Skip to main content

View Diary: RKBA: Myth Busting (253 comments)

Comment Preferences

  •  Be prepared for people citing TomP's recent (8+ / 0-)

    diary as a refutation.

    The problem is, definitions. The definition of mass shooting in this article is the FBI one - 4 deaths. The Mother Jones definition is that, plus not crime-related, in a public place, and a single shooter. The "active shooter" report TomP diaried requires that the shooting be not crime-related, in a public place, but doesn't specify the number of deaths (but the median was 2).

    The Mother Jones-defined "mass shootings" will be a small faction of the FBI-defined ones, but are at least a subset. The problem is, when you're dealing with a "small fraction of ~18 a year", you're in territory where statistical uncertainties go through the roof. But it does confirm something I've suspected - that most of the FBI-defined "mass shootings" fall into the criminal category, or not in a public place, and were not the sort that people imagine when discussing these incidents as a political issue.

    The "active shooter" report was interesting. First I have to question the possibility of early undercounting - even the phrase "active shooter" is a new one for our media, so I have to wonder how much of the apparent rise is due to a change in reporting rates by local law enforcement to the national level of incidents they might have not done so before the post-Newtown/Navy Yard awareness.

    But let's say that that's not an issue, and that prior to 2011 there were 4 incidents a year and afterwards there were 16. If true, that would represent an increase in the number of this type of event, how much of a change does it represent to the rate seen in either the Mother Jones or the FBI definitions?

    The answer is, not much. The "active shooter" report included all such incidents, regardless of the number of deaths, meaning that with a median of 2 (considering that a number of incidents would have included no deaths), and the nature of small-number statistics, only a tiny fraction of either the pre-2011 4 per year or post 2011 16 per year would reach the 4-death threshold. In other words, especially for the FBI definition, any increase using the "active shooter" definition would disappear in the year-to-year statistical uncertainty.

    Non enim propter gloriam, diuicias aut honores pugnamus set propter libertatem solummodo quam Nemo bonus nisi simul cum vita amittit. -Declaration of Arbroath; μολὼν λαβέ - att. Leonidas I

    by Robobagpiper on Fri Jan 17, 2014 at 07:55:34 AM PST

    •  I don't like the Mother Jones definition since it (4+ / 0-)

      leaves out Columbine, which is a mass shooting by every other definition I've ever seen.

    •  It's probably more true in gun statistics (3+ / 0-)
      Recommended by:
      Robobagpiper, ColoTim, 88kathy

      than most things that one must be very careful in citing or interpreting the numbers, for the reason you illustrate above.  Not just re: "mass shootings" but in any number of other definitions used to promote one viewpoint or another regarding "gun violence".  Even terms such as "deterrence" or "accident" can include greatly disparate datasets depending on how one wishes to shape the argument.

      You can't spell CRAZY without R-AZ.

      by rb608 on Fri Jan 17, 2014 at 08:38:07 AM PST

      [ Parent ]

      •  Undercounting is a huge ignored issue (4+ / 0-)

        People ask "why do you focus on deaths when the true cost of guns is much greater!". Leaving aside the assuming the conclusion bit, it's because deaths are the stat least likely to suffer from undercounting, or changes in reporting. As such, they end up being a proxy for trends in other related types of incident where changes in reporting rates could muddle things.

        We have to remember when we look at lists of the number of incidents of something a year, that this requires someone to have recorded the incident, and reported the recording through a chain of custody to the agency that tabulates it. The greater the impact of the incident, the more likely it gets reported up the chain and lands in the final counted pile. The smaller the impact, the more likely reporting on it will remain confined to where the incident happened.

        Case in point: the number of gunshot injuries a year. The CDC lists a number of gunshot injuries per year. It's from the CDC, it must be valid. Except...

        When you look at how the CDC comes up with the number, things become less clear. Basically they have a group of hospitals they sample. These hospitals report the number of cases that come through their doors, the CDC tabulates them, does a statistical analysis to extrapolate that (admittedly large) sample of hospitals to the general population, and voila! The number of gunshot injuries in the US per year.

        But there are several problems:
        1) This completely excludes any gunshot injury that is not treated in a hospital or emergency room. Of those that are, only something like 10-15% are life-threatening, and one would expect that those where the person shot sought hospital treatment would be biased toward this type. Indeed, of people who admitted to government interviewers in the National Crime Survey (parent to the National Crime Victimization Survey) to having been injured by gunshot, only half said they got any treatment at all beyond, presumably, basic first aid.

        2) In addition to the "no treatment at all" option, there will be all those treatment options that don't involve a hospital or emergency room - whether above board or underground.

        3) Even for those treated in a hospital or emergency room, the historic reporting rates to the CDC are pretty poor, when you look at them closely. Why? Because not reporting an incident takes zero effort, while reporting one does. And collecting data for the CDC is not the hospital's primary mission. It's not even really its secondary mission. There is a fairly strict requirement that gunshot injuries have to be reported to the police, and for immediately obvious reasons beyond this requirement, reporting to law enforcement would be something one would expect to be more routinely done. Indeed, a study of Galveston hospitals performed in the 90s collected all gunshot injury reports for a several year period from police, hospitals, and emergency rooms. The total number of incidents was more than double those reported by hospitals. Emergency rooms had a better rate, and police reports captured 90% of all reported incidents. Meaning that the number reported by the CDC, which again only represents gunshot injuries treated in hospitals, is probably low by a factor of two! Now, this is form of underreporting that could easily be affected by changes in record-keeping technology, so it would be important to conduct this study again to see how things stand today.

        Needles to say, though, the actual number of gunshot injuries in the country could be 4x - or more - that reported by the CDC, simply from the method by which incidence rates are collected.

        Non enim propter gloriam, diuicias aut honores pugnamus set propter libertatem solummodo quam Nemo bonus nisi simul cum vita amittit. -Declaration of Arbroath; μολὼν λαβέ - att. Leonidas I

        by Robobagpiper on Fri Jan 17, 2014 at 09:09:37 AM PST

        [ Parent ]

      •  First rule of statistics (2+ / 0-)
        Recommended by:
        rb608, greengemini

        Now they have the 2nd (safety net for sloppy) Amendment, and can't be infringed to actually treat their gun like a gun and not a video game controller.

        by 88kathy on Fri Jan 17, 2014 at 09:17:52 AM PST

        [ Parent ]

Subscribe or Donate to support Daily Kos.

  • Recommended (183)
  • Community (81)
  • Baltimore (76)
  • Bernie Sanders (57)
  • Civil Rights (47)
  • Freddie Gray (43)
  • Elections (35)
  • Culture (34)
  • Hillary Clinton (32)
  • Law (29)
  • Racism (28)
  • Labor (27)
  • Education (24)
  • Economy (24)
  • 2016 (24)
  • Media (23)
  • Politics (22)
  • Rescued (22)
  • Texas (21)
  • Barack Obama (19)
  • Click here for the mobile view of the site