In 2012 I put together some data for Radley Balko on the purported rise of police killings. Last week I saw that Dara Lind prepared something similar at Vox. My data goes back a little bit further than Dara’s (her’s goes to 1996, mine to 1961), and I thought to put up what I have here.
The FBI keeps track of two types of police deaths: Accidental deaths and felonious killings, which involves the deliberate killing of law enforcement officers in the line of duty. I’ve collected three statistics related to the latter. First, number of officers feloniously killed since 1961; second, the rate of felonious deaths per 100,000 officers since 1989; and finally, average felonious deaths per five-year period since 1961. I present these statistics in chart form here, and at the end of the post I share my data file and talk about the process of obtaining these figures.
Everything is collected from the Uniform Crime Reports compiled by the FBI. One comment from the 1990 UCR report I found very interesting: “The 1990 total was the lowest since the FBI started collecting such data in the 1960s.” I was able to find online older UCR report up until 1961, and that has made me somewhat confident that my data goes back to the first years that the FBI started to keep track of this number. I’d like to keep updating this as new data comes in so that it can be a complete and easily-searchable source of for these numbers. Your help and feedback is appreciated.
Here’s the summary: In general, the job of policing has become much safer since 1961. Here are a few interesting points.
- More officers were feloniously killed in the 11 years between 1970 and 1980 (1228 deaths) than in the 21 years between 1993 and 2013 (1182 deaths).
- The rate of felonious killings per 100,000 officers has declined from about 18 in 1989 to about 5 in 2013. It was over 3 times safer to be a police officer in 2013 than 26 years ago.
- In the five years between 1971 and 1975, an average of 125 officers were feloniously killed per year. Most recently, between 2006 and 2010, the equivalent number is 50. That’s more remarkable given that the number of officers employed has increased considerably since the ‘70s.
Now the data. Click on these pictures to zoom.
Number of officers feloniously killed since 1961
I’ve put in a trendline to better illustrate the decline. The peak year for deaths was 132 killings in 1972. The safest year recorded was the most recent: 27 deaths in 2013. That’s nearly an 80% drop. The number of deaths has steadily decreased since the ‘70s, with two spikes in 2001 and 2011.
Next, felonious killings per 100,000 officers since 1989
You’ll see from the data source in the next section that the number of officers has grown from about 400,000 officers in 1990 to about 530,000 officers in 2000. Still, this decline in the rate of killings isn’t just driven by an expanding denominator (number of officers), but also a declining numerator (number of killings). The number of killings has decreased even when the number of officers grew by over 25%.
The data on the number of officers serving is really difficult to find, which is why my cutoff has been 1989, the last year for which I can get reliable data. I’ll talk more about this in the next section.
Finally, five-year averages of felonious killings
This is just an aggregation of the first chart, useful for seeing the decline of felonious killings in half-decade chunks.
Every time a police killing makes it to national headlines, voices pipe up warning of an ominous trend in the rise of police officer killings. (See Radley’s recent compilation of some of these articles.) This data indicates that policing is much safer than in the past.
2013 was the safest year recorded for felonious killings of police. It’s hard to go down from 27 deaths. Consider that an increase of 9 felonious killings of police in 2014 would be a 33% rise from the year before; meanwhile, 9 felonious deaths over the 1972 peak would be only a 5% increase.
I’ve compiled everything I’ve found into a Google Doc that you can find here. The first sheet holds the data I’ve collected, along with the source of every year’s UCR report. The next three sheets hold each of the three charts above. You’re very welcome to use it as you like, but please link to this original post or mention @danwwang.
Now some remarks about how I got the data. It was a big challenge to find some of these data points because collections are so haphazard, so I especially welcome feedback and corrections if you catch any errors.
This data comes from the Law Enforcement Officers Killed and Assaulted subreport of the FBI’s Uniform Crime Reports. I have to note as Dara has that these figures are voluntarily submitted to the FBI from local law enforcement agencies, and anywhere between 20 to 25 percent of agencies don’t do that. These figures aren’t complete, and we can be mindful that systematic under-reporting is a possibility. But I suspect that may not be a big problem; if a department lost an officer that year, it’s more likely that it will be reported to the FBI, so the rate of fatalities may in fact be even lower.
Now on to the reports. The FBI’s website links to UCRs from 1996, and I had difficulty finding earlier reports online. In the end I stumbled on what appear to be digitized copies of these reports from the Boston Public Library hosted on archive.org. Here’s a sample report from 1981: http://www.archive.org/stream/uniformcrimerepo1981unit/uniformcrimerepo1981unit_djvu.txt
There’s no page that lists all UCR reports. I found a way to get around this by manually changing URLs; thankfully, you can get from the 1981 UCR to the 1982 UCR by changing the two dates in the URL above.
Still, there are other difficulties. For example, I was just not able to find the 1988 UCR, and I’ve had to leave that row blank in the data.
More importantly, much of the digitizing on pre-1996 reports was done with software. That means that sometimes gibberish ended up in the text files. They get more difficult to read the earlier the reports go. I worry about the risk that some of these numbers have been distorted when they were digitized.
I relied on a lot of “command + F” to find various pieces of data. Hunting for some of these numbers was a real challenge, try it out for yourself. Usually, the number of felonious killings will be found after you look around the world “felonious,” and the total number of officers employed can be found around the word “civilian.”
The area of great unreliability came in the figures for the number of officers employed. Take a look at Table 30 of the 1996 report, which lists the total number of officers employed going back to 1987. For 1987, it states that the country employed 378,977 officers; meanwhile, the 1987 report counted 480,383 officers. Rather than hunt down older sources, I decided to make 1989 my cutoff. If you’re curious, though, I’ve included the number of officers from original UCR reports in comment boxes.
Feedback, questions, corrections? Leave a comment, or reach out at firstname.lastname@example.org or to @danwwang.