If you want to get a handle on the numbers involved in gun violence, you can go to two sources: the CDC or the FBI.  The numbers aggregated by the CDC come from coroner’s reports received by state health departments and then forwarded, analyzed and presented on the CDC website WISQARS, which tracks fatal and non-fatal injuries since 1999 and 2001, respectively. The other method is to use the crime data from the FBI, whose numbers begin in 1960 but become state-based beginning in 1985.

The data in these two reports is, to put it politely, somewhat diffuse.  Take one year for example, in this case 2005.  According to the FBI, 16,740 people were victims of murder or manslaughter, the CDC listed the total number of homicides as 18,124.  This 10% difference between the two numbers is more or less the same for every year in which both agencies report their numbers, and it reflects both different definitions (one is reporting medical events, the other reporting crimes) and both numbers are estimates reflecting the fact that state and local agencies which report the raw totals are not necessarily required to report anything at all.

Where things really get crazy is when we look at CDC and FBI numbers for what is referred to as homicide by ‘legal intervention,’ which is a polite way of saying that someone got shot by a cop.  In 2010, to choose a different year for comparison, the FBI put this number at 397; for the CDC it was 412.  For the years 2010 – 2014, the FBI says that 2,142 people were killed by law enforcement, the CDC number is 2,485.  So now we have a gap between the two estimates of nearly 15%, but that’s not even scratching the veritable surface when it comes to figuring out what’s what.

I was tipped off to this problem by a story in MedScape that focused on the research of a group at the Harvard School of Public Health who have been looking at the data on cop killings since 1960. They recently published an op-ed on this problem citing an enormous discrepancy between the ‘official’ numbers on legal intervention deaths and what is now being reported by, of all media outlets, The Guardian, which happens to be a newspaper published in the U.K. The reason I find this interesting is because cop killings in England are so rare that in 2013, police in the U.K. only shot off their duty weapons three times and, by the way, didn’t kill anyone at all.

The Guardian has created a website, The Counted, which has been collecting and publishing stories about legal interventions since 2015, and I have to tell you that the numbers are frighteningly higher than anything posted by the CDC or the FBI. In 2015 the site lists 1,140 persons killed by the police, so far in 2016 the number has reached 136.  At this rate the total for 2016 will only be 1,013, a 10% decrease from last year, but still more than twice as high as what we get from our usual sources at the FBI or the CDC. Actually, my friends at the Gun Violence Archive also post a daily count on what they call “officer involved shootings,’ and so far this year their death toll stands at 145.

I’ll leave the two aggregators to figure out whose number is more exact, but the bottom line is that cop killings are much higher than what is usually assumed to be the case, and they occur most frequently in African-American ghetto neighborhoods – gee, what a surprise! The problem with the data found in the Guardian’s website, however, is that it is very incomplete. Try filtering for any attribute – race, age, gender, weapon – the numbers fall way short. Deriving stories from media notices is one thing, aggregating objective data is something else. If public health researchers want to get their hands on real data they better be prepared to wait, and wait, and wait.