Kellermann and X times more likely: he's baaack

[rainbowbob]

In essence, as I recall the methodology, they called up some number of people...

In fact, 4,978 households were contacted. That is considered to be a powerful statistical sample size by folks in the stats business. I believe this study is the largest and methodologically strongest study to date.

This is a link to an interview with the author of that study if you're interested:
http://www.vcdl.org/new/kleck.htm

The biggest fallacy of Kellermann's study is that it ONLY considers deaths as criteria for defensive gun use. Given that in excess of 90% of defensive gun uses do NOT include a death - or even a shot fired - this seems to me a particularly blood-thirsty requirement for the use of our guns. Kellermann isn't satisfied that we are killing enough bad guys?

 

[HK G3]

Freedom =/= Safety

In fact, Freedom usually implies increased risk. Risk which is well worth it. Unless you hate freedom and just want to be protected from yourself under penalty of death

 

[jaak]

while he tries to "save" the world from guns, who is going to save us from this idiot and his stupid opinions?

oh thats right... hes protected under the first ammendment, just like guns are under the second.

 

[RPCVYemen]

In fact, 4,978 households were contacted.That is considered to be a powerful statistical sample size by folks in the stats business.

It would seem to me that the size of the sample needed would need to be related to the frequency of the behavior . I don't now much about stats - so I may be wrong.

Does sample size need to be adjusted for the frequency of the behavior?

For example, 5000 household might be a very large sample for a political sample.

If they talked to 5,000 households and found that 1500 of them were McCain supporters and 1500 were Obama supporters, I would be less skeptical - since supporting one or the other is a relatively frequent event. I would guess that 60% of people support on or the other.

I am skeptical 1.5 million gun uses based on 50 gun uses.

Have other more recent studies replicated Kleck's results? Any of them with larger sample sizes?

Mike

 

[TallPine]

Kellermann isn't satisfied that we are killing enough bad guys?

Apparently

Maybe we need to work to change that

 

[RPCVYemen]

By the way, just goofing around on the net, I found that my intuitions about skepticism over sample size was not completely off base. The first part of the articles some of the issues that make me skeptical:

http://www.saf.org/LawReviews/Hemenway1.htm

While I have no idea if Hemenway is right or not, I think the following captures my skepticism in more precise language:

The fact that the survey is trying to estimate a low probability event also means that a small percentage bias, when extrapolated, can lead to extreme overestimates.

Note that I started asking questions on this thread not because I disagree with the estimates, but because I hoped bdickens had a better study - by that I mean one with less extrapolation.

Mike

 

[Monkeyleg]

Kleck's study is backed up by a 1994 US Bureau of Justice Statistics report which put the number of defensive uses of firearms at roughly 1.5 million.

Kellerman, on the other hand, looked only at homicides, which we know make up a small fraction of the defensive uses of firearms.

 

[Aguila Blanca]

But Scalia ignored a substantial body of public health research that contradicts his assertions. A number of scientific studies, published in the world’s most rigorous, peer-reviewed journals, show that the risks of keeping a loaded gun in the home strongly outweigh the potential benefits.

Kellerman should stick to diseases. Gunshot wounds are neither a disease nor an infection, and discussing them in terms of epidemiology is intellectually dishonest.

 

[Deanimator]

intellectually dishonest

The ONLY people I've ever seen more dishonest than anti-gunners are Holocaust deniers and not by much.

 

[Conqueror]

Kellerman's studies are the worst form of junk science. In medical school my entire class spent 2 weeks studying his second paper and punching holes in it.

First of all, he approached the studies with a foregone conclusion. A trauma surgeon at Emory, he tired of suturing GSWs and sought to prove why guns are a blight on society.

Second, his methodology is deeply flawed. As mentioned above, he only looks at bad guys KILLED instead of crimes PREVENTED. His math also contains unexplained and deeply troubling flaws. For example, his adjusted odds ratio for gun deaths in the home is HIGHER than his crude odds ratio. This is, generally speaking, impossible with a properly-conducted study.

Lastly, he draws conclusions which are not supported by his research, even if you believe his research as reliable (which is unlikely).

 

[HK G3]

Just about every public health study ever conducted is guilty of the same intellectually dishonest tactics that Kellerman uses - it makes for a more interesting (sensationalist) article. Why do you think everything both cures you and gives you cancer at the same time?