Wednesday, December 4, 2013

USA Today Actually Checking Their Sources: Excellent!

This is what journalists are supposed to do.  From December 3, 2013 USA Today:
When USA Today started investigating mass killings, it seemed a fairly straightforward thing to count: How many times have at least four people died at the hands of another in a single incident?

Yet marking the death toll of mass killings in America is anything but simple. It’s hampered by the FBI’s voluntary reporting system that gets it right a little more than half the time, and by advocacy groups who may count only incidents that support their cause, ignoring killings that don’t involve a gun or did not get heavy media coverage.
It is an amazing story -- it means that the data that the FBI released on mass murders 2006-2011 was pretty much worthless (only 61% accurate), with multiple individual murders lumped together into a single mass murder; a mass murder that was not:
In another incident, the FBI data included the deaths of two adults and three children in a Cleveland suburb in 2009, shot to death by a 28-year-old man. There was a drive-by shooting at that date and time, but no one was killed. ...
In nearly a dozen cases, USA Today — searching media reports and interviewing local law enforcement agencies — could find no record of a murder, even when the FBI data showed as many as seven killed. Among them: a quintuple murder in Newark in 2010. What actually happened: Police arrested two men in connection with the murder of five teens in 1978 — 32 years earlier.


  1. It seems the quintuple murder was in fact right, then. It was just the date that was wrong.

  2. asdf, I think that listing the quintuple murder was not right. Apparently the bodies were never located and the case hinged on an inmate's testimony. The person charged was found Not Guilty.

    Thus, when it comes down to it there is the possibility that there may have never been any murders in the first place. Or if there were murders, they were not in that location.

    A link to one of the articles:

  3. First, off, I would point out to asdf that we need the years to be right as much as anything else: one of the things we are trying to do is figure out the rate that these things happen.

    Second, I would point out that intent matters as well. If someone shows up at a school and shoots people, and only three die, but a dozen are shot, I would consider that a mass murder; similarly, if someone shows up at a school with six guns, lots of ammo, and starts shooting, but gets shot, and no one dies...that should count as a mass murder too (or at least, as an attempted mass murder). Counting only situations where four or more people die skews the results against situations where people successfully defend themselves against a potential mass murderer.

    Somewhat off topic, it drives me nuts when people point to Australia, and say "They banned guns, and they haven't had a mass shooting since then!" I've checked up on mass murders in Australia, and it so happens that they get, on average, about one incident every fourteen years. They've *had* one incident since they've banned guns, but it didn't involve shooting. This isn't to say that they can't have three mass murders in the next month...but what it *does* show, is that mass murder is so rare in Australia, that you can't attribute the absence of it to their gun laws.

    And this blog post makes me wonder how accurate this "mass murder incidence report" (in this case, Wikipedia) is all that accurate, one way or the other...

  4. Thanks for the extra info, guys. I see your point now.

  5. Re the FBI Cleveland item: one thing that clueful commenters on cross-national stats point out is the different concepts of what is a homicide. In Britain, which should be one of our closest polities in both ancestry and cultural aspects, they don't count homicides until the cases are resolved.

    This one about the long-ago Cleveland murders is just a reminder that our system, though much better in conceptual terms, can still end up making mistakes like this one. But at least it's not "working as designed".