Skip to main content

View Diary: RKBA: More guns...less crime? (146 comments)

Comment Preferences

  •  The lowest figure (6+ / 0-)

    for defensive gun uses is what? 500,000 per year? So there are over 5 million murders per year caused by average joe's losing their cool in arguments?

    And we are to take you seriously? Bullshit. Pure bullshit. Please.. just be honest and not resort to such blatantly outrageous dishonest hyperbole.

    •  Ok.. so 108k (4+ / 0-)

      as stated by KV. That makes over a Million murders then. My point still stands.

    •  That's not the lowest figure. That's the lowest (7+ / 0-)

      mean minus uncertainty of all the studies.

      The means of the various studies cluster around 2 million.

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

      by Robobagpiper on Thu Aug 29, 2013 at 07:51:31 AM PDT

      [ Parent ]

      •  with error bars (2+ / 0-)
        Recommended by:
        Kasoru, DefendOurConstitution

        at least that high.

        The best way to pin down the number is work backwards from justifiable homicides, since that's the only reliable figure we have.

        This numbers in the ~200ish range for civilians. Then, you divide by the ~15-20% mortality rate for gunshots to get ~1-1.5k justifiable woundings.

        Then, the hardest part is trying to pin down how many DGUs result in an actual shooting, where a bullet hit the target. From Kleck this was a bit over 8%, putting actual, justifiable DGUs around 10-20k, though I could easily see this number being off a factor of two each way.

        However, to suggest that ~2 million justifiable DGUs occur each year is to strain credibility.

        •  However - DGU (8+ / 0-)

          does not have to mean that a person was shot. Let's say that a person or a group of people is about to assault me. I pull my handgun and the threat evaporates- they leave.  That is a defensive gun use. But it would not be included in your numbers as no shots were fired.

          •  You didn't read what I wrote: (0+ / 0-)

            so I'll repeat it:

            Then, the hardest part is trying to pin down how many DGUs result in an actual shooting, where a bullet hit the target. From Kleck this was a bit over 8%, putting actual, justifiable DGUs around 10-20k, though I could easily see this number being off a factor of two each way.

        •  Nope; the error bars for the individual studies (8+ / 0-)

          are about 60% of the mean value, not "at least that high". And it should be noted that it's an abuse of statistics to suggest that the high value of the uncertainty range is "less likely" than the low value. Statistically, they are equally likely, whatever your bias.

          But that's for individual studies. When you combine 19 studies with similar sample sizes, all of which are producing in-family mean values, the uncertainty drops by about three-quarters, because the aggregate sample size of positive respondents goes way up.

          The number really is about two million, plus or minus a couple hundred thousand.

          Working backward from justifiable homicides is a poor methodology, because it fails to come close to the number estimating DGUs where no one is wounded, much less the weapon not fired.

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

          by Robobagpiper on Thu Aug 29, 2013 at 08:48:55 AM PDT

          [ Parent ]

          •  I addressed that (1+ / 0-)
            Recommended by:
            oldpunk

            in my last point. Working backwards from the justifiable homicides only requires ONE number from those surveys, the percentage of time a DGU results in a shooting, and if you use the number from Kleck you get ~8%. If you want to find a better number, than go through those 19 surveys and come up with a mean. However, since Kleck estimated 2.5 million DGUs, and you're saying around 2 million, chances are your mean percentage would be close to 6-8% as well, which means that the projected DGU working backwards wouldn't be significantly different from the 10-20k estimated above.

            Using your methodology, you rely on a self-reporting survey for ALL of the numbers, and based on the small sample size and the danger of false-positives, you get widely varying results with extremely high error bars.

            Those 60% error bars you mentioned are just from the base statistics, not from a consideration of systematic errors like false positives, and could easily be off by a factor of 2, making the estimate worthless.

            Your statement that 'working backwards' is a poor methodology relies on the circular argument that your DGU estimate is actually correct to begin with, i.e. you're assuming your conclusion.

            If Kleck's survey indicating 8% DGU results in a shooting is off by x100, as it would have to be to make the numbers consistent with your estimate, then how the hell can you trust any number from a self-reported survey?

            •  You are ignoring three factors, both of which (5+ / 0-)

              are pointed out by Kleck.

              1) Only about 10% of gunshot injuries are life-threatening. So even if those 8% are hit, a criminal without a life-threatening injury has an incentive to go to alternative medical facilities, since all gun shot wounds in hospitals are reported to the police. That's a potential for an order of magnitude error there.

              2) The 8% Kleck cites as believing they hit their target, Kleck would be the first to tell you, is based on an absolute number smaller than the error bars. Kleck's primary criticism with the widespread misuse of his survey is that while the complete sample is statistically valid (because the error is related to the total sample size), the error of most subgroups is higher than their actual value (because their uncertainty is determined by the much smaller sample size of positive respondents). It's why you also can't work back to the estimated number of thwarted crimes (other than the biggest, burglary).

              3) And, of course, this assumes that the majority of people who think they hit their target did.

              So right there, your working backwards from justifiable homicides has got systematic errors that introduce errors up to two orders of magnitude.

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

              by Robobagpiper on Thu Aug 29, 2013 at 12:48:38 PM PDT

              [ Parent ]

              •  I ignored no such thing (0+ / 0-)

                1) the percentage is 15-20%, not 10%, and this is based on hospital admitting statistics. If you consider people shot during a DGU may be less likely to seek medical attention, this number could plausibly even be higher. However, I didn't 'ignore' this issue as you claim, the percentage is baked right into the numbers and I was explicit in the discussion, how did you miss this? It's how you go from ~200 killed to 1000-1500 shot.

                2) Unless you can cite a better sourced number, we use what numbers we have. This is the only number that isn't 'solid' so to speak, for the comparison, and therefore would have to be a factor of 100!! lower for the validation to check out.

                Do you want to claim that people miss their targets 99% of the time, even when they claim otherwise? Is that supposed to make people feel good about DGU?

                So, no, there are no '2 orders' of magnitude wiggle room unless you claim that the average DGU misses 99% of the time that they think they hit.

                And if there are two orders of magnitude discrepency in the only number of this validation that relies on the self-reported survey numbers, well, that should tell you a little bit about the DGU numbers from self-reported surveys, now shouldn't it.

                You're in the position of saying that the DGU surveys are reliable even though individual parts are off by two orders of magnitude. That's pretty wacky.

Subscribe or Donate to support Daily Kos.

Click here for the mobile view of the site