Skip to main content

View Diary: Repeal the Ban on Gun Violence Research (82 comments)

Comment Preferences

  •  Let me begin to list the problems with (2+ / 0-)
    Recommended by:
    MGross, Neon Mama

    the study.

    Kellerman used a case control methodology and compared incidents of shootings in homes in three counties, to similar demographics where shootings didn't occur. He based his data on the shootings on police reports, and used them to resolve a few questions:

    Was there a shooting in the house?
    Was the homeowner shot?
    Was an intruder shot?
    Was there a gun in the house?

    He did not ask if the shooting was justified, or if it was a police action, or any details beyond this.

    He then compared this to nearby households where no shooting took place, and surveyed them to see if they had guns in the home to get a baseline measure of gun ownership in homes where no shooting took place.

    His initial study found the risk factor 47x as great, nearly all of which were suicides, for which he took withering criticism. When he excluded suicides in a second study, he got a risk factor 2.7x as great. He has since gone on the lecture circuit and was videotaped claiming (without supporting evidence) a risk factor 18x as great.

    Let's look at the problems:

    1) Kellerman used a case control methodology. It's a cheap way to do research, but one thing it can not do is establish causality, nor can it account for reverse causality.

    2) Kellerman's papers were "special articles" and were not peer-reviewed. He has never released his raw data, citing only percentages of the demographic groups in his shooting and control groups, so the research can not be independently verified.

    3) Kellerman's study group, and control group, were in poor, high-crime, minority communities. 51% of his control group had arrest records, and 60% were African-American. This is not a sample group that can be extrapolated to the nation at large.

    4) Kellerman's study of the defensive shootings only counted those where a shot was fired, and a police report was generated. It did not account for defensive uses where no shots were fired, which (as other studies have shown) make up the vast majority. It did not account for cases where the homeowner was a criminal shot by cops.

    5) Kellerman's study of the control group revealed a consistent error medical professionals make when dabbling outside their field into criminology - the failure to account for false negative responses when asking about criminal behavior. The reported gun ownership rate in the control group was contaminated by the fact that half of them had arrest records, and of those, most would have been prohibited possessors; and those that were not were in a bad neighborhood. He did not account for the number of people who would not admit to a stranger on the phone that they had a gun in the house.

    6) Kellerman didn't account for reverse causality - that high rates of criminal violence would drive up gun ownership rates, and only assumed the reverse, as a risk factor.

    7) Kellerman's results can be completely reproduced by taking a small high-risk population and conflating it with a large low-risk population; where his "risk factor" is actually a measure of homicide rate with gun ownership as a confounding variable.

    8) Kellerman had close but undisclosed ties to the lobbying organization Handgun Control (the old name of Brady).

    Basically, his studies are so methodologically flawed as to be worthless. This did not stop the scientifically worthless study from being widely cited by policymakers as justification for strict gun control.

    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 Tue Jan 15, 2013 at 04:58:22 AM PST

    [ Parent ]

Subscribe or Donate to support Daily Kos.

Click here for the mobile view of the site