Deltoid

Kleck reckons that 97% of defensive gun users lie to the census bureau about it. Are we to suppose that 97% of the people don't believe legal guarantee of confidentiality? And yet those same people will tell a complete stranger (who may be a government agent posing as a pollster working for Kleck) about it? Come now.

John Briggs writes:

Careful, please, I cite you in another post. Are you saying that Kleck actually says this or are you interpreting his criticisms of the NCVS as compelling this conclusion about his "reckonings".

p 168 Kleck says "only about 3% of DGUs among NCVS Rs are reported to interviewers." On pp 154-6 he argues that this is because Rs are worried they might get into trouble if the authorites find out about the DGU. And yet 64.2% said that the police were aware of the incident. (Table 3) Doesn't make sense...

The fact is that the NCVS is the most comprehensive, complete, accurate crime survey ever done.

Even accepting your assertion, so what? As Kleck suggests, the NCVS is not designed to elicit information specifically about DGU's,

Oh, another point. Kleck's survey is inconsistent with NCVS measures of crime, not just of DGUs. For example, Kleck's estimate of the number of DGUs against robbery is only slightly less that the NCVS estimate for ALL robberies and attempted robberies.

although the NCVS is used (some would say, abused] for that purpose.

Used for that purpose by Kleck himself when it suits his purposes.

Yes. The question is not use but abuse.

Well he uses the NCVS to argue that guns are the most effective means of resistance. He can't have his cake and eat it. If the NCVS only finds 3% of DGUs then it is useless for measuring the effectiveness of DGU.

Where are their boundaries? Of course, the DGU surveys are also subect to abuse. Not that it matters to anyone but I have never referred to Kleck's 2.4 million number conclusive even as to the order of magnitude of DGU's. The clustering of DGU surveys at an order of magnitude greater than that derived from the NCVS survey interests me more than Klecks 2x number compared to other DGU surveys. The fact that he tried to avoid the problems that attend those sorts of surveys is no guarantee that he succeded.

I'm not sure if clustering is the right word. Most of them can't be used to produce any sort of estimate unless a lot of fudge factors are applied, and different fudge factors produce wildly different numbers. For example, Kleck has derived estimates ranging from 340,000 to 1.7M from the Hart poll. About the best you can say is that these other surveys yield estimates somewhere between 300,000 and 3 million. Over the 25 years that it has been running, the NCVS has consistently given estimates under 100,000. I don't think either the high or low number can be dismissed as an aberration --- we need some reasonable explanation as to why the NCVS can be consistently wrong, or why the other surveys can be somewhat consistently wrong.

That is what Kleck focuses on. He notes that indpendent (of each other and of the government) surveys, some by pro-gun, some by pro-control, some by neutral groups, are all wildly at variance with the NCVS. These should at least raise questions about the NCVS, just as the NCVS should raise questions about the various DGU surveys.

Yes, it should, but it doesn't prove the NCVS to be wrong as Kleck claims.

I agree. I do not regard Kleck as infallible. But, he is not the only critic of the NCVS. Let us say I am leery about the "conclusiveness" of the NCVS with respect to something it is not even intended to measure when criticisms have been leveled against it with repect to, e.g., rape and domestic violence, which it is designed to measure. I have my doubts about some of the research on both sides of this issue.

I don't believe that the NCVS is conclusive. It's not going to count DGUs that people don't want to talk about. Trouble is, Kleck's study isn't going to pick those up either.

Since I have my own pro-gun-rights biases I try to listen more carefully to the antis. On the legal/constitutional questions I am probably beyond hope -- I are a law skool graduate. I have enough experience in practice and reading of law to evaluate the arguments of judges and commentators. But, then any reasonably literate person is pretty much qualified to follow the legal arguments. At the risk of drastic oversimplification, the NRA is right, HCI doesn't know squat.

Gee, the constitutional stuff seems pretty simple to me. The state parliament has the power to control guns, the federal parliament does not. Mind you, the feds are slugging me for an extra $100 to fund the gun buy back component of the stupid new gun laws.

Do you deny that criminals have strong incentives to avoid contact with doctors who will report the gunshot wound? It is not ridiculous at all. The question is how many ...

Let's look at some numbers: Medical data indicate that about 50,000 gun-shot wounds from assaults are treated in hospitals each year. The NCVS indicates that about 80%

The accuracy of these with respect to woundings as a result of a DGU are probably lower than woundings of crime victims as the result of a crime. The criminal is less likely to check in to a hospital. The criminal is also less likely to give accurate answers should they wind up being interviewed by a government agent, even one from the Census Bureau. I think you understate the uncertainty.

of people with gun-shot wounds get hospital treatment. That suggest there are about 60,000 gun-shot wounds altogther. With roughly 10,000 homicides that means that the death rate from gun-shot wounds is around 15%. If criminals avoid medical treatment, their death rate will be somewhat higher. Kleck's survey implies that 200,000 criminals are shot each year which should produce at least 30,000 dead criminals. Where are the bodies?

Your 200,000 figure is, as you have said, 100,000 to 300,000 if all of Kleck's respondents were accurate in their assessment of the results of their shootings. You account for statistical variance due to sampling error. What I am quite willing to agree is that some of that small number who reported killing or wounding their assailants were wrong.

Almost all of them must have been wrong. Kleck's own estimate (in "Point Blank") of the number of criminals shot is 10,000 to 20,000. That means that you would only expect one or two Rs who really did shoot the perp.

You and RR say they were all lying so none of Kleck's conclusions are warranted, even as regards DGU more broadly defined and reported.

I say nothing of the sort. This particular point only shows that most of those that reported shooting the perp did not tell the truth. There are many other problems - inconstency with NCVS measures of burglaries and robberies, UCR records of these, surveys with criminals, a wildly disproportinate number of DGUs by the R as compared to the number by another household member and so on.

I say that the accuracy of a response of some DGU is much higher (because less uncertain -- respondent will unlikely be wrong about whether they drew a gun and fired) than is the specific response (I hit him, he died). Everybody, police, civilians, fighter pilots, overstates their marksmanship. But they are all more accurate in their report that they got into some sort of a fight. A five- or even ten-fold overstate of woundings/killings would not surprise me. A hundred-fold overstatement would.

If the discrepency is just because they mistakenly thought they hit when they missed, that implies the hit rate when DGUs tried to shoot the perp was around 3%. It might be safer being the target rather than a bystander...