And now, the NY Post channels bad reporting by the Associated Press. What you do is take a press release and add some two-bit "analysis."
WASHINGTON — The number of law enforcement officers killed by firearms in the US jumped by 56 percent this year ... The annual report by the nonprofit National Law Enforcement Officers Memorial Fund found that 50 officers were killed by guns this year, compared to 32 in 2013. - AP, 12/30/14
and that
126 federal, local, tribal and territorial officers were killed in the line of duty in 2014. That’s a 24 percent jump from last year’s 102 ...
The sharp increase in gun-related deaths among officers followed a dramatic dip in 2013, when the figure fell to levels not seen since the 19th century.
Really? "Shootings were the leading cause of officer deaths in 2014 followed by traffic-related fatalities, at 49." Yes. 50 versus 49 makes that the "leading cause." Try "Officers were as likely to die in a traffic accident as they were to be shot."
And how about this: "This year’s uptick comes amid increased tension between police and the public following the high-profile deaths of unarmed black men by white police officers, including Eric Garner in New York and Michael Brown in Ferguson, Missouri.
I hate to tell the good folks churning out bad copy at the AP, but the two are completely unrelated. Apparently the Memorial Foundation and the AP think there's some sort of link to be made between the "uptick," police killing unarmed people, and "ambush" police killings (there were 15; no data is presented to show that this is greater or lesser than in any previous year).
In the one recent case in NYC, two officers were shot in one ambush. I don't know if that counts as a single ambush or two ambushes. Whichever, it's the only one where there has been any hint of a connection with Garner or Brown. Whether it was a real connection, or Facebook fabulism by a sick man, probably won't ever be answered.
The Moral About Lies, Damned Lies, and Statistics
Dear AP writers and editors,
1. ... don't write stories based only on a press release.
That's not journalism.
2. Four words: regression to the mean.
Is This News?
Not the way the story slants it.
If you can, go to the source. That's what I did. And that's why I find it infuriating that major news outlets don't puta link to the reporting organization in their stories as a matter of journalistic integrity.
Unlike many organizations issuing press releases based on reports based on their own data the Memorial Fund has a table of all the officers killed, by year, back to 1791, and they do put their analyses on their website. Obviously the numbers from 105 or 200 years ago aren't very reliable.
The Scary Jagged Graph
Chart from the Memorial Fund website.
Up By 50 Percent! OMG!! By 24 Percent !!!... not
A quick look at the numbers shows that there's nothing at all out of line with 126 deaths in the profession, in fact, that number is still below the long-run average. Why the number for 9/11 is emphasized in their Scary Chart, I don't know. It's clearly an outlier.
Put one way, the number of deaths returned to the level it was in 2009 (125), which historically was as low as it has been since 1960 (129). The fact is that the number was 102 in 2013 may well have been a fluke. Nevertheless the AP characterized it as "levels not seen since the 19th century." I think reaching back 115 years or more puts you in the realm of taking those numbers with a grain of salt. And that still may not tell the real story -- see below.
In fact the overall trend is downward, even as the number of police has increased.
Average Annual Number of Officers Killed
-- in the past 100 years: 175
-- for the past 75 years: 161
-- for the past 50 years: 181
-- for the past 25 years: 161
-- for the past 15 years: 157
-- for the past 10 years: 150
-- for the past 5 years: 136
Put pictorially in my
cheap but useful infographic,
The rightmost green column represents the 126 reported deaths (motor vehicle, guns, other). The line carries that number across the cloumns showing long-run and recent averages. The line is below average for any period (based on Memorial Fund data).
Hence: regression to the mean. In plain English, a measurement of some phenomenon that occurs in a population will move up and down due to random factors. In the long run, a series of low measurements will tend to move back to the mean (average) value. High measurements, ditto. Over time, the mean may shift some, but there will always be higher and lower values.
So going from last year to this year as the only comparison is misleading.
The Memorial Fund would probably be best off reporting something like a moving average of five, six, eight years. Moving averages smooth out random variations.
Are Police Safer Today?
Another very useful measure is "per 1,000" - since in a given occupation, 100 deaths per 100,000 is the same as 200 per 200,000.
The difference in the simple total looks horrendous (100,000 leaps to 200,000) or like an amazing improvement (death "dramatically" cut in half). It's also a standard measure in where incidence within the population is being discussed.
The MF doesn't provide this data. So we can surmise that a cop is much safer now -- since there are many more cops -- than they were in 1960 when there were a comparable number of deaths. And certainly safer than they were in the 19th century.
Killed By Cars
The real story might well be deaths per 1,000 in car crashes. Is gun violence really the killer the story tries to picture?
While the crash numbers aren't available on the website, the MF's chart hints that a police officer is as likely, or more likely, to die in a traffic accident than they are to be shot. Look at the difference between the uppermost line on their chart and the middle line. That seems to show a lot less variation than the total, but since the three causes are stacked, it's visually difficult to decipher accurately.
This wasn't the case back in the late 60s and early 70s, but starting around 1985 the two causes appear to track very closely. The story hopes you never look at the website and the MF's chart and never gain historical perspective. You're just supposed to flip out that this year was so much worse.
Presentation Is Perception
It would be interesting to see the same chart with three data series superimposed rather than stacked, or with Guns on the bottom and "Other" on the top, to see if the reader takes away the same message.
See also:
Fatalities Report 2014
http://www.nleomf.org/...
Press Release
http://www.nleomf.org/...