 39%
1998 35%
2008 38%

What is our level of confidence in this data? Low. From How many police shootings a year? No one knows.:

“What’s there is crappy data,” said David A. Klinger, a former police officer and criminal justice professor at the University of Missouri who studies police use of force.

Several independent trackers, primarily journalists and academics who study criminal justice, insist the accurate number of people shot and killed by police officers each year is consistently upwards of 1,000 each year.

“The FBI’s justifiable homicides and the estimates from (arrest-related deaths) both have significant limitations in terms of coverage and reliability that are primarily due to agency participation and measurement issues,” said Michael Planty, one of the Justice Department’s chief statisticians, in an email.

Are there other sources we might use? Well, yes. Wikipedia is one place to start. It has lists of killings by law enforcement officers in the U.S.. The introductory page says:

Listed below are lists of people killed by nonmilitary law enforcement officers, whether in the line of duty or not, and regardless of reason or method. Inclusion in the lists implies neither wrongdoing nor justification on the part of the person killed or the officer involved. The listing merely documents the occurrence of a death.

The lists below are incomplete, as the annual average number of justifiable homicides alone is estimated to be near 400.

Each entry cites a source, typically a newspaper report. About once a day (or maybe about twice day), a police officer shoots a civilian somewhere in the U.S. That’s a rare and dramatic event that will almost certainly be noted in a local newspaper. The report may or may not provide complete details, but it’s an independent data point. Everything else we know about the phenomenon is based on self-reporting by law enforcement.

To an analyst, the data that lives in Wikipedia tables is semi-structured. It can be extracted into a spreadsheet or a database, but extracting fully structured data almost always requires some massaging. The script I wrote to massage Wikipedia’s lists of police homicides handles the following irregularities:

1. From 2009 to 2009 the table lives in a single per-year page. From 2010 onward, the per-year pages subdivide into per-month pages.
2. The city and state are usually written like this: Florida (Jacksonville). But for the first five months of 2012 they are written like this: Jacksonville, Florida.
3. The city name, and or the city/state combination, is sometimes written as plain text, and sometimes as a link to the corresponding Wikipedia page.
4. The city name is sometimes omitted.

The script produces a CSV file. The data prior to 2009 is sparse, so I’ve omitted it. Here’s a yearly summary since 2009:

year  count
----  -----
2009     60
2010     82
2011    157
2012    580
2013    309
2014    472

It looks like this listmaking process didn’t really kick into high gear until 2012. Since then, though, it has produced a lot of data. And for one of those years, 2012, the count of police homicides is 580, versus the Uniform Crime Report’s 426. Each of those 580 incidents cites a source. Here’s one of them I’ve picked randomly:

One person is dead following an officer-involved shooting in Anderson, according to police.

Investigators said they were called to Kings Road about 2:45 a.m. Thursday for a domestic-related incident between a husband and wife.

Coroner Greg Shore said three officers went inside the home, and that’s when a man pointed what appeared to be a gun at officers. At least one officer fired his gun, and the suspect died on the scene, Shore said.

Investigators said the man’s wife, who was inside the home, was taken to the hospital due to injuries from a physical fight with the suspect. Shore said she was doing OK.

Shore said the man officers shot was 47-year-old Paul Leatherwood. Sgt. David Creamer said officers have been called to the same house six times in the past six months for domestic-related issues.

The State Law Enforcement Division is headed to the scene to investigate, which is standard protocol for officer-involved shootings in South Carolina.

– One killed in Anderson officer-involved shooting, FoxCarolina.com

We don’t know the race of the shooter or the shootee. We don’t know whether what appeared to be a gun turned out to be a gun. We don’t know whether this was or wasn’t reported as a justifiable homicide. But we could find out.

Every week, a million people listen to the blockbuster podcast Serial, an investigation into a cold case from 1999. A staggering amount of cognitive surplus has been invested in trying to figure out whether Adnan Syed did or did not murder Hae Min Lee. In this blog post, which I picked randomly from the flood of Reddit commentary, a lawyer named Susan Simpson has produced a 14000-word densely-illustrated “comparison of Adnan’s cell phone records to the witness statements provided by Adnan, Jay, Jenn, and Cathy.”

With a fraction of the investigative effort being poured into that one murder, we could know a lot more about many others.