Thursday, September 27, 2007

Final stake in Kellermann

There is no greater anti-gun-owner BS that is constantly cited than the claim that you are "X times more likely to be killed by the gun in your home used by someone in the home than by an intruder." This is even more damaging PR than the "Assault weapons are super-deadly" nonsense.

The best debunking of the Kellermann "realstats" was posted way back when as a comment to the Deltoid blog, run by famed Aussie anti-gun-owner statistician Tim Lambert, who never hesitates to criticize pro-gun-owner stats while pulling every book off the shelf to defend anti-gun-owner stats.

Kellermann's data never included an ID on which "gun" was used in the homicide, and if it was legally owned to begin with.

Taking Lambert below at face value (which the commenter below disputes), Kellermann's data also shows that only 1 in 4 homicides "in the home" were committed by a resident using any gun.

Taken together, these data points show there is no support for the idea that merely owning a legal gun is a "high risk factor" for homicides in the home.

In the interests of keeping this Deltoid exchange from falling down The Memory Hole, I am posting the best debunking of Kellermann here. I edited out the extraneous stuff about Kellermann's data not being available (it actually is) or other stuff about John Lott's credibility. When I do that, you will see ellipses (" . . . .").

1. Andrew Salemme Says:
June 9th, 2005 at 12:00 am

Questions and some Answers

Also the foundation of Kellermann’s conclusion falls apart if the guns used were guns brought from outside the home. According to of all of the homicides used in the three county study 4.6 % of the victims were killed by a gun that was from their own home. Kellermann’s data of the matched pairs (316 cases I believe) shows that “no more than 34% were killed by a gun from the victims home,” which falls to 3.5% when talking about matched cases, according to GunCite. That means the conclusion (family fights turn deadly and that guns “pose a substantial threat to members of the household”) of Kellermann’s study is largely unsupportable does it not? Mr. Lambert has previously stated that the studys findings make it impossible for a significant number of guns to have been brought from the outside- however this does not appear to be the case upon closer review, therefore should this study not be cited very critically.

I am confused as to why this study has been defended so strongly by Mr. Lambert when the conclusion cannot be supported by the subsequent data that shows the guns used turn out to be not from the victims home in the majority of cases. This study in my opinion is an attempt to prove that “normal” people committ crime, but this cannot be justified by the evidence to the contrary.

(Kellermann does state that 62 % of the proxy respondents said a gun was kept in the house of the victim (as to 93%)which also could lessen the strength of the conclusion, right?)


2. Tim Lambert Says:
June 9th, 2005 at 12:00 am


I did not say that it was impossible for a significant number of guns to have been brought from outside. I said that only a small percentage were by intruders. An intruder is not the same as a non-resident. I examined the data and found that 96 out of 196 gun homicides were committed by family or roommates — people likely to have used the gun in the home. As was clearly shown in the paper and consistently ignored by critics, all of the extra risk associated with gun ownership was associated with homicide by residents.

3. Andrew Salemme Says:
June 9th, 2005 at 12:00 am


I do however, believe that in one post you wrote in response to someone about how the “gun in the home” is so dangerous- “Because a significant number of the homicides were committed with the gun in the home.” Which implies that the guns used were not brought from the outside. Also you do say significant and I don’t believe 49% is significant in the sense that an ordinary observer sees it.

Also in an article written by Steven Kangas (which I assume you looked over and helped with since he does say he is deeply indebted to you for your help) Mr. Kangas writes that “the study’s findings make it logically IMPOSSIBLE for a significant number of guns to have been brought from the outside. I believe that this point needs to be constructively criticized by you. You also say that the people were “likely to have used the gun in the home,” but the percentage 49% is still not a large majority its not a majority at all, thus the conclusion is weakened and thus should be cited as such. Also “likely” is much different then the strong statement of “a significant number of homicides were committed with the gun in the home.” While the paper does show there is extra risk associated with gun ownership a person still cannot conclusively state based on this survey that families owning guns are more likely to be murdered by their own gun across the United States, which is what the study is so often cited for. The study shows a correlation between guns in the home and homicide with a gun in these three urban counties. It is possible that if someone were to conduct a study in three rural counties guns in the home and homicide with a gun would not show the correlation that the Kellermann study shows. My point is the use of the study without citing it critically as do Cook and Ludwig in their book, Evaluating Gun Policy, is misleading to the general public.

So is it not true that the Kellermann studies conclusion should be cited critically (just as Klecks, Kates, Kopel, Cook and Ludwig, Hemenway, Donahue etc… should be)? No matter which side of the debate someone is on they should be somewhat consistent in their criticism so as to not give the appearance of double standards. This pertains to the Lott debate, so while Lott deserves to be criticized for what has stated so to should those who come up with different findings. Thus my question- Does it not follow that if there were not more guns as you argue, than the paper by Duggan “More Guns, More Crime” and the paper by Black and Nagin and even the Ayers, Donahue critique are equally invalid in saying that GUNS lead to more CRIME?

Thank you for your previous response and clearing up the Kates confusion.

4. Andrew Salemme Says:

June 9th, 2005 at 12:00 am

I think that using the 96 out 196 is a little misleading to the causal onlooker. I believe that of the gun homicides committed in the home that owned a gun the number was 75, am I mistaken?

Are you including gun homicides committed by family/roommates in homes that did not have a gun as well? I ask because the number of homicides in the home is 388 and shouldn’t the comparison be brought to gun homicides in the home and not gun homicides in both homes with guns and without them? That is should the comparison be 96 gun homicides committed by family/roommates to 388 homicides in the home if you want to imply that guns in the home are dangerous?

5. Andrew Salemme Says:
June 9th, 2005 at 12:00 am

correcting myself

In my first comment I used the number 4.6% incorrectly, by accident. It was 4.6% of all homicides in the three counties not 4.6% of all homicides used in the study, which is a big difference.

Also I have omitted a couple of words from a couple of comments which could confuse some people. In my previous comment I meant to say, “shouldn’t the comparison be brought to gun homicides in a home with a gun and not gun homicides in both homes etc…” I think that someone would be able to infer what I meant by reading the rest of the sentence and the next one, but better safe than sorry.

(I did something similar in my comment to On Sharing Data- I meant to write- “So 75 of 388 homicides in the home were committed in a home that kept a gun by a family member/roommate, (which is 19%).” Sorry if I caused any confusion.



Post a Comment

<< Home

Locations of visitors to this page