Kellermann's Responce to Criticism
Sam A. Kersh
Dr. Arthur Kellermann had a paper published in Vol 329, No. 15, October
7, 1993 issue of the NEJM. This paper brought the Journal and
Kellermann much criticism. He responded and the letter was published in
Vol 330, No. 5, Feb 3, 1994. Robert L. Ray claims "it's too long to
post." I disagree--- sooooo
The authors reply:
To the Editor:
Large-scale cohort studies are usually the best way to explore the
relation between a potential risk factor and an outcome of
interest. Since this approach was not feasible, we conducted a
case-control study. By ascertaining the rate of gun ownership in
households where a homicide had occurred and comparing this rate
with that noted in a random sample of neighboring households that
contained a person of the same age group, sex, and race as the
victim, we obtained a good approximation of relative risk (1). This
is the same research technique that was used to identify the link
between cigarette smoking and lung cancer (2).
The exaggerated claim that guns are used in self-defense more than
a million times a year has not withstood scientific scrutiny (3,4).
If a gun in the home affords substantial protection from homicide
(whether it is used to injure, kill, or frighten intruders or
simply discourage them from entering), we should have found that
homes in which a homicide occurred were less likely to contain a
gun than similar households in which a homicide did not occur. The
opposite was true.
We restricted our study to homicides in the home because the risk
or protective benefit of a readily available firearm should be most
plausibly demonstrated where it is kept. All such homicides were
included, whether or not they involved a person at high risk for
violence because of various behavioral factors. Although we noted
a degree of association among several behavioral risk factors, each
contributed independently to the risk of homicide. Therefore, we
took these effects into consideration in our final model.
A comparable ascertainment of exposure is crucial in any
case-control study, which is why we based our analysis on
interviews rather than on-the-scene reports. Ninety-three percent
of the homicides involving firearms occurred in homes where a gun
was kept, according to the proxy respondents. In 8 of the other 14
homicides, the investigating officer specifically noted that the
gun involved had been kept in the home.
Although we tried to interview a proxy respondent for each control,
this was often impossible. However, the rate of gun ownership
reported by the proxy respondents was actually higher than the rate
reported by the controls themselves.
We are confident that our findings will be corroborated. The early
studies of smoking and lung cancer were confirmed by subsequent
studies. Nonetheless, some doctors still smoke. Old habits -- and
deeply held beliefs -- die hard.
Arthur L. Kellermann, M.D., M.P.H.
Emory University
Atlanta, GA 30329
Grant Somes, Ph.D.
University of Tennessee, Memphis
Memphis, TN 38103
Frederick P. Rivara, M.D., M.P.H.
University of Washington
Seattle, WA 98195
>.................................................................................................................
Sam A. Kersh
NRA Life Member
TSRA Life Member
L.E.A.A., JPFO
=======================================================
Read the book UNINTENDED CONSEQUENCES.
Reviews are at
http://www.amazon.com/exec/obidos/ISBN%3D1888118040
or
http://www.nguworld.com/vindex/95/123195vs.htm
uspc...@my-dejanews.com
In article <35731b2c...@news.flash.net>,
csmk...@flash.net (Sam A. Kersh) wrote:
> A very misleading subject title <
I expected something along the lines of:
"boo hoo hoo, waa waa waa"
Then again, after reading Kellermann's written response, it did.
Jim
-----== Posted via Deja News, The Leader in Internet Discussion ==-----
http://www.dejanews.com/ Now offering spam-free web-based newsreading
David Veal
In article <35731b2c...@news.flash.net>,
Sam A. Kersh <csmk...@flash.net> wrote:
[Dr. Kellermann responds to criticism.]
>The authors reply:
>
>To the Editor:
>The exaggerated claim that guns are used in self-defense more than
>a million times a year has not withstood scientific scrutiny (3,4).
>If a gun in the home affords substantial protection from homicide
>(whether it is used to injure, kill, or frighten intruders or
>simply discourage them from entering), we should have found that
>homes in which a homicide occurred were less likely to contain a
>gun than similar households in which a homicide did not occur. The
>opposite was true.
This is nonsense. In order to conclude this it would be necessary
for each control household to be subject to the same danger as the
household in which the murder took place. The study didn't attempt to
pair homes of equal risk ... in fact, the study more than amply
demonstrated that homes in which someone was murdered suffer from a whole
host of danger-predicting characteristics that their control pairs didn't.
In fact, firearms could still be very effective for self-defense and
still produce exactly the situation the study found ... if the
experimental homes were more likely to be subject to a potentially lethal
attack. And given the differences in characteristics, you would be hard
pressed to argue that homes in which larger percentage of occupants used
illegal drugs and had criminal records were not exactly that.
--
David Veal ve...@utk.edu
WYSIWYG - What You See Is What You Get
WYGIWYD - What You Get Is What You Deserve
Bill B.
The exaggerated claim that guns are used in self-defense more than
a million times a year has not withstood scientific scrutiny (3,4).
If a gun in the home affords substantial protection from homicide
(whether it is used to injure, kill, or frighten intruders or
simply discourage them from entering), we should have found that
homes in which a homicide occurred were less likely to contain a
gun than similar households in which a homicide did not occur. The
opposite was true.
>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>
And homes in which homicides occurred also had toilets. Hence toilets kill!
RAY
David Veal wrote:
Kellermann wrote:
> >To the Editor:
> [...]
> >The exaggerated claim that guns are used in self-defense more than
> >a million times a year has not withstood scientific scrutiny (3,4).
> >If a gun in the home affords substantial protection from homicide
> >(whether it is used to injure, kill, or frighten intruders or
> >simply discourage them from entering), we should have found that
> >homes in which a homicide occurred were less likely to contain a
> >gun than similar households in which a homicide did not occur. The
> >opposite was true.
>
DV> This is nonsense. In order to conclude this it would
be necessary
> for each control household to be subject to the same danger as the
> household in which the murder took place. The study didn't attempt to
> pair homes of equal risk ... in fact, the study more than amply
> demonstrated that homes in which someone was murdered suffer from a whole
> host of danger-predicting characteristics that their control pairs didn't.
RR: David, you forget that Kellermann then factored out
these risks for homicide, and STILL came up with a 2.7 times
added risk with firearms in the home, even with the risks
of both cases and controls evened out.
RAY
Bill B. wrote:
>
> The exaggerated claim that guns are used in self-defense more than
> a million times a year has not withstood scientific scrutiny (3,4).
> If a gun in the home affords substantial protection from homicide
> (whether it is used to injure, kill, or frighten intruders or
> simply discourage them from entering), we should have found that
> homes in which a homicide occurred were less likely to contain a
> gun than similar households in which a homicide did not occur. The
> opposite was true.
>
> >>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>
>
Hence toilets kill!
RR: However, after a final double-blind multi-factorial
deep-regression analysis, Professor Alphonse HighBottom has
concluded that there is a 6.34422 chance that homes
containing guns are more likely to have an outhouse than an
'inhouse' :-)
David Friedman
> In fact, firearms could still be very effective for self-defense and
> still produce exactly the situation the study found ... if the
> experimental homes were more likely to be subject to a potentially lethal
> attack. And given the differences in characteristics, you would be hard
> pressed to argue that homes in which larger percentage of occupants used
> illegal drugs and had criminal records were not exactly that.
I think the argument against the study is a little stronger then that. The
fact that someone has a gun in his home is in itself evidence that he is
at risk--not because he is at risk from his gun, but because being at risk
is one of the reasons people have guns.
Someone else writes (responding to David Veal, not to me)
> RR: David, you forget that Kellermann then factored out
> these risks for homicide, and STILL came up with a 2.7 times
> added risk with firearms in the home, even with the risks
> of both cases and controls evened out.
I haven't seen the study, but the methodological problem is that he can
only factor out those risks he has data on. If being at risk causes gun
ownership, which seems plausible, then gun ownership will correlate with
the unobservable risks as well--and they will correlate with being a gun
victim.
--
David Friedman
http://www.best.com/~ddfr/
David Friedman
After reading the thread on Kellerman, I did a bit of web browsing. Steve
Kangas has a good summary of the Kellerman paper at:
http://www.scruz.net/~kangaroo/L-kellermann.htm
The one crucial thing he does not say, and that is asserted in a piece
from _Reason_ elsewhere on the web, is that Kellerman has refused to make
his data available to other researchers. If that is true, then his study
cannot be trusted, since he can say anything he likes about the results,
true or false, without the risk that someone else will rerun his
regressions and prove them false.
I note, incidentally, that John Lott and Isaac Ehrlich, two economists who
did (widely separated) statistical studies heavily criticized (by
opponents of handgun ownership and the death penalty, respectively) both
made their data available to others.
Steve Hix
In article <ddfr-03069...@129.210.78.3>, dd...@best.com (David
Friedman) wrote:
> > RR: David, you forget that Kellermann then factored out
> > these risks for homicide, and STILL came up with a 2.7 times
> > added risk with firearms in the home, even with the risks
> > of both cases and controls evened out.
>
> I haven't seen the study, but the methodological problem is that he can
> only factor out those risks he has data on. If being at risk causes gun
> ownership, which seems plausible, then gun ownership will correlate with
> the unobservable risks as well--and they will correlate with being a gun
> victim.
What RR neglected to bring out from the study in question was
that Kellerman found even *higher* risks associated with
- Renting, rather than buying your home.
- Using any illegal drugs.
and a couple of other factors.
He also neglected to note that Kellerman counted homicides that
either didn't involve any firearm (but the victim either owned
one or resided in a home where one was owned), or the firearm
owned was not connected with the homicide.
The study, as an argument for restricting legal gun ownership,
is severely lacking.
RAY
David Friedman wrote:
> I haven't seen the study, but the methodological problem is that he can
> only factor out those risks he has data on. If being at risk causes gun
> ownership, which seems plausible, then gun ownership will correlate with
> the unobservable risks as well--and they will correlate with being a gun
> victim.
RR: What 'unobservable risks' are you talking about? As I
said, Kellermann already has factored out the societal
risks. The only remaining homicide risk might be a personal
threat from specific individuals. Kellermann also takes
that into account with the observation that if these
homicide victims bought guns to protect themselves from
these personal threats, then the purchase of guns for this
self-defense purpose was in itself self-defeating.
Afterall, they were MURDERED!
BTW, you can locate Arthur Kellermann's study at your
local library. Just ask for the New England Journal of
Medicine dated 8/7/93, page 1084.
RAY
David Friedman wrote:
>
> After reading the thread on Kellerman, I did a bit of web browsing. Steve
> Kangas has a good summary of the Kellerman paper at:
> from _Reason_ elsewhere on the web, is that Kellerman has refused to make
> his data available to other researchers. If that is true, then his study
> cannot be trusted, since he can say anything he likes about the results,
> true or false, without the risk that someone else will rerun his
> regressions and prove them false.
RR: Actually, it is NOT true. Kellermann has provided his
data to the Inter-university Consortium for Political and
Social Research at Ann Arbor, Michigan. It is available to
the public.
John R. Lott, Jr.
RAY wrote:
Not all Kellermann's data has yet to be released, and he waited over five
years to release what he did. Recent work by Gary Kleck using what has been
released confirms that at absolute most 4 percent of the homicides could be
attributed to the gun owned by those in the residence. Kellermann's study
incorrectly assumed that all the gun deaths were caused by the gun in the
home. Obviously, this assumption is crucial for his claim. At least 96
percent of his deaths were falsely assigned.
Another big problem with Kellermann's work is that he uses a statistical
method (the case control method) which was designed for a completely different
type of problem. Gun ownership was not randomly assigned in the same way that
a drug study for example assigns drugs to people. Many people may have bought
guns because they were at particularly high risk of being attacked.
Finally, unlike Kellermann's studies, my earlier study and my new book did not
focus on data from only one or three cities for only one year. I studied data
for all the counties in the entire United States over many years. I studied
murder rates, accidental gun deaths, and suicides. There is in fact
absolutely no evidence that any of Kellermann's claims are correct. See More
Guns, Less Crime (University of Chicago Press, 1998).
Steve Fischer
In article <35752E4F...@Interaccess.com> RAY <Ki...@Interaccess.com> writes:
>David Veal wrote:
>
>Kellermann wrote:
>> >To the Editor:
>> [...]
>> >a million times a year has not withstood scientific scrutiny (3,4).
>> >If a gun in the home affords substantial protection from homicide
>> >(whether it is used to injure, kill, or frighten intruders or
>> >simply discourage them from entering), we should have found that
>> >homes in which a homicide occurred were less likely to contain a
>> >gun than similar households in which a homicide did not occur. The
>> >opposite was true.
>>
>be necessary
>> for each control household to be subject to the same danger as the
>> household in which the murder took place. The study didn't attempt to
>> pair homes of equal risk ... in fact, the study more than amply
>> demonstrated that homes in which someone was murdered suffer from a whole
>> host of danger-predicting characteristics that their control pairs didn't.
>
>these risks for homicide, and STILL came up with a 2.7 times
>added risk with firearms in the home, even with the risks
>of both cases and controls evened out.
... and it is the way he "factored out" that is in question.
--
/Steve D. Fischer/Atlanta, Georgia/str...@netcom.com/
Steve Fischer
In article <35761E87...@Interaccess.com> RAY <Ki...@Interaccess.com> writes:
>David Friedman wrote:
>
>> I haven't seen the study, but the methodological problem is that he can
>> only factor out those risks he has data on. If being at risk causes gun
>> ownership, which seems plausible, then gun ownership will correlate with
>> the unobservable risks as well--and they will correlate with being a gun
>> victim.
>
>RR: What 'unobservable risks' are you talking about? As I
>said, Kellermann already has factored out the societal
>risks.
Exactly how did he do that? Be specific. Use numbers.
Let's see, the societal risk = 2.537584, so that means the
retarded monkey typing the great American novel will only
be enticed into eating a banana if I wear a dress and rub
my stomach while singing the French national anthem. Hence,
more guns = more murders.
> BTW, you can locate Arthur Kellermann's study at your
>local library. Just ask for the New England Journal of
>Medicine dated 8/7/93, page 1084.
Been there, done that.
Dan Z
In <35761F7B...@Interaccess.com> RAY <Ki...@Interaccess.com>
writes:
>David Friedman wrote:
>>
>> After reading the thread on Kellerman, I did a bit of web browsing.
Steve
>> Kangas has a good summary of the Kellerman paper at:
>> The one crucial thing he does not say, and that is asserted in a
piece
>> from _Reason_ elsewhere on the web, is that Kellerman has refused to
make
>> his data available to other researchers. If that is true, then his
study
>> cannot be trusted, since he can say anything he likes about the
results,
>> true or false, without the risk that someone else will rerun his
>> regressions and prove them false.
>
>RR: Actually, it is NOT true. Kellermann has provided his
>data to the Inter-university Consortium for Political and
>Social Research at Ann Arbor, Michigan. It is available to
>the public.
As has been pointed out to Ray before (sigh), AFTER several years, AND
"cleaning up" of the data.
--
antispam address list, spammers listed will be added to other spammer's lists; poetic justice....
kle...@optsys.com , wrzi...@optsys.com , weigh...@giselp.com , gis...@hotmail.com , avra...@ragis.com ,
Global...@hotmail.com , knic...@internetman.com , o...@online-group.com , peeka...@juno.com ,
non...@juno.com , da...@linkus.com , cei...@mailexcite.com , neta...@ml.org , m...@ml.org , kki...@vvi.net ,
spid...@pompano.net , ab...@myownmail.com , 34...@pcpostal.com , ja...@post.omnitel.net , h...@clark.net ,
netcom...@ns1.gonepostal.com , bonvoya...@worldnet.att.net , force...@msn.com , mar...@aware.nl ,
eth...@spiritnetwork.com , lor...@global-homebiz.com , web...@azstarnet.com , play...@xoom.com ,
webm...@globaltekgrp.com , ad...@sysna.com , the...@biznessweb.net , dal...@hoststar.com ,
da...@linkus.com , sa...@123bizweb.com , webm...@123bizweb.com , re...@123bizweb.com , t...@wolsi.com ,
sysa...@titaniccollectibles.com , sun...@safari.net , ad...@pornacopia.com , sig...@mwci.net ,
rm0...@replinets-max.com , ever...@TheOffice.net , labl...@wekan.com , s...@ras.nexgate.net ,
dan...@aol.com , da...@linkus.com , webm...@nucredit.com , dom...@website-hostings.com , st...@woz.org ,
st...@woz.org , apry...@usa.net , hoo...@xoom.com , sco...@hothotmail.com , micr...@doctorpc.com ,
ke...@hudsonet.com , va...@clubvault.com ,
Dan Z
In <35761E87...@Interaccess.com> RAY <Ki...@Interaccess.com>
writes:
>David Friedman wrote:
>
>> I haven't seen the study, but the methodological problem is that he
can
>> only factor out those risks he has data on. If being at risk causes
gun
>> ownership, which seems plausible, then gun ownership will correlate
with
>> the unobservable risks as well--and they will correlate with being a
gun
>> victim.
>
>RR: What 'unobservable risks' are you talking about? As I
>said, Kellermann already has factored out the societal
>threat from specific individuals. Kellermann also takes
>that into account with the observation that if these
>homicide victims bought guns to protect themselves from
>these personal threats, then the purchase of guns for this
>self-defense purpose was in itself self-defeating.
>Afterall, they were MURDERED!
>local library. Just ask for the New England Journal of
>Medicine dated 8/7/93, page 1084.
Third, it is not surprising that Kellerman and Reay's analysis shows
that homicide victims are armed in disproportionate numbers, for it
appears that a large and growing proportion of homicide victims are
criminals themselves. [74] In other words, by focusing his analysis on
households in which homicides occurred, Kellerman and Reay may be
finding out information only about the characteristics of homicide
victims, and people who commit murder, without any assurance that such
information may be safely generalized to the gun-owning public at
large. It is possible that the households in which homicides occur are
far from representative of typical or average households in which guns
are present. If so, treating the 43 times statistic as though it were a
universal law applicable to all gun owners, rather than as descriptive
of a discrete, aberrant subset, is simply wrong and misleading.
[74]. See Daniel D. Polsby, "Firearms Costs, Firearms Benefits and
the Limits of Knowledge," Journal of Law and Criminology 86
(1995): 211.
(From John R. Lott, Jr)
James F. Mayer
In <35761F7B...@Interaccess.com> RAY <Ki...@Interaccess.com>
writes:
>David Friedman wrote:
>>
>> After reading the thread on Kellerman, I did a bit of web browsing.
Steve
>> Kangas has a good summary of the Kellerman paper at:
>> The one crucial thing he does not say, and that is asserted in a
piece
>> from _Reason_ elsewhere on the web, is that Kellerman has refused to
make
>> his data available to other researchers. If that is true, then his
study
>> cannot be trusted, since he can say anything he likes about the
results,
>> true or false, without the risk that someone else will rerun his
>> regressions and prove them false.
>
>RR: Actually, it is NOT true. Kellermann has provided his
>data to the Inter-university Consortium for Political and
>Social Research at Ann Arbor, Michigan. It is available to
>the public.
Only after being badgered by congress to do so and after it have
been "cleaned up".
David Friedman
In article <35761E87...@Interaccess.com>, RAY <Ki...@Interaccess.com>
wrote:
> RR: What 'unobservable risks' are you talking about?
Any factors that both make people more likely to own guns and more likely
to be murdered.
>As I
> said, Kellermann already has factored out the societal
> risks.
Insofar as he has data on them.
>The only remaining homicide risk might be a personal
> threat from specific individuals. Kellermann also takes
> that into account with the observation that if these
> homicide victims bought guns to protect themselves from
> these personal threats, then the purchase of guns for this
> self-defense purpose was in itself self-defeating.
> Afterall, they were MURDERED!
If he makes that argument, he is obviously missing the point.
Consider the striking correlation between visiting a hospital and dying.
The "obvious" explanation is that going to a hospital increases your
chance of dying. If someone pointed out that the reason for the
correlation was that when people were seriously ill they went to a
hospital in the hope of being cured, would it be an adequate response to
say that going to the hospital was obviously self-defeating, since they
died?
Steve Kangas, incidentally, has a good (sympathetic) summary of the
Kellerman paper webbed on his site, from which it sounds as though
Kellerman did everything he reasonably could to control for the sort of
problems we are discussing--although everything might still not be enough.
The one critical fact that Steve omits is that (at least according to a
_Reason_ article I found online that discusses Kellerman's work) Kellerman
has refused to make his data available to other researchers. If that is
true, it undercuts all of his claims about his research, since it means
that he has made it impossible for other people to find out whether or not
they are true.
Does anyone know whether it is true that he has kept his data secret, and
if so whether he has offered any explanation for doing so?
David Friedman
In article <35761F7B...@Interaccess.com>, RAY <Ki...@Interaccess.com>
wrote:
> David Friedman wrote:
> >
> > After reading the thread on Kellerman, I did a bit of web browsing. Steve
> > Kangas has a good summary of the Kellerman paper at:
> > The one crucial thing he does not say, and that is asserted in a piece
> > from _Reason_ elsewhere on the web, is that Kellerman has refused to make
> > his data available to other researchers. If that is true, then his study
> > cannot be trusted, since he can say anything he likes about the results,
> > true or false, without the risk that someone else will rerun his
> > regressions and prove them false.
>
> RR: Actually, it is NOT true. Kellermann has provided his
> data to the Inter-university Consortium for Political and
> Social Research at Ann Arbor, Michigan. It is available to
> the public.
Can one get at it online? Do you know if anyone, in particular anyone on
the other side, has reanalyed it (as Black and Nagin recently did in the
JLS for Lott's data) and published the results?
How do you know it is available, by the way? I ask, not to question your
honesty, but because one of the virtues of the net is that it moves us
closer to a situation where everyone can check out facts for himself at
fairly low cost. A lot of political argument consists of one siding
asserting "A," the other asserting "not A," and no easy way for spectators
to tell who is telling the truth.
Incidentally, I apologize for reposting my "fact" about the data not being
available just before reading your post.
Bill Bailey
John R. Lott, Jr. <john...@law.uchicago.edu> wrote in article
<3576BE82...@law.uchicago.edu>...
>
>
> RAY wrote:
>
> > David Friedman wrote:
> > >
> > > After reading the thread on Kellerman, I did a bit of web browsing.
Steve
> > > Kangas has a good summary of the Kellerman paper at:
> > > The one crucial thing he does not say, and that is asserted in a
piece
> > > from _Reason_ elsewhere on the web, is that Kellerman has refused to
make
> > > his data available to other researchers. If that is true, then his
study
> > > cannot be trusted, since he can say anything he likes about the
results,
> > > true or false, without the risk that someone else will rerun his
> > > regressions and prove them false.
> >
> > RR: Actually, it is NOT true. Kellermann has provided his
> > data to the Inter-university Consortium for Political and
> > Social Research at Ann Arbor, Michigan. It is available to
> > the public.
>
Kellermann's research model (as medical models typically do) excludes
rational choice factors. (Yes, Ray, there are unobserved factors in
Kellerman's research!) Similarly, the earlier social models of firearms
research--which were very derivative of physical science models--excluded
factors of rational choice. They counted guns and counted crimes (somewhat
selectively) and declared more guns correlated with more crime. There was
not much difference between that model and the medical contagion model.
You caught crime from guns. The idea has found sybolic representation in a
number of movies--the "evil" gun that goes from hand to hand causing
murders and misery, and no person has sufficient strength of character to
stop it. The idea that more crime (or at least more, the media artifact of
more crime information) might cause more people in a given region to
purchase more firearms wasn't even a consideration. From Zimring onward, I
think that "mechanistic model" was pretty much the standard model for gun
research.
Are we seeing a shift toward rational choice models in firearms research?
I think of James D. Wright and colleagues as being some of the first (I may
be wrong on that point), who began by examining rational choices made by
criminals amid an armed citizenry. More recently, we have Gary Kleck and
John Lott's research. Dare we hope for more?
Bill Bailey
jackson dryden
--
jackson
I don't know about everybody else, but I would love to see Arthur Kellerman
come and debate you here. Oracle, at least, would be chagrined at the
prospect that his favorite "expert" would expose himself to the possibility
of being undone in honest, intellectual combat. If that happened, Oracle
would, himself, be undone.
Hope to see your posts on a regular basis.
jackson
>
>
RAY
RR: I must say, I'm glad to see one of the academic players
take a roll in tpg. Lott first sent this to my e-mail.
This is how I responded:
John Lott wrote:
> > RR: Actually, it is NOT true. Kellermann has provided his
> > data to the Inter-university Consortium for Political and
> > Social Research at Ann Arbor, Michigan. It is available to
> > the public.
>
JL> Not all Kellermann's data has yet to be released, and
> he waited over five years to release what he did.
> Recent work by Gary Kleck using what has been released
> confirms that at absolute most 4 percent of the homicides
> could be attributed to the gun owned by those in the
> residence. Kellermann's study incorrectly assumed
> that all the gun deaths were caused by the gun in the
> home. Obviously, this assumption is crucial for his
> claim. At least 96> percent of his deaths were falsely
> assigned.
RR: Thank you for your response. I had seen you make this
claim in the Chicago Tribune (although I did not see any
attribution to Kleck there). Perhaps you can tell me what
data source this 96% figure comes from.
As I've posted in TPG, the only source I have found
which comes close is a letter Kellermann wrote in the NEJM
responding to his critics, where he said police had noted
that 8 of 14 cases of homicide were committed with a gun
that 'had been' in the home. If this is the only source
that Kleck or you have, I would suggest that your
interpretation is speculative, to say the least.
I have e-mailed Kellermann with your Tribune paragraph
(was it Myth #5?), and he says your contention is a
distortion.
JL> Another big problem with Kellermann's work is that he
> uses a statistical method (the case control method)
> which was designed for a completely different type of
> problem. Gun ownership was not randomly assigned in the
> same way that a drug study for example assigns drugs to
> people. Many people may have bought guns because they
> were at particularly high risk of being attacked.
RR: However, if Kellermann has correctly factored out
societal risks of homicide, then IMM, the remaining
correlation with guns (in this case 2.7X) might still be
considered causal. The only risks of attack that might
remain would be a personal attack threat... something so
individualized that it could not be part of the overall
analysis. And Kellermann commented on this potential
drawback in the study, saying that if Joe Blow bought a gun
because Sam Smith had threatened to kill him, then quite
obviously the gun had failed as a self-defense tool since
Sam Smith accomplished the deed.
JL> Finally, unlike Kellermann's studies, my earlier study
> and my new book did not focus on data from only one or
> three cities for only one year. I studied data
> for all the counties in the entire United States over
> many years. I studied murder rates, accidental gun
> deaths, and suicides. There is in fact absolutely no
> evidence that any of Kellermann's claims are correct.
> See More Guns, Less Crime (University of Chicago Press, 1998).
RR: I hope to soon. I've read your earlier study, and look
forward to seeing your book.
Finally, as a layman in this debate, I'd sure
appreciate it if you would comment on the absolute rancor
that's surfaced in recent years from academics on both sides
of the issue. While I'm quite use to seeing the flames
erupt in tpg, I personally find it disheartening when MDs
and Phds go toe to toe in allegedly prestigious Law Reviews
and Professional Journals.
I'm sure Gun Control advocates are not immune from this
disease, but I really thought that the Kates/Schaeffer
article in the Tennessee Law Review hit a new low in
rhetoric and name calling, without adding much of
substance.
Perhaps your guys need to communicate more among each
other.. have some multi-disciplinary round-tables and even
go out for a few drinks together. I wonder, did you even
attempt to talk with Kellermann to get his meaning before
you wrote your Myth #5?
David Friedman
In article <357739FA...@Interaccess.com>, RAY <Ki...@Interaccess.com>
wrote:
>RR: However, if Kellermann has correctly factored out
>societal risks of homicide, then IMM, the remaining
>correlation with guns (in this case 2.7X) might still be
>considered causal. The only risks of attack that might
>remain would be a personal attack threat... something so
>individualized that it could not be part of the overall
>analysis. And Kellermann commented on this potential
>drawback in the study, saying that if Joe Blow bought a gun
>because Sam Smith had threatened to kill him, then quite
>obviously the gun had failed as a self-defense tool since
>Sam Smith accomplished the deed.
As I just pointed out in this thread, that is a very poor argument--if
Kellerman really made it, that is one bit of evidence that he is a
demagogue rather than a scientist (although, of course, even scientists
are not immune to the temptation to occasional demagoguery). Does anyone
have the quote? Did he imply that he was answering the argument, or merely
commenting on the (obvous) fact that self-defense does not always work.
Consider two people, otherwise similar, save that A has been threatened
with murder by someone and B has been not. A, not surprisingly, obtains a
gun. Which of them is more likely to be a murder victim?
Again, consider two people, otherwise similar, one of whom is a
professional criminal and the other not. Which is more likely to own a
gun? To be a murder victim?
> Finally, as a layman in this debate, I'd sure
>appreciate it if you would comment on the absolute rancor
>that's surfaced in recent years from academics on both sides
>of the issue. While I'm quite use to seeing the flames
>erupt in tpg, I personally find it disheartening when MDs
>and Phds go toe to toe in allegedly prestigious Law Reviews
>and Professional Journals.
1. I think you should distinguish between professional scholars, such as
Black and Nagin (the authors of the only scholarly criticism of the Lott
and Mustard paper that I have seen), and professional political activists
such as Stephen Terek. The latter have a clear personal stake in
denigrating studies that produce results they don't like, and suffer
virtually no professional loss from making bad, or even dishonest,
arguments. I think the Black and Nagin article is in large part
unconvincing, and it might be dishonest in the very weak sense of
attempting to make their case look better than it is, but it isn't
particularly rancorous.
2. I stumbled into the aftermath of a previous dispute of a similar sort
some years back, and I think the story is instructive. As you may know,
Isaac Ehrlich did some very controversial research on the deterrent effect
of the death penalty, and was criticized in much the same fashion that
John Lott has been.
Some years ago, I had a colleague, a law professor, who is a very
prominent figure in the criminalogical community. He is also, in my
judgement, an unusually gentle and reasonable man. I happened to mention
Ehrlich's work, and the response was just short of a diatribe. He claimed
that the work had been refuted, that Ehrlich was dishonest, that he had
refused to provide his data to other researchers, ... .I asked for
evidence, and he lent me a volume of articles on deterrence edited by some
friends of his.
I read the volume, and it provided no support at all for any of his
claims--on the contrary, the one article that dealt with one of Ehrlich's
articles was by people who had been given Ehrlich's data by Ehrlich and
had replicated his results. The summary article on studies of deterrence
(not the death penalty specifically) found, as I recall, that out of
thirteen such studies twelve had come up with a positive result. The only
stuff in the book that was critical of deterrence, as far as I remember,
was the introduction, written by the editors--and it consisted essentially
of arguing that all of the evidence for deterrence could conceivably be
explained away in one way or another. While that was true--statistical
evidence can practically always be explained away if you are sufficiently
ingenious--the editors offered no tests at all of their conjectural
explanations.
I pointed all of this out to my colleague, and his response was apologetic
(as I say, he is an unusually reasonable man). Basically what he said was
that Ehrlich's work had been used for evil purposes (to argue for the
death penalty, which my colleague opposed), that he was sure its
conclusion was wrong (not on statistical grounds--I don't think he knows
much about statistics--but from dealing with criminals), and he thus had
strong hostile feelings towards Ehrlich. Pretty clearly, his specific
charges were things he had gotten from other people who shared his biases,
and had never bothered to check.
--
David Friedman
DD...@Best.com
http://www.best.com/~ddfr/
Tim Lambert
dd...@best.com (David Friedman) writes:
> In article <35761E87...@Interaccess.com>, RAY <Ki...@Interaccess.com>
> wrote:
>
> > RR: What 'unobservable risks' are you talking about?
>
> Any factors that both make people more likely to own guns and more likely
> to be murdered.
>
> >As I
> > said, Kellermann already has factored out the societal
> > risks.
>
> Insofar as he has data on them.
>
> >The only remaining homicide risk might be a personal
> > threat from specific individuals. Kellermann also takes
> > that into account with the observation that if these
> > homicide victims bought guns to protect themselves from
> > these personal threats, then the purchase of guns for this
> > self-defense purpose was in itself self-defeating.
> > Afterall, they were MURDERED!
>
> If he makes that argument, he is obviously missing the point.
>
> Consider the striking correlation between visiting a hospital and dying.
> The "obvious" explanation is that going to a hospital increases your
> chance of dying. If someone pointed out that the reason for the
> correlation was that when people were seriously ill they went to a
> hospital in the hope of being cured, would it be an adequate response to
> say that going to the hospital was obviously self-defeating, since they
> died?
Ray has misstated Kellermann's argument. He merely points out that in
those cases the gun failed as protection.
> Steve Kangas, incidentally, has a good (sympathetic) summary of the
> Kellerman paper webbed on his site, from which it sounds as though
> Kellerman did everything he reasonably could to control for the sort of
> problems we are discussing--although everything might still not be enough.
> The one critical fact that Steve omits is that (at least according to a
> _Reason_ article I found online that discusses Kellerman's work) Kellerman
> has refused to make his data available to other researchers. If that is
> true, it undercuts all of his claims about his research, since it means
> that he has made it impossible for other people to find out whether or not
> they are true.
>
> Does anyone know whether it is true that he has kept his data secret, and
> if so whether he has offered any explanation for doing so?
The data is publicly available from the ICPSR
(http://www.icpsr.umich.edu/) study #6898.
James F. Mayer
In <tn1zt4e...@oolong.orchestra.cse.unsw.EDU.AU> Tim Lambert
According to Mr. John Lott, not all of it.
Tim Lambert
"John R. Lott, Jr." <john...@law.uchicago.edu> writes:
> RAY wrote:
>
> > David Friedman wrote:
> > >
> > > After reading the thread on Kellerman, I did a bit of web browsing. Steve
> > > Kangas has a good summary of the Kellerman paper at:
> > > The one crucial thing he does not say, and that is asserted in a piece
> > > from _Reason_ elsewhere on the web, is that Kellerman has refused to make
> > > his data available to other researchers. If that is true, then his study
> > > cannot be trusted, since he can say anything he likes about the results,
> > > true or false, without the risk that someone else will rerun his
> > > regressions and prove them false.
> >
> > RR: Actually, it is NOT true. Kellermann has provided his
> > data to the Inter-university Consortium for Political and
> > Social Research at Ann Arbor, Michigan. It is available to
> > the public.
>
> Not all Kellermann's data has yet to be released, and he waited over five
> years to release what he did.
Hmmm, it's been over five years since Kleck did his survey on
defensive gun use. When is he going to release the data?
> Recent work by Gary Kleck using what has been
> released confirms that at absolute most 4 percent of the homicides could be
> attributed to the gun owned by those in the residence. Kellermann's study
> incorrectly assumed that all the gun deaths were caused by the gun in the
> home. Obviously, this assumption is crucial for his claim. At least 96
> percent of his deaths were falsely assigned.
This is wrong from beginning to end. As far as I can tell, this
"work" by Gary Kleck amounts to misreading a statement by Kellermann
that that there only 14 gun homicides in homes where no guns were kept
according to the proxy and in 8 of those cases the police noted that
the gun involved had been kept in the home (i.e. the proxy was
mistaken in those cases.) Kleck seems to have misunderstood this as
meaning that in only 8 out of the 209 gun homicides did the police
state that the gun involved was kept in the home.
It's not hard to estimate an upper bound on the number of homicides
committed with guns brought from outside. Six out of the 214 cases
where no gun was kept were committed with guns brought from outside.
It seems reasonable to suppose that the proportion would be similar
for the other 174 cases, giving us a total of at most 11 homicides
committed with outside guns. This is an upper bound because we have
assumed that all the guns where the police could not decide whether or
not the gun was brought from the outside were, in fact, from the
outside.
To summarize: at least 95 percent of the homicides can be attributed
to the gun owned by those in the residence. Kleck's "at absolute most 4
percent" is wildly incorrect.
> Another big problem with Kellermann's work is that he uses a statistical
> method (the case control method) which was designed for a completely different
> type of problem.
Wrong. I suggest you read Scelesselman's book on case control studies.
> Gun ownership was not randomly assigned in the same way that
> a drug study for example assigns drugs to people.
A drug study that randomly assigns drugs to people is not a case
control study.
> Many people may have bought
> guns because they were at particularly high risk of being attacked.
>
> Finally, unlike Kellermann's studies, my earlier study and my new book did not
> focus on data from only one or three cities for only one year.
Your study, however, is correlational just like Kellermann's. States
were not randomly assigned concealed carry laws.
> There is in fact
> absolutely no evidence that any of Kellermann's claims are correct.
Welcome to talk.politics.guns, Dr Lott. I see you are getting into
the swing of things. Overblown rhetoric is what we do best here.
Tim
Jim McCulloch
>Another big problem with Kellermann's work is that he uses a statistical
>method (the case control method) which was designed for a completely different
>type of problem. Gun ownership was not randomly assigned in the same way that
>a drug study for example assigns drugs to people. Many people may have bought
>guns because they were at particularly high risk of being attacked.
This seems like a common-sense statement, but do we really know this
to be true? Here it Texas, the last time I checked, one class of gun
owners, at least, the people who were getting permits to carry
handguns, were overwhelmingly middle-aged middle-class white males,
who statistically are not at particularly high risk being attacked.
That is not to say they do not *believe* themselves at high risk ;-).
In any case, you can very well say the people in Kellermann's sample
*may* have bought guns because they were in danger, but do you
actually know what percentage of the gun owners bought guns because
they were, objectively, "at high risk"? Seems like a hard thing to
prove, to me. You certainly can't use the occaional homicide in the
sample as proof, because the cause of that is what is in
question--that reasoning would not only be circular, but very tightly
circular. You would need some sort of auxiliary evidence, like, say,
prior police reports of threats to the gun owners, wouldn't you? Do
you have any actual evidence for your statement?
--Jim McCulloch
Dan Z
In <ddfr-04069...@129.210.78.3> dd...@best.com (David Friedman)
writes:
>The one critical fact that Steve omits is that (at least according to
a
>_Reason_ article I found online that discusses Kellerman's work)
>has refused to make his data available to other researchers. If that
is
means
>that he has made it impossible for other people to find out whether or
not
>they are true.
>
>Does anyone know whether it is true that he has kept his data secret,
and
>if so whether he has offered any explanation for doing so?
>
From John Lott, Jr's post of yesterday:
"Not all Kellermann's data has yet to be released, and he waited over
five years to release what he did."
--
antispam address list, spammers listed will be added to other spammer's lists; poetic justice....
Global...@hotmail.com , knic...@internetman.com , o...@online-group.com , peeka...@juno.com ,
non...@juno.com , da...@linkus.com , cei...@mailexcite.com , neta...@ml.org , m...@ml.org , kki...@vvi.net ,
spid...@pompano.net , ab...@myownmail.com , 34...@pcpostal.com , ja...@post.omnitel.net , h...@clark.net ,
netcom...@ns1.gonepostal.com , bonvoya...@worldnet.att.net , force...@msn.com , mar...@aware.nl ,
eth...@spiritnetwork.com , lor...@global-homebiz.com , web...@azstarnet.com , play...@xoom.com ,
webm...@globaltekgrp.com , ad...@sysna.com , the...@biznessweb.net , dal...@hoststar.com ,
da...@linkus.com , sa...@123bizweb.com , webm...@123bizweb.com , re...@123bizweb.com , t...@wolsi.com ,
sysa...@titaniccollectibles.com , sun...@safari.net , ad...@pornacopia.com , sig...@mwci.net ,
rm0...@replinets-max.com , ever...@TheOffice.net , labl...@wekan.com , s...@ras.nexgate.net ,
dan...@aol.com , da...@linkus.com , webm...@nucredit.com , dom...@website-hostings.com , st...@woz.org ,
st...@woz.org , apry...@usa.net , hoo...@xoom.com , sco...@hothotmail.com , micr...@doctorpc.com ,
ke...@hudsonet.com , va...@clubvault.com , gw...@yahoo.com , rich...@webt.com , mic...@netparty.com ,
cha...@aol.com , mike...@aol.com , remo...@aol.com , nappm...@yahoo.com ,
Dan Z
being
>available just before reading your post.
>
>--
>David Friedman
>http://www.best.com/~ddfr/
From John Lott Jr's post of yesterday:
"Not all Kellermann's data has yet to be released, and he waited over
five years to release what he did."
I suspect John Lott is a much more credible source than Ray.
John R. Lott, Jr.
>John Lott wrote:
>> > RR: Actually, it is NOT true. Kellermann has provided his
>> > data to the Inter-university Consortium for Political and
>> > Social Research at Ann Arbor, Michigan. It is available to
>> > the public.
>>
>waited over five
>> years to release what he did. Recent work by Gary Kleck using what
has been
>> released confirms that at absolute most 4 percent of the homicides
could be
>> attributed to the gun owned by those in the residence. Kellermann's
study
>> incorrectly assumed that all the gun deaths were caused by the gun in
the
>> home. Obviously, this assumption is crucial for his claim. At least
96
>> percent of his deaths were falsely assigned.
>
>claim in the Chicago Tribune (although I did not see any
>attribution to Kleck there). Perhaps you can tell me what
>data source this 96% figure comes from.
> As I've posted in TPG, the only source I have found
>which comes close is a letter Kellermann wrote in the NEJM
>responding to his critics, where he said police had noted
>that 8 of 14 cases of homicide were committed with a gun
>that 'had been' in the home. If this is the only source
>that Kleck or you have, I would suggest that your
>interpretation is speculative, to say the least.
> I have e-mailed Kellermann with your Tribune paragraph
>(was it Myth #5?), and he says your contention is a
>distortion.
Gary Kleck has looked into this quite closely and should be able to
provide you information on this if you write him at Florida State. The
reference in th NEJM is not the source for this 4 percent, it is from
what data that Kellermann has released. Kellermann refused to release
any data to me while I was still working on the book so that the only
information that I had was from the difficult to interpret response that
Kellermann had in the NEJM. When I called Kellermann and coauthors to
ask them to ellaborate on the quote they refused to do so. It was the
only information that was available, and as I say I tried talking to
them about it when I wrote that section of my book.
As to Kellermann saying that I distorted his point, he should have been
willing to talk to me about it when I called him. I would also suggest
that he specifically state what the true correct number is and give us
all the hard data so that we can evaluate his claim.
>
>JL> Another big problem with Kellermann's work is that he
>uses a statistical
>> method (the case control method) which was designed for a completely
different
>> type of problem. Gun ownership was not randomly assigned in the same
way that
>> a drug study for example assigns drugs to people. Many people may
have bought
>> guns because they were at particularly high risk of being attacked.
>
>RR: However, if Kellermann has correctly factored out
>societal risks of homicide, then IMM, the remaining
>correlation with guns (in this case 2.7X) might still be
>considered causal. The only risks of attack that might
>remain would be a personal attack threat... something so
>individualized that it could not be part of the overall
>analysis. And Kellermann commented on this potential
>drawback in the study, saying that if Joe Blow bought a gun
>because Sam Smith had threatened to kill him, then quite
>obviously the gun had failed as a self-defense tool since
>Sam Smith accomplished the deed.
>
He has not correctly factored this out because the probability should be
expected to systematically differ across subgroups. I will be
interested in your response after you read Chapter 2 in my book because
I spend some time trying to deal with this issue. When I debated
Kellermann he refused to discuss this issue with me. The discussion in
my chapter is taken directly from our debate.
>JL> Finally, unlike Kellermann's studies, my earlier study
>and my new book did not
>> focus on data from only one or three cities for only one year. I
studied data
>> for all the counties in the entire United States over many years. I
studied
>> murder rates, accidental gun deaths, and suicides. There is in fact
>> absolutely no evidence that any of Kellermann's claims are correct.
See More
>> Guns, Less Crime (University of Chicago Press, 1998).
>
>RR: I hope to soon. I've read your earlier study, and look
>forward to seeing your book.
> Finally, as a layman in this debate, I'd sure
>appreciate it if you would comment on the absolute rancor
>that's surfaced in recent years from academics on both sides
>of the issue. While I'm quite use to seeing the flames
>erupt in tpg, I personally find it disheartening when MDs
>and Phds go toe to toe in allegedly prestigious Law Reviews
>and Professional Journals.
> I'm sure Gun Control advocates are not immune from this
>disease, but I really thought that the Kates/Schaeffer
>article in the Tennessee Law Review hit a new low in
>rhetoric and name calling, without adding much of
>substance.
> Perhaps your guys need to communicate more among each
>other.. have some multi-disciplinary round-tables and even
>go out for a few drinks together. I wonder, did you even
>attempt to talk with Kellermann to get his meaning before
>you wrote your Myth #5?
> --- Robert L. Ray
Kellermann went over a year without returning my telephone calls, and I
was only was able to talk with him because I think that he accidentally
answered the telephone one time. When I would ask him for the data he
was evassive and noncommital on a date. I will be interested in your
reaction to Chapter 7 in my book where I talk about the debate. I have
been in a lot of academic debates before, but I have never seen anything
like the one over guns. As is true in most of academia, I have made all
my data immediately available. I have not waited 5 years to release a
portion of my data. I have spent dozens of hours working with those
that I have given my massive data set to that I put together with my own
money.
You need to evaluate this debate yourself, and I will be interested in
your final conclusion.
Sincerely,
John
Steve Fischer
In article <tnu360c...@oolong.orchestra.cse.unsw.EDU.AU> Tim Lambert <lam...@cse.unsw.EDU.AU> writes:
>"John R. Lott, Jr." <john...@law.uchicago.edu> writes:
>
>> RAY wrote:
>>
>> > David Friedman wrote:
>> > >
>> > > After reading the thread on Kellerman, I did a bit of web browsing. Steve
>> > > Kangas has a good summary of the Kellerman paper at:
>> > > The one crucial thing he does not say, and that is asserted in a piece
>> > > from _Reason_ elsewhere on the web, is that Kellerman has refused to make
>> > > his data available to other researchers. If that is true, then his study
>> > > cannot be trusted, since he can say anything he likes about the results,
>> > > true or false, without the risk that someone else will rerun his
>> > > regressions and prove them false.
>> >
>> > data to the Inter-university Consortium for Political and
>> > Social Research at Ann Arbor, Michigan. It is available to
>> > the public.
>>
>> years to release what he did.
>
>defensive gun use. When is he going to release the data?
>
>> released confirms that at absolute most 4 percent of the homicides could be
>> attributed to the gun owned by those in the residence. Kellermann's study
>> incorrectly assumed that all the gun deaths were caused by the gun in the
>> home. Obviously, this assumption is crucial for his claim. At least 96
>> percent of his deaths were falsely assigned.
>This is wrong from beginning to end. As far as I can tell, this
>"work" by Gary Kleck amounts to misreading a statement by Kellermann
>that that there only 14 gun homicides in homes where no guns were kept
>according to the proxy and in 8 of those cases the police noted that
>the gun involved had been kept in the home (i.e. the proxy was
>mistaken in those cases.) Kleck seems to have misunderstood this as
>meaning that in only 8 out of the 209 gun homicides did the police
^^^^^^
>state that the gun involved was kept in the home.
^^^^^^
You know, I looked through Kellerman's original paper twice
and I can't find a single statement by any cop anywhere in the
paper. Not only that, I can't find a single statement in the
paper in which Kellerman even reports the number of crime guns
that originated in the home. Why not? Supposedly he had police
reports? Why speculate?
So why is there no table showing something like:
Circumstances # cases Weapon Used # cases (weapon in home)
------------- ------- ----------- ------------------------
Rape 34 knife 9 ( 3)
handgun 6 ( 3)
other gun 2 ( 2)
physical force 6 (NA)
blunt instrument 8 ( 5)
burns, etc 3 (NA)
Robbery 13 knife 2 ( 0)
handgun 6 ( 4)
other gun 0 ( 0)
physical force 3 (NA)
blunt instrument 2 ( 1)
other 0 ( 0)
etc
Quarrel 160 knife
gun
etc
Romantic Triangle etc etc etc
Homicide only
Police Intervention
etc
John Johnson
In <3577f463....@newshost.cc.utexas.edu>
mccu...@mail.utexas.edu (Jim McCulloch) writes:
> On Thu, 4 Jun 1998 15:34:59 GMT, "John R. Lott, Jr."
> <john...@law.uchicago.edu> wrote:
>
>> Another big problem with Kellermann's work is that he uses a
>> statistical method (the case control method) which was designed
>> for a completely different type of problem. Gun ownership was
>> not randomly assigned in the same way that a drug study for
>> example assigns drugs to people. Many people may have bought
>> guns because they were at particularly high risk of being attacked.
>
> This seems like a common-sense statement, but do we really know this
> to be true? Here it Texas, the last time I checked, one class of gun
> owners, at least, the people who were getting permits to carry
> handguns, were overwhelmingly middle-aged middle-class white males,
> who statistically are not at particularly high risk being attacked.
> That is not to say they do not *believe* themselves at high risk ;-).
If one were to research this very deeply, I would be almost
willing to bet they would find this is the same group of Texans
who provide for the welfare, safety and security of themselves
and their familes with:
Life Insurance,
Health Insurance,
Homeowners Insurance,
Auto Insurance
A Texan's CHL is merely "Personal Protection Insurance."
--John Johnson/TX Peace Officer (15+ Years) supporting the
Texas and U.S. Constitutions, the BoR, the 2ndAmnd and the RKBA
"If you're 20 and not a liberal, you have no heart.
If you're 40 and not a conservative, you have no brains."
--Sir Winston Churchill
"The young man who is not liberal is a dullard.
The old man who is not conservative is a fool."
--Samuel Langhorn Clemens, aka: Mark Twain
"Conservatives are afraid the people WON'T understand;
Liberals are afraid the people WILL understand!!!"
--Unknown
"If liberals interpreted the Second Amendment the way they
interpret the rest of the Bill of Rights, there would
be law professors arguing that gun ownership is mandatory"
--Mickey Kaus, Editor: "New Republic"
"The greatest dangers to liberty lurk in insidious encroachment
by men of zeal, well-meaning; but without understanding."
--Justice Louis D. Brandeis, U.S. Supreme Court
--
John_Johnson
TXJo...@ix.netcom.com
© 1998 All rights reserved
Dan Z
In <3577f463....@newshost.cc.utexas.edu>
mccu...@mail.utexas.edu (Jim McCulloch) writes:
>
>On Thu, 4 Jun 1998 15:34:59 GMT, "John R. Lott, Jr."
><john...@law.uchicago.edu> wrote:
>
>>Another big problem with Kellermann's work is that he uses a
statistical
>>method (the case control method) which was designed for a completely
different
>>type of problem. Gun ownership was not randomly assigned in the same
way that
>>a drug study for example assigns drugs to people. Many people may
have bought
>>guns because they were at particularly high risk of being attacked.
>
>This seems like a common-sense statement, but do we really know this
>to be true? Here it Texas, the last time I checked, one class of gun
>owners, at least, the people who were getting permits to carry
>handguns, were overwhelmingly middle-aged middle-class white males,
>who statistically are not at particularly high risk being attacked.
>That is not to say they do not *believe* themselves at high risk ;-).
>
>*may* have bought guns because they were in danger, but do you
>actually know what percentage of the gun owners bought guns because
>they were, objectively, "at high risk"? Seems like a hard thing to
>prove, to me. You certainly can't use the occaional homicide in the
>sample as proof, because the cause of that is what is in
>question--that reasoning would not only be circular, but very tightly
>circular. You would need some sort of auxiliary evidence, like, say,
>prior police reports of threats to the gun owners, wouldn't you? Do
>you have any actual evidence for your statement?
>
You missed the entire point by concentrating on the last sentence
instead of the two prior to it. The last sentence was simply a possible
example (note the use of the word "may") to illustrate the point in the
prior sentences.
uspc...@my-dejanews.com
In article <3576BE82...@law.uchicago.edu>,> RAY wrote:
> > RR: Actually, it is NOT true. Kellermann has provided his
> > data to the Inter-university Consortium for Political and
> > Social Research at Ann Arbor, Michigan. It is available to
> > the public.
> Not all Kellermann's data has yet to be released, and he waited over five
> years to release what he did. Recent work by Gary Kleck using what has been
> released confirms that at absolute most 4 percent of the homicides could be
> attributed to the gun owned by those in the residence. Kellermann's study
> incorrectly assumed that all the gun deaths were caused by the gun in the
> home. Obviously, this assumption is crucial for his claim. At least 96
> percent of his deaths were falsely assigned.
>
> Another big problem with Kellermann's work is that he uses a statistical
> method (the case control method) which was designed for a completely different
> type of problem. Gun ownership was not randomly assigned in the same way that
> a drug study for example assigns drugs to people. Many people may have bought
> guns because they were at particularly high risk of being attacked.
>
> Finally, unlike Kellermann's studies, my earlier study and my new book did not
> focus on data from only one or three cities for only one year. I studied data
> for all the counties in the entire United States over many years. I studied
> murder rates, accidental gun deaths, and suicides. There is in fact
> absolutely no evidence that any of Kellermann's claims are correct. See More
> Guns, Less Crime (University of Chicago Press, 1998).
Kellermann's biggest problem was that he decided to make a study proving that
guns are bad, in spite of the results. What he didn't do was decide to make a
study to see if guns are bad, then make an honest effort on it, regardless of
the results.
Jim
-----== Posted via Deja News, The Leader in Internet Discussion ==-----
http://www.dejanews.com/ Now offering spam-free web-based newsreading
julia.c...@iint.com
In article <3577f463....@newshost.cc.utexas.edu>,
mccu...@mail.utexas.edu (Jim McCulloch) wrote:
>
> On Thu, 4 Jun 1998 15:34:59 GMT, "John R. Lott, Jr."
> <john...@law.uchicago.edu> wrote:
>
> >Another big problem with Kellermann's work is that he uses a statistical
> >method (the case control method) which was designed for a completely
different
> >type of problem. Gun ownership was not randomly assigned in the same way
that
> >a drug study for example assigns drugs to people. Many people may have
bought
> >guns because they were at particularly high risk of being attacked.
>
> to be true? Here it Texas, the last time I checked, one class of gun
> owners, at least, the people who were getting permits to carry
> handguns, were overwhelmingly middle-aged middle-class white males,
> who statistically are not at particularly high risk being attacked.
> That is not to say they do not *believe* themselves at high risk ;-).
>
> In any case, you can very well say the people in Kellermann's sample
> *may* have bought guns because they were in danger, but do you
> actually know what percentage of the gun owners bought guns because
> they were, objectively, "at high risk"? Seems like a hard thing to
> prove, to me. You certainly can't use the occaional homicide in the
Okay, this is where you've got a problem in your reasoning.
The object of scientific research is to prove things. Your study
only proves things if your assumptions are statistically valid.
If you presume a population is "random" in your study, and someone
criticizing your study points out that that sample "may" not be
random, whammo, all your statistics become invalid and your study
doesn't prove squat-----which means some other bright researcher
with a grant has to either figure out a way to conclusively rule
out that "may", or you have to go with a study design whose statistical
tools don't require that particular population to be random.
Lott's showing a plausible reason why the sample "may" not be
random is all he needs to do to completely torpedo the study's
credibility, because as a result of the study design used if
that sample isn't random, the researcher doesn't have squat.
Lott's shown that because of that "may", Kellerman's study doesn't
*prove* what it says it proves.
The point is that not only doesn't Lott know what percentage of people
bought their guns because they were at high risk, *Kellerman doesn't
either*----and Kellerman's unproven assumption that the data is random
is critical for his study to prove what he says it does.
> sample as proof, because the cause of that is what is in
> question--that reasoning would not only be circular, but very tightly
> circular. You would need some sort of auxiliary evidence, like, say,
> prior police reports of threats to the gun owners, wouldn't you? Do
> you have any actual evidence for your statement?
You're backwards on who has to have the evidence. When you publish a
study, that study's claim to being worth the paper it's printed on is
that it supposedly adds some new scientific knowledge to the world---it
supposedly proves something. Someone reviewing that study doesn't have
to *disprove* it to shoot it down, all they have to do is find a hole
in its reasoning, design, or execution that causes it not to prove what
it says it proves.
Lott's done that.
The typical academic response at this point, among real social scientists,
is to either document why your reasoning doesn't rely on that population
being random (can't be done with the study style Kellerman used), or to
redesign your study so it doesn't have to rely on that and then do it
over, or to try to measure the degree to which the sample isn't random
and use messy statistical techniques to factor it out.
The ideal study would be to pick 1000 random households and get them to
get rid of all their guns, give a random half of them guns and get them to
agree to keep them, make everyone promise not to change their gun-owning
status from the preset one over 40 years (set it up double-blind) and
follow them and see what happens. Nobody does it because compliance is
a problem and it's prohibitively expensive, not to mention a long wait to
get your results. Even with that you run into real world problems that
some people to whom you give guns might have self-selected themselves out
of owning guns if left alone, perhaps knowing they had a hot temper.
Anyway, it's the guy doing the study's job to account for all the maybe's
and perhaps's that might be statistically relevant in his study. All the
guys reading the study looking to pick it apart are supposed to look for
maybe's and perhaps's that the guy doing the study forgot to account for.
If they find one, the guy doing the study loses. This is made "fair" in
that all the guys reading the study looking to pick it apart also do and
publish their own studies that are each subject to being picked apart the
exact same way.
Clear?
Julie
>
> --Jim McCulloch
Sam A. Kersh
RAY <Ki...@Interaccess.com> wrote:
>David Veal wrote:
>
>Kellermann wrote:
>> >To the Editor:
>> [...]
>> >The exaggerated claim that guns are used in self-defense more than
>> >a million times a year has not withstood scientific scrutiny (3,4).
>> >If a gun in the home affords substantial protection from homicide
>> >(whether it is used to injure, kill, or frighten intruders or
>> >simply discourage them from entering), we should have found that
>> >homes in which a homicide occurred were less likely to contain a
>> >gun than similar households in which a homicide did not occur. The
>> >opposite was true.
>>
>DV> This is nonsense. In order to conclude this it would
>be necessary
>> for each control household to be subject to the same danger as the
>> household in which the murder took place. The study didn't attempt to
>> pair homes of equal risk ... in fact, the study more than amply
>> demonstrated that homes in which someone was murdered suffer from a whole
>> host of danger-predicting characteristics that their control pairs didn't.
>
>RR: David, you forget that Kellermann then factored out
>these risks for homicide, and STILL came up with a 2.7 times
>added risk with firearms in the home, even with the risks
>of both cases and controls evened out.
What Kellerman did was ignore his own findings. He wrote:
"The ...characteristics of the victims and controls were similar
except that the case subjects were more likely to have rented their
homes (70.4% v 47.3%).. lived alone (26.8 v 11.9%) ...
Alcohol was more commonly consumed by one or more members of
household of case subjects... (and)..by case subjects themselves....
Illegal drug-use by case subject....was also reported more commonly by
case households than control...
Previous episodes of violence were reported more frequently by
..case households.. "
On final analysis, Kellermann found(1):
1. Rented 5.9 times
2. Lived alone 3.4
3. Alcohol related problems 7.0
4.Use of illicit drugs 9.0
5. Gun kept for self-defense 1.7(2)
1. Table 3, page 1088, NEJM, Vol 329, No15, Oct 7, 1993. At this point,
the 444 had become 388 matched pairs used in the univariate analysis.
2. No, this isn't a typo, the 2.7 says "any gun kept loaded." When
Kellermann finally reached this number, he had "adjusted" the original
444 down to 316 matched pairs case/control in the logistic-regression.
Sam A. Kersh
NRA Life Member
TSRA Life Member
L.E.A.A., JPFO
http://www.flash.net/~csmkersh/csmkersh.htm
=======================================================
Read the book UNINTENDED CONSEQUENCES.
Reviews are at
http://www.amazon.com/exec/obidos/ISBN%3D1888118040
or
http://www.nguworld.com/vindex/95/123195vs.htm
Sam A. Kersh
RAY <Ki...@Interaccess.com> wrote:
>Bill B. wrote:
>>
>> The exaggerated claim that guns are used in self-defense more than
>> a million times a year has not withstood scientific scrutiny (3,4).
>> If a gun in the home affords substantial protection from homicide
>> (whether it is used to injure, kill, or frighten intruders or
>> simply discourage them from entering), we should have found that
>> homes in which a homicide occurred were less likely to contain a
>> gun than similar households in which a homicide did not occur. The
>> opposite was true.
>>
>> >>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>
>>
>BB> And homes in which homicides occurred also had toilets.
>Hence toilets kill!
>
>RR: However, after a final double-blind multi-factorial
>deep-regression analysis, Professor Alphonse HighBottom has
>concluded that there is a 6.34422 chance that homes
>containing guns are more likely to have an outhouse than an
>'inhouse' :-)
Ah, but Kellermann himself the very strong likelihood of reverse
causation.
Sam A. Kersh
RAY <Ki...@Interaccess.com> wrote:
> I have e-mailed Kellermann with your Tribune paragraph
>(was it Myth #5?), and he says your contention is a
>distortion.
>
We only have your word on that and you know how much your word carries
here.....
>JL> Another big problem with Kellermann's work is that he
>> uses a statistical method (the case control method)
>> which was designed for a completely different type of
>> problem. Gun ownership was not randomly assigned in the
>> same way that a drug study for example assigns drugs to
>> people. Many people may have bought guns because they
>> were at particularly high risk of being attacked.
>
>RR: However, if Kellermann has correctly factored out
>societal risks of homicide, then IMM, the remaining
>correlation with guns (in this case 2.7X) might still be
>considered causal. The only risks of attack that might
>remain would be a personal attack threat... something so
>individualized that it could not be part of the overall
>analysis. And Kellermann commented on this potential
>drawback in the study, saying that if Joe Blow bought a gun
>because Sam Smith had threatened to kill him, then quite
>obviously the gun had failed as a self-defense tool since
>Sam Smith accomplished the deed.
>
>JL> Finally, unlike Kellermann's studies, my earlier study
>> and my new book did not focus on data from only one or
>> three cities for only one year. I studied data
>> for all the counties in the entire United States over
>> many years. I studied murder rates, accidental gun
>> deaths, and suicides. There is in fact absolutely no
>> evidence that any of Kellermann's claims are correct.
>> See More Guns, Less Crime (University of Chicago Press, 1998).
>
>RR: I hope to soon. I've read your earlier study, and look
>forward to seeing your book.
The book is out and can be bought either through Amazon or Barnes Noble.
Have my copy, as you are aware of, Robert L, remember. You accused Prof
Lott of lying about Kellermann not being open that only 8 of the
firearms could be specifically assigned as the one kept in the case
household....
David Friedman
>On Thu, 4 Jun 1998 15:34:59 GMT, "John R. Lott, Jr."
><john...@law.uchicago.edu> wrote:
>
>>Another big problem with Kellermann's work is that he uses a statistical
>>method (the case control method) which was designed for a completely different
>>type of problem. Gun ownership was not randomly assigned in the same way that
>>a drug study for example assigns drugs to people. Many people may have bought
>>guns because they were at particularly high risk of being attacked.
>
>This seems like a common-sense statement, but do we really know this
>to be true? Here it Texas, the last time I checked, one class of gun
>owners, at least, the people who were getting permits to carry
>handguns, were overwhelmingly middle-aged middle-class white males,
>who statistically are not at particularly high risk being attacked.
>That is not to say they do not *believe* themselves at high risk ;-).
Remember that Kellerman was trying to use controls who matched the
observable characteristics of the murder victims. So the question is,
between two middle-aged, middle-calss white males, is the one who bought a
gun more likely to be the one at risk of being murdered?
>In any case, you can very well say the people in Kellermann's sample
>*may* have bought guns because they were in danger, but do you
>actually know what percentage of the gun owners bought guns because
>they were, objectively, "at high risk"?
Obviously not for Kellerman's sample--he (or one of his coauthors), after
all, was the one who wrote the questionaire. The question is not "do we
know that his result is wrong" but "how good is our reason for thinking it
is right?" John has offered an argument for the claim that we do not have
have a very strong reason for believing it is right.
Oracle
Kellermann is not my "favorite expert." I know little of his
work. In fact, because I'm not conversant in the details of any
statistical analysis, I really don't have a favorite. Whatever
opinion I've developed of the validity of any such work is based
solely on the reputation and credentials of its author.
--
xxxxxxxxxxxxxxx
HANDGUN CONTROL
xxxxxxxxxxxxxxx
xxxxxxxxxxxxxxx
jackson dryden <jackson...@nashville.com> wrote in article
<01bd9012$f592eda0$77b441cf@jackson>...
>
> --
Bill Bailey
Well, if that's truly your criteria, you've been treated to a good dose of
the reputation and credentials of John Lott. What do you think?
Bill Bailey
Oracle <aa1...@tseinc.com> wrote in article
<01bd90e6$5db89e60$188653d1@xbwerrsn>...
James F. Mayer
In <01bd90e6$5db89e60$188653d1@xbwerrsn> "Oracle" <aa1...@tseinc.com>
writes:
>Kellermann is not my "favorite expert." I know little of his
>work. In fact, because I'm not conversant in the details of any
>statistical analysis, I really don't have a favorite. Whatever
>opinion I've developed of the validity of any such work is based
>solely on the reputation and credentials of its author.
>
No, your opinions have been developed by whether or not the
person's work supports your socialistic agenda and has nothing at all
to do with credentials or reputation.
Jim McCulloch
> Okay, this is where you've got a problem in your reasoning.
> The object of scientific research is to prove things. Your study
> only proves things if your assumptions are statistically valid.
> If you presume a population is "random" in your study, and someone
> criticizing your study points out that that sample "may" not be
> random, whammo, all your statistics become invalid
That's nonsense. The criticism has to be, well, plausible, to perform
this magic. The tobacco industry for years could think of a multitide of
*possible* weaknesses, including possible sampling non-randomness --which
however were only *remotely* probable-- in the various studies indicating
links between smoking and several illnesses, and they did indeed hope that
the magic you mention, "whammo, all your statistics become invalid", would
occur. But it didn't.
My question to Mr. Lott had to do with the plausibility of his criticism
concerning nonrandomness. It seems to me from what I have read that many
gun owners do not buy guns because they are objectively in danger.
Now Mr. Lott is perfectly correct that such a nonrandomness *may* exist.
On the other hand, it *may not*. I was bringing to his attention some
possible evidence against the plausibility of his claim. He may have an
answer. I don't know. If he doesn't, and if this potential nonrandomness
is essential to his claim that Kellermann's study is wrong, then he
certainly needs to show us why his criticism in not in the same league
with the criticisms of smoking studies by the tobacco industry, which
everyone agreed identified possible weaknesses, but very few people really
thought to be very likely.
(more of the same snipped)
--Jim McCulloch
John R. Lott, Jr.
Jim McCulloch wrote:
> In article <6l9na6$r9b$1...@nnrp1.dejanews.com>, julia.c...@iint.com wrote:
>
> > Okay, this is where you've got a problem in your reasoning.
> > The object of scientific research is to prove things. Your study
> > only proves things if your assumptions are statistically valid.
> > If you presume a population is "random" in your study, and someone
> > criticizing your study points out that that sample "may" not be
> > random, whammo, all your statistics become invalid
>
> That's nonsense. The criticism has to be, well, plausible, to perform
> this magic. The tobacco industry for years could think of a multitide of
> *possible* weaknesses, including possible sampling non-randomness --which
> however were only *remotely* probable-- in the various studies indicating
> links between smoking and several illnesses, and they did indeed hope that
> the magic you mention, "whammo, all your statistics become invalid", would
> occur. But it didn't.
>
> My question to Mr. Lott had to do with the plausibility of his criticism
> concerning nonrandomness. It seems to me from what I have read that many
> gun owners do not buy guns because they are objectively in danger.
>
> Now Mr. Lott is perfectly correct that such a nonrandomness *may* exist.
> On the other hand, it *may not*. I was bringing to his attention some
> possible evidence against the plausibility of his claim. He may have an
> answer. I don't know. If he doesn't, and if this potential nonrandomness
> is essential to his claim that Kellermann's study is wrong, then he
> certainly needs to show us why his criticism in not in the same league
> with the criticisms of smoking studies by the tobacco industry, which
> everyone agreed identified possible weaknesses, but very few people really
> thought to be very likely.
>
> (more of the same snipped)
>
> --Jim McCulloch
My original paper with Mustard as well as my book provides a great deal of
evidence that the places with the highest crime rates issued the most permits
and had the biggest drops in violent crime rates.
Jim is correct that this is a potential problem. Simply pointing to the
possibility is not the same thing as proving it. However, there is also a
large amount of evidence from areas other than concealed handgun use that those
who face the largest risks from crime take the greatest efforts to protect
themselves from those threats. The point about the endogeniety problem is that
Kellermann ignores this issue, and he uses tests that are particularly
vulnerable to this problem. What makes an academic paper a good paper is
whether it can anticipate potential objections and deal with them. I deal with
this endogeniety problem in my book and I look at all the counties in the
United States over many years (not just three places in one year), and I find
no evidence to support any of Kellermann's claims.
Here is a portion of the relevant discussion on Kellermann from my book (More
Guns, Less Crime, University of Chicago Press, 1998):
". . . suppose that this same statistical method — with a matching control
group — was used to do an analogous study on the efficacy of hospital care.
Assume that we collected data in the same way these authors did, that is, we
get a list of all the individuals who died in a particular county over the
period of a year and we asked their relatives whether they had been admitted to
a hospital during the previous year. We would also put together a control
sample with people of similar ages, sex, race, and neighborhoods, and ask these
men and women whether they had been in a hospital during the past year. My bet
is that we would find a very strong positive relationship between those who
spent time in hospitals and those who died, quite probably a stronger
relationship than in Kellermann’s study on homicides and gun ownership. If so,
would we take that as evidence that hospitals kill people? Hopefully not. We
would understand that despite controlling for age, sex, race, and neighborhood,
the people who had visited a hospital during the past year and the people in
the 'control' sample who did not visit a hospital were really not the same
types of people. The difference is pretty obvious: those hospitalized were
undoubtedly sick and thus it should come as no surprise that they would face a
higher probability of dying.
"The relationship between homicides and gun ownership is no different. The
finding that those who are more likely to own guns suffer a higher homicide
rate makes us ask: why were they more likely to own guns? Could it be that
they were at greater risk of being attacked? Is it possible that this
difference arose because of a higher rate of illegal activities by those in the
case study group than in the control group? Owning a gun could lower the
probability of attack but still leave it higher than the probability faced by
those who never felt the need to buy a gun to begin with. The fact that all or
virtually all the homicide victims died from a weapon brought into their home
by an intruder makes this all the more plausible.
"Unfortunately, the case studies method was not designed to study these types
of social issues. Compare these endogeneity concerns with a laboratory
experiment to test the effectiveness of a new drug. Some patients with the
disease are provided with the drug, while others are given a placebo. The
random assignment of who gets the drug and who receives the placebo is
extremely important. The comparable case to the homicide and guns link would
have researchers randomly place guns inside certain households and also
randomly determine in which households guns would be forbidden. Who receives a
gun would not be determined by other factors that might themselves be related
to whether a person faces a high probability of being killed.
"So how does one solve this causation problem? Think for a moment about the
hospital example that we just discussed. One way is to examine a change in
something like the cost of going to hospitals. For example, one could see if
the cost of going to hospitals fell, some people who would otherwise have not
gone to the hospital would now seek help there. As we observed an increase in
the number of people going to hospitals, we could then check to see whether
this was associated with an increase or decrease in the number of deaths. By
examining changes in hospital care prices, we could see what happens to people
who either choose to go to the hospital or not who were otherwise similar in
terms of those characteristics that would determine their probability of
living."
My empirical work on murder, accidental death rates, and suicide is spread
throughout the book. I hope that this answers your questions.
pyotr filipivich
"Oracle" <aa1...@tseinc.com> writes:
>Kellermann is not my "favorite expert." I know little of his
>work. In fact, because I'm not conversant in the details of any
>statistical analysis, I really don't have a favorite. Whatever
>opinion I've developed of the validity of any such work is based
>solely on the reputation and credentials of its author.
In other words, Oracle doesn't concern himself with facts, only
with how he feels about the subject.
--
pyotr filipivich, AKA Nickolai Petrovich.
When Napoleon was my age, he was Emperor.
David Friedman
In article <mcculloch-060...@dial-47-96.ots.utexas.edu>,
mccu...@mail.utexas.edu (Jim McCulloch) wrote:
>My question to Mr. Lott had to do with the plausibility of his criticism
>concerning nonrandomness. It seems to me from what I have read that many
>gun owners do not buy guns because they are objectively in danger.
Sure. The question is not how large a fraction of people who buy guns are
objectively in danger, but how large a fraction of people who are
objectively in danger buy guns, relative to those who don't--and how large
a fraction of each get murdered.
Suppose the average, not objectively in danger, person has a .2
probability of keeping a handgun in the house, and the average,
objectively in danger person has a .4 probability, which doesn't seem
particularly implausible to me. Further suppose that the average not
objectively in danger person has a .001 chance of being murdered, and the
average objectively in danger person a .002 chance, seems a very modest
estimate (of the ratio--all of the numbers are actually too high). Since
being objectively in danger is an unobserved variable, we attribute all of
the difference (after controlling on other characteristics--make them
identical for purposes of the exampe) to gun ownership. So an increase of
.2 guns appears to increase the probability of being killed by .001. So an
increase of one gun increases it by .005. So having a gun in your home
makes you 5 times more likely to get killed than the average person--a
much stronger result than Kellerman actually found. (All of this is a bit
oversimplified; I don't know the details of his regression, whether he was
using lots, etc., but it gives you a rough idea).
Which demonstrates that it doesn't take very strong effects to give the result.
RAY
John R. Lott, Jr. wrote:
> >RR: Thank you for your response. I had seen you make this
> >claim in the Chicago Tribune (although I did not see any
> >attribution to Kleck there). Perhaps you can tell me what
> >data source this 96% figure comes from.
> > As I've posted in TPG, the only source I have found
> >which comes close is a letter Kellermann wrote in the NEJM
> >responding to his critics, where he said police had noted
> >that 8 of 14 cases of homicide were committed with a gun
> >that 'had been' in the home. If this is the only source
> >that Kleck or you have, I would suggest that your
> >interpretation is speculative, to say the least.
> > I have e-mailed Kellermann with your Tribune paragraph
> >(was it Myth #5?), and he says your contention is a
> >distortion.
>
JL> Gary Kleck has looked into this quite closely and should
be able to
> provide you information on this if you write him at Florida State. The
> reference in th NEJM is not the source for this 4 percent, it is from
> what data that Kellermann has released. Kellermann refused to release
> any data to me while I was still working on the book so that the only
> information that I had was from the difficult to interpret response that
> Kellermann had in the NEJM. When I called Kellermann and coauthors to
> ask them to ellaborate on the quote they refused to do so. It was the
> only information that was available, and as I say I tried talking to
> them about it when I wrote that section of my book.
> As to Kellermann saying that I distorted his point, he should have been
> willing to talk to me about it when I called him. I would also suggest
> that he specifically state what the true correct number is and give us
> all the hard data so that we can evaluate his claim.
RR: Now I'm a little confused on the data for this 4%
comment. First you've said that Kellermann has not released
ALL of his data to the Michigan consortium. Now you say
that Kleck has "looked into it," presumably using enough of
Kellermann's data to form a 4% opinion. So, either
Kellermann released sufficient data, or Gary Kleck is not
telling you the truth. Regardless, I still think it was
speculative on your part to publicise the 4% opinion when
the only documented evidence you personally had was the
Kellermann letter, which you have admitted to being
"difficult to interpret."
Next time I get a chance to write to Kellermann I shall
ask him about your phone bill, and about the guns used in
the remaining 430 homicides. From the study itself we know
that only about half of the studied 420 cases were committed
with guns. While it would more clear if we knew which of
these 209 gun homicides were committed with the guns in the
home, I don't think this knowledge is essential in order to
show that guns in the home raise the risk of homicide in the
home.
For instance, those who own guns may be more amenable to
their friends, family or acquaintences bringing their own
guns into the home. Should a deadly argument ensue, the
friend's gun might have been used. Had the homeowner not
owned a gun, friends would be less likely to bring one with
them for a visit, and such a shooting would not occur.
Furthermore, should an enemy wish to do you harm, or even
wish to engage you in what may become an ugly argument, and
should that enemy have knowledge that you own a gun, he/she
may be more inclined to bring a gun themselves for
protection. In the course of the fight, that gun from
outside may be the murder weapon, although it was the
homeowner's possession of a gun which resulted in the murder
weapon originally being brought into the home.
> >JL> Another big problem with Kellermann's work is that he
> >uses a statistical method (the case control method)
> >which was designed for a completely different
> > type of problem. Gun ownership was not randomly
>> assigned in the same way that a drug study for example
> > assigns drugs to people. Many people may have bought
> > guns because they were at particularly high risk
> > of being attacked.
> >
> >RR: However, if Kellermann has correctly factored out
> >societal risks of homicide, then IMM, the remaining
> >correlation with guns (in this case 2.7X) might still be
> >considered causal. The only risks of attack that might
> >remain would be a personal attack threat... something so
> >individualized that it could not be part of the overall
> >analysis. And Kellermann commented on this potential
> >drawback in the study, saying that if Joe Blow bought a gun
> >because Sam Smith had threatened to kill him, then quite
> >obviously the gun had failed as a self-defense tool since
> >Sam Smith accomplished the deed.
> >
JL> He has not correctly factored this out because the
> probability should be expected to systematically differ
> across subgroups. I will be interested in your response
> after you read Chapter 2 in my book because
> I spend some time trying to deal with this issue.
> When I debated Kellermann he refused to discuss this
> issue with me. The discussion in
> my chapter is taken directly from our debate.
RR: Although I have not yet read your book, this
"subgrouping" complaint sounds an awful lot like what Henry
Schafer has said. However, Schafer could provide no
evidence that Kellermann did not factor out subgroup
variables, nor did Schafer provide evidence that
Kellermann's control homes differed significantly
(accounting for the variables) in subgroups from the case
homes.
> >JL> Finally, unlike Kellermann's studies, my earlier
> > study and my new book did not focus on data from only
> > one or three cities for only one year. I
> > studied data for all the counties in the entire
> > United States over many years. I studied
> > murder rates, accidental gun deaths, and suicides.
> > There is in fact absolutely no evidence that any of
> > Kellermann's claims are correct. See More
> >> Guns, Less Crime (University of Chicago Press, 1998).
RR: How could you know? Your research did not ask the same
questions that Kellermann asked. Did you study homicides in
homes with guns as opposed to homes without guns? My
understanding of your study was that it compared crime at
the county level between states with concealed carry systems
vs states without concealed carry systems. That's quite
different from the gun-in-the-home population studied by
Kellermann.
Besides you yourself have written that CCW permits
would be most effective "in a place where victims otherwise
would not be allowed to carry firearms" (i.e. away from
their homes).
RR> > Finally, as a layman in this debate, I'd sure
> >appreciate it if you would comment on the absolute rancor
> >that's surfaced in recent years from academics on both sides
> >of the issue. While I'm quite use to seeing the flames
> >erupt in tpg, I personally find it disheartening when MDs
> >and Phds go toe to toe in allegedly prestigious Law Reviews
> >and Professional Journals.
> > I'm sure Gun Control advocates are not immune from this
> >disease, but I really thought that the Kates/Schaeffer
> >article in the Tennessee Law Review hit a new low in
> >rhetoric and name calling, without adding much of
> >substance.
> > Perhaps your guys need to communicate more among each
> >other.. have some multi-disciplinary round-tables and even
> >go out for a few drinks together. I wonder, did you even
> >attempt to talk with Kellermann to get his meaning before
> >you wrote your Myth #5?
JL> Kellermann went over a year without returning my
telephone calls, and I
> was only was able to talk with him because I think that he accidentally
> answered the telephone one time. When I would ask him for the data he
> was evassive and noncommital on a date. I will be interested in your
> reaction to Chapter 7 in my book where I talk about the debate. I have
> been in a lot of academic debates before, but I have never seen anything
> like the one over guns. As is true in most of academia, I have made all
> my data immediately available. I have not waited 5 years to release a
> portion of my data. I have spent dozens of hours working with those
> that I have given my massive data set to that I put together with my own
> money.
> You need to evaluate this debate yourself, and I will be interested in
> your final conclusion.
RR: Tell you what. Your book will be next on my plate.
The new UC Business school building is downtown relatively
near where I work.. and it has a bookstore. I'll see if
they carry it. In the meantime I had a number of questions
which I wrote in the margins of "Crime, Deterrence, and
Right to Carry Concealed Handguns." If you don't mind I'll
ask you a few in the next post.
Regarding the gun debate.. how did a U-C economist like
yourself get interested in it in the first place?
Steve Fischer
In article <3578601b...@news.flash.net> csmk...@flash.net (Sam A. Kersh) writes:
>RAY <Ki...@Interaccess.com> wrote:
>
>>David Veal wrote:
>>
>>Kellermann wrote:
>>> >To the Editor:
>>> [...]
>>> >a million times a year has not withstood scientific scrutiny (3,4).
>>> >If a gun in the home affords substantial protection from homicide
>>> >(whether it is used to injure, kill, or frighten intruders or
>>> >simply discourage them from entering), we should have found that
>>> >homes in which a homicide occurred were less likely to contain a
>>> >gun than similar households in which a homicide did not occur. The
>>> >opposite was true.
>>>
truly "independent" variables.
RAY
Dan Z wrote:
> >RR: What 'unobservable risks' are you talking about? As I
> >said, Kellermann already has factored out the societal
> >threat from specific individuals. Kellermann also takes
> >that into account with the observation that if these
> >homicide victims bought guns to protect themselves from
> >these personal threats, then the purchase of guns for this
> >self-defense purpose was in itself self-defeating.
> >Afterall, they were MURDERED!
> >local library. Just ask for the New England Journal of
> >Medicine dated 8/7/93, page 1084.
>
DZ quoting Polsby> Third, it is not surprising that
Kellerman and Reay's analysis shows
> that homicide victims are armed in disproportionate numbers, for it
> appears that a large and growing proportion of homicide victims are
> criminals themselves. [74] In other words, by focusing his analysis on
> households in which homicides occurred, Kellerman and Reay may be
> finding out information only about the characteristics of homicide
> victims, and people who commit murder, without any assurance that such
> information may be safely generalized to the gun-owning public at
> large. It is possible that the households in which homicides occur are
> far from representative of typical or average households in which guns
> are present. If so, treating the 43 times statistic as though it were a
> universal law applicable to all gun owners, rather than as descriptive
> of a discrete, aberrant subset, is simply wrong and misleading.
>
> [74]. See Daniel D. Polsby, "Firearms Costs, Firearms Benefits and
> the Limits of Knowledge," Journal of Law and Criminology 86
> (1995): 211.
RR: Of course Polsby was not talking about Kellermann's
more recent study (1993). Instead, Polsby was discussing
Kellermann's 1986 study. Two points:
1. The 1986 study was not just about homicides in the
home. It also added up gun deaths from accident and
suicide.
2. The 1993 study fixed the criticism that Polsby has of
the 1986 study. That's because the 1993 study factors out
the variables that allowed Polsby to claim that the only
gun-homicide connection was due to the "criminal" element.
Steve Fischer
In article <01bd90e6$5db89e60$188653d1@xbwerrsn> "Oracle" <aa1...@tseinc.com> writes:
>Kellermann is not my "favorite expert." I know little of his
>work. In fact, because I'm not conversant in the details of any
>statistical analysis, I really don't have a favorite. Whatever
>opinion I've developed of the validity of any such work is based
>solely on the reputation and credentials of its author.
of any of the researchers involved here. It was just a day or two
ago that John Lott was kind enough to allow us to look at HIS
Curriculum Vitae? Insofar as I am aware, none of the anti-gunners
have allowed the public to have access to theirs.
I wrote to Arthur Kellerman about a year ago and I never even
got an acknowledgement that he received my letter. I was polite.
I simply asked a few questions about his research. This is in
keeping with what others have said about him. It appears he doesn't
like having his work questioned.
I did have a short E-mail correspondence with Brian Wiersema of
the Univ of Maryland anti-violence (in truth, an anti-gun group)
research group some time ago. The back and forth lasted until it
was clear that I wasn't interested in buying what he had to sell.
Their group was kind enough, however, to send me a preprint of
one of their papers, including the technical appendix which was
left out of the published paper. Unhappily for them, the technical
appendix is what convinced me they were just playing with numbers.
I've twice written to Gary Kleck requesting literature and
comments and he was kind enough to respond each time.
Just the other day, after John Lott posted here, I wrote him
and he promptly answered me.
Some researchers, it seems, like to engage in the interchange
of idea, whereas others simply like to advance the cause of gun
control and shut down when their work is questioned.
What one thinks of a person's "reputation" often correlates
directly with his feelings on gun control and whether or not a
given author's work backs up those feelings.
Steve
Steve Fischer
In article <DDFr-05069...@ddfr.vip.best.com> DD...@best.com (David Friedman) writes:
>In article <3577f463....@newshost.cc.utexas.edu>,
>mccu...@mail.utexas.edu (Jim McCulloch) wrote:
>
>>On Thu, 4 Jun 1998 15:34:59 GMT, "John R. Lott, Jr."
>><john...@law.uchicago.edu> wrote:
>>
>>>method (the case control method) which was designed for a completely different
>>>type of problem. Gun ownership was not randomly assigned in the same way that
>>>a drug study for example assigns drugs to people. Many people may have bought
>>>guns because they were at particularly high risk of being attacked.
>>
>>to be true? Here it Texas, the last time I checked, one class of gun
>>owners, at least, the people who were getting permits to carry
>>handguns, were overwhelmingly middle-aged middle-class white males,
>>who statistically are not at particularly high risk being attacked.
>>That is not to say they do not *believe* themselves at high risk ;-).
>
>Remember that Kellerman was trying to use controls who matched the
>observable characteristics of the murder victims. So the question is,
Having just re-read the paper, it looks to me like the ONLY
things they matched for were age, sex, race and area of residence
(NEJM, 7 Oct 93, pg 1085, paragraph 4).
----- Steve Fischer
>>In any case, you can very well say the people in Kellermann's sample
>>*may* have bought guns because they were in danger, but do you
>>actually know what percentage of the gun owners bought guns because
>>they were, objectively, "at high risk"?
>Obviously not for Kellerman's sample--he (or one of his coauthors), after
>all, was the one who wrote the questionaire. The question is not "do we
>know that his result is wrong" but "how good is our reason for thinking it
>is right?" John has offered an argument for the claim that we do not have
>have a very strong reason for believing it is right.
--
/Steve D. Fischer/Atlanta, Georgia/str...@netcom.com/
RAY
David Friedman wrote:
> >RR: However, if Kellermann has correctly factored out
> >societal risks of homicide, then IMM, the remaining
> >correlation with guns (in this case 2.7X) might still be
> >considered causal. The only risks of attack that might
> >remain would be a personal attack threat... something so
> >individualized that it could not be part of the overall
> >analysis. And Kellermann commented on this potential
> >drawback in the study, saying that if Joe Blow bought a gun
> >because Sam Smith had threatened to kill him, then quite
> >obviously the gun had failed as a self-defense tool since
> >Sam Smith accomplished the deed.
>
DF> As I just pointed out in this thread, that is a very
poor argument--if
> Kellerman really made it, that is one bit of evidence that he is a
> demagogue rather than a scientist (although, of course, even scientists
> are not immune to the temptation to occasional demagoguery). Does anyone
> have the quote? Did he imply that he was answering the argument, or merely
> commenting on the (obvous) fact that self-defense does not always work.
> Consider two people, otherwise similar, save that A has been threatened
> with murder by someone and B has been not. A, not surprisingly, obtains a
> gun. Which of them is more likely to be a murder victim?
RR: The NRA would say that "B" is more likely to be
murdered, because "A" has had the 'foresight' to protect
himself with an 'equalizer.' Here's the Kellermann quote:
"Third, it is possible that reverse caustion accounted
for some of the association we observed between gun
ownership and homicide (ie in a limited number of cases,
people may have acquired a gun in response to a specific
threat.) If the source of that threat subsequently caused
the homicide, the link between guns in the home and homicide
may be due at least in part to the failure of these weapons
to provide adequate protection from the assailants."
DF> Again, consider two people, otherwise similar, one of
whom is a
> professional criminal and the other not. Which is more likely to own a
> gun? To be a murder victim?
RR: Hmmm... the NRA would say that the non-criminal is more
likely to own a gun. (200 million guns are out there.. there
can't be THAT many criminals).
And of course all this "criminal" stuff is irrelevant
because Kellermann factored out the criminal element and
STILL came up with guns being 2.7 times more likely to
increase the risk of homicide in the home.
RR> Finally, as a layman in this debate, I'd sure
> >appreciate it if you would comment on the absolute rancor
> >that's surfaced in recent years from academics on both sides
> >of the issue. While I'm quite use to seeing the flames
> >erupt in tpg, I personally find it disheartening when MDs
> >and Phds go toe to toe in allegedly prestigious Law Reviews
> >and Professional Journals.
>
DF> 1. I think you should distinguish between professional
scholars, such as
> Black and Nagin (the authors of the only scholarly criticism of the Lott
> and Mustard paper that I have seen), and professional political activists
> such as Stephen Terek. The latter have a clear personal stake in
> denigrating studies that produce results they don't like, and suffer
> virtually no professional loss from making bad, or even dishonest,
> arguments. I think the Black and Nagin article is in large part
> unconvincing, and it might be dishonest in the very weak sense of
> attempting to make their case look better than it is, but it isn't
> particularly rancorous.
>
> 2. I stumbled into the aftermath of a previous dispute of a similar sort
> some years back, and I think the story is instructive. As you may know,
> Isaac Ehrlich did some very controversial research on the deterrent effect
> of the death penalty, and was criticized in much the same fashion that
> John Lott has been.
>
> Some years ago, I had a colleague, a law professor, who is a very
> prominent figure in the criminalogical community. He is also, in my
> judgement, an unusually gentle and reasonable man. I happened to mention
> Ehrlich's work, and the response was just short of a diatribe. He claimed
> that the work had been refuted, that Ehrlich was dishonest, that he had
> refused to provide his data to other researchers, ... .I asked for
> evidence, and he lent me a volume of articles on deterrence edited by some
> friends of his.
>
> I read the volume, and it provided no support at all for any of his
> claims--on the contrary, the one article that dealt with one of Ehrlich's
> articles was by people who had been given Ehrlich's data by Ehrlich and
> had replicated his results. The summary article on studies of deterrence
> (not the death penalty specifically) found, as I recall, that out of
> thirteen such studies twelve had come up with a positive result. The only
> stuff in the book that was critical of deterrence, as far as I remember,
> was the introduction, written by the editors--and it consisted essentially
> of arguing that all of the evidence for deterrence could conceivably be
> explained away in one way or another. While that was true--statistical
> evidence can practically always be explained away if you are sufficiently
> ingenious--the editors offered no tests at all of their conjectural
> explanations.
>
> I pointed all of this out to my colleague, and his response was apologetic
> (as I say, he is an unusually reasonable man). Basically what he said was
> that Ehrlich's work had been used for evil purposes (to argue for the
> death penalty, which my colleague opposed), that he was sure its
> conclusion was wrong (not on statistical grounds--I don't think he knows
> much about statistics--but from dealing with criminals), and he thus had
> strong hostile feelings towards Ehrlich. Pretty clearly, his specific
> charges were things he had gotten from other people who shared his biases,
> and had never bothered to check.
RR: Sigh... Just goes to show that even the most
knowledgable among us still can suffer from human bias and
hubris.
RAY
Bill Bailey wrote:
> Kellermann's research model (as medical models typically do) excludes
> rational choice factors. (Yes, Ray, there are unobserved factors in
> Kellerman's research!) Similarly, the earlier social models of firearms
> research--which were very derivative of physical science models--excluded
> factors of rational choice. They counted guns and counted crimes (somewhat
> selectively) and declared more guns correlated with more crime. There was
> not much difference between that model and the medical contagion model.
> You caught crime from guns. The idea has found sybolic representation in a
> number of movies--the "evil" gun that goes from hand to hand causing
> murders and misery, and no person has sufficient strength of character to
> stop it. The idea that more crime (or at least more, the media artifact of
> more crime information) might cause more people in a given region to
> purchase more firearms wasn't even a consideration. From Zimring onward, I
> think that "mechanistic model" was pretty much the standard model for gun
> research.
RR: Perhaps you could explain what "rational choice factors"
were left out of the Kellermann 'model.'
Kellermann has factored out criminal intent, and
commented on the protective purchase of guns. What other
'rational choices' are you interested in?
Steve Fischer
In article <mcculloch-060...@dial-47-96.ots.utexas.edu> mccu...@mail.utexas.edu (Jim McCulloch) writes:
>In article <6l9na6$r9b$1...@nnrp1.dejanews.com>, julia.c...@iint.com wrote:
>
>> The object of scientific research is to prove things. Your study
>> only proves things if your assumptions are statistically valid.
>> If you presume a population is "random" in your study, and someone
>> criticizing your study points out that that sample "may" not be
>> random, whammo, all your statistics become invalid
>
>this magic. The tobacco industry for years could think of a multitide of
>*possible* weaknesses, including possible sampling non-randomness --which
>however were only *remotely* probable-- in the various studies indicating
>links between smoking and several illnesses, and they did indeed hope that
>the magic you mention, "whammo, all your statistics become invalid", would
>occur. But it didn't.
>
>concerning nonrandomness. It seems to me from what I have read that many
>gun owners do not buy guns because they are objectively in danger.
>Now Mr. Lott is perfectly correct that such a nonrandomness *may* exist.
>On the other hand, it *may not*. I was bringing to his attention some
>possible evidence against the plausibility of his claim. He may have an
>answer. I don't know. If he doesn't, and if this potential nonrandomness
>is essential to his claim that Kellermann's study is wrong, then he
>certainly needs to show us why his criticism in not in the same league
>with the criticisms of smoking studies by the tobacco industry, which
>everyone agreed identified possible weaknesses, but very few people really
>thought to be very likely.
I bought my first handgun in response to an attempted rape that
occurred in the parking lot of my apartment complex. That was way
back in 1984 or 85. Chris C. whom you know, bought his first gun in
response to being attacked one night at a Marta Station. Glenn you-
know-who was attacked on his way home when he lived near Ga Tech.
He started collecting guns then.
Steve
Steve Fischer
John, Kellerman didn't even use the same reporting period as
a base in all three counties (NEJM 7 oct 93, see pg 1084, column 2
under "Methods"). The study period was Aug 23, 1987 - Aug 23, 1992
in King County and Shelby County, and Jan 1, 1990 - Aug 23, 1992 in
Cuyahoga County.
I've always enjoyed watching people squirm when trying to
justify the "correctness" of a study using such a limited population
when compared with a study which encompasses the entire country.
If you want to study GUNS IN THE UNITED STATES, you can't have a
sample database better than the whole country.
RAY
John R. Lott, Jr. wrote:
> Guns, Less Crime, University of Chicago Press, 1998):
>
> ". . . suppose that this same statistical method — with a matching control
> group — was used to do an analogous study on the efficacy of hospital care.
> Assume that we collected data in the same way these authors did, that is, we
> get a list of all the individuals who died in a particular county over the
> period of a year and we asked their relatives whether they had been admitted to
> a hospital during the previous year. We would also put together a control
> sample with people of similar ages, sex, race, and neighborhoods, and ask these
> men and women whether they had been in a hospital during the past year. My bet
> is that we would find a very strong positive relationship between those who
> spent time in hospitals and those who died, quite probably a stronger
> relationship than in Kellermann’s study on homicides and gun ownership. If so,
> would we take that as evidence that hospitals kill people? Hopefully not. We
> would understand that despite controlling for age, sex, race, and neighborhood,
> the people who had visited a hospital during the past year and the people in
> the 'control' sample who did not visit a hospital were really not the same
> types of people. The difference is pretty obvious: those hospitalized were
> undoubtedly sick and thus it should come as no surprise that they would face a
> higher probability of dying.
RR: On the other hand, what if you also asked the proxies
and controls for the person's medical history, and also
whether they had suffered any fatal out-of-hospital
accidents or injuries?
What you might discover, for instance, is that when
hospitalized patients are allowed to go home, they STILL are
at a higher risk for death than the general population.
In fact, this is the very conclusion of a study in the
June 3rd Journal of the American Medical Association. It
looked at hospital trauma cases, and found that patients
who'd presumably been treated and 'cured' still died at a
higher rate when they were back at home than the average
person.
This may sound common sensical to many of us, but it
has profound implications for many of the models used by
health professionals, government agencies and insurance
companies.
Getting back to Kellermann, he DID ask far more
questions than merely whether there was a gun in the home.
As a result, he was able to factor out the variables that
you complain about.
JL> "The relationship between homicides and gun ownership is
no different. The
> finding that those who are more likely to own guns suffer a higher homicide
> rate makes us ask: why were they more likely to own guns? Could it be that
> they were at greater risk of being attacked? Is it possible that this
> difference arose because of a higher rate of illegal activities by those in the
> case study group than in the control group? Owning a gun could lower the
> probability of attack but still leave it higher than the probability faced by
> those who never felt the need to buy a gun to begin with. The fact that all or
> virtually all the homicide victims died from a weapon brought into their home
> by an intruder makes this all the more plausible.
>
> "Unfortunately, the case studies method was not designed to study these types
> of social issues.
RR: Alcohol, illicit drugs, physical fights, arrests..
don't you think these factors might just have something to
do with the 'risk of being attacked?' Don't you think that
people whose environment contains high scores in these
indices are more likely to become both perpetrators and
victims of such attacks? And yet you say that Kellermann
has not taken such risks into account?
JL> Compare these endogeneity concerns with a laboratory
> experiment to test the effectiveness of a new drug. Some patients with the
> disease are provided with the drug, while others are given a placebo. The
> random assignment of who gets the drug and who receives the placebo is
> extremely important.
RR: A retrospective case-control study cannot be
randomized. Why do you insist that it should have been?
JL> The comparable case to the homicide and guns link would
Steve Fischer
If Kellerman were more forthcoming with people, he wouldn't
get himself into these situations. He CLAIMED to have police
reports on each and every homicide. If a gun was found at the
scene, surely they would have had it tested ballistically to see
if it was the murder weapon. If no gun were ever found, that
would have been part of the report as well. It was his respon-
sibility to see to it that he counted how many "guns in the home"
turned out to be murder weapons. He presented no such table or
data.
> Next time I get a chance to write to Kellermann I shall
>ask him about your phone bill, and about the guns used in
>the remaining 430 homicides.
The "good stuff" (6-variable regression) came from 316
matched pairs, not 430.
Good luck. I wrote him about a year ago and he never
answered back. I was polite - I'm always polite when I
contact other academics about their work. Maybe he didn't
like having his results questioned?
>From the study itself we know
>that only about half of the studied 420 cases were committed
>with guns. While it would more clear if we knew which of
>these 209 gun homicides were committed with the guns in the
>home, I don't think this knowledge is essential in order to
>show that guns in the home raise the risk of homicide in the
>home.
> For instance, those who own guns may be more amenable to
>their friends, family or acquaintences bringing their own
>guns into the home. Should a deadly argument ensue, the
>friend's gun might have been used. Had the homeowner not
>owned a gun, friends would be less likely to bring one with
>them for a visit, and such a shooting would not occur.
You're grasping at straws now, Robert. Let's take the
last, most preposterous part of that statement first. When
people feel they have to bring a gun with them when they
visit, these people are not called friends, they're called
ENEMIES. Drug dealers approach each other this way - neither
trusting the other.
I often get into spirited arguments with friends as part
of political or religious discussions, etc. It would never
cross my mind, no matter how angry I got, to bring a gun with
me because my friends also own guns (most of them do). The
idea of shooting someone over an argument would never enter
our minds. Of course, my friends are not spouse-beating,
drug-dealing ex-convicts.
[NOTE to readers: If your friends ARE such
people, I counsel you to get new friends]
Finally, in the immediate heat of a killing rage, the idea
that taking the final step of taking a life would depend upon
whether the resident had a gun in another room is laughable.
If you're mad enough to kill, you're going to kill right then
and there regardless of other circumstances.
> Furthermore, should an enemy wish to do you harm, or even
>wish to engage you in what may become an ugly argument, and
>should that enemy have knowledge that you own a gun, he/she
>may be more inclined to bring a gun themselves for
>protection. In the course of the fight, that gun from
>outside may be the murder weapon, although it was the
>homeowner's possession of a gun which resulted in the murder
>weapon originally being brought into the home.
Now you're talking about people who are likely to be
criminals, not your ordinary whitebread, middle-class,
suburbanites. Drug dealers are good examples, as I mentioned
above. They know that other drug dealers carry weapons and
have large wads of cash around. People like me buy guns to
protect myself from people like that when we meet them on the
street. I don't visit their homes.
Why didn't Kellerman show exactly HOW he made the
corrections? Why do we have to guess at these things?
In a field filled with such contentiousness, you should
illustrate everything you do. Stick some equations in.
Give an example or two. Don't assume that everyone knows
what you did.
Doesn't it bother you that Kellerman refused to
discuss it? Why would an honest academic refuse to
answer the criticisms of a colleague? I deal with
experimentalists all the time in my work. The good
ones will discuss their work in detail - especially
the tests they did to rule out certain complications.
>> >JL> Finally, unlike Kellermann's studies, my earlier
>> > study and my new book did not focus on data from only
>> > one or three cities for only one year. I
>> > studied data for all the counties in the entire
>> > United States over many years. I studied
>> > murder rates, accidental gun deaths, and suicides.
>> > There is in fact absolutely no evidence that any of
>> > Kellermann's claims are correct. See More
>> >> Guns, Less Crime (University of Chicago Press, 1998).
>RR: How could you know? Your research did not ask the same
>questions that Kellermann asked. Did you study homicides in
>homes with guns as opposed to homes without guns?
Kellerman sampled a small fraction of all homicides,
namely 3 counties out of the entire United States, and
studied two of them over a 60 month period and another
over a 32 month period. His most extensive regressions
used only 316 matched pairs out of a total of 430 homicides.
The case samples were skewed in various ways with respect
to the controls, and the way he "corrected" for these
discrepancies is not clear to the reader. But most
scandalously, he failed to demonstrate that gun in the
home = homicide gun for the range of possible homicide
circumstances.
>My
>understanding of your study was that it compared crime at
>the county level between states with concealed carry systems
>vs states without concealed carry systems. That's quite
>different from the gun-in-the-home population studied by
>Kellermann.
> Besides you yourself have written that CCW permits
>would be most effective "in a place where victims otherwise
>would not be allowed to carry firearms" (i.e. away from
>their homes).
True. The case that convinced me to buy a handgun occurred
outside my apartment. The two instances in which I had to defend
myself with a gun were both outside the home. It is precisely
this venue that most anti-gunners want to ban guns, however.
>RR> > Finally, as a layman in this debate, I'd sure
>> >appreciate it if you would comment on the absolute rancor
>> >that's surfaced in recent years from academics on both sides
>> >of the issue. While I'm quite use to seeing the flames
>> >erupt in tpg, I personally find it disheartening when MDs
>> >and Phds go toe to toe in allegedly prestigious Law Reviews
>> >and Professional Journals.
>> > I'm sure Gun Control advocates are not immune from this
>> >disease, but I really thought that the Kates/Schaeffer
>> >article in the Tennessee Law Review hit a new low in
>> >rhetoric and name calling, without adding much of
>> >substance.
>> > Perhaps your guys need to communicate more among each
>> >other.. have some multi-disciplinary round-tables and even
>> >go out for a few drinks together. I wonder, did you even
>> >attempt to talk with Kellermann to get his meaning before
>> >you wrote your Myth #5?
>JL> Kellermann went over a year without returning my
>telephone calls, and I
>> was only was able to talk with him because I think that he accidentally
>> answered the telephone one time. When I would ask him for the data he
>> was evasive and noncommital on a date. I will be interested in your
>> reaction to Chapter 7 in my book where I talk about the debate. I have
>> been in a lot of academic debates before, but I have never seen anything
>> like the one over guns. As is true in most of academia, I have made all
>> my data immediately available. I have not waited 5 years to release a
>> portion of my data. I have spent dozens of hours working with those
>> that I have given my massive data set to that I put together with my own
>> money.
>> You need to evaluate this debate yourself, and I will be interested in
>> your final conclusion.
>RR: Tell you what. Your book will be next on my plate.
>The new UC Business school building is downtown relatively
>near where I work.. and it has a bookstore. I'll see if
>they carry it.
If they don't, I got my copy from http://www.amazone.com!
>In the meantime I had a number of questions
>which I wrote in the margins of "Crime, Deterrence, and
>Right to Carry Concealed Handguns." If you don't mind I'll
>ask you a few in the next post.
> Regarding the gun debate.. how did a U-C economist like
>yourself get interested in it in the first place?
I can't wait to read the back and forth on this. Ray does
ask good questions and I'm sure John's responses will be
full and forthcoming.
I'm sure Tim Lambert will want in on this too.
If only we could coax Gary Kleck to post here. <wish>
If you have any influence, Mr Lott??
I've given up on Kellerman. I don't think he'd be civil
anyway. All those axes out there grinding away .....
Steve Fischer
Steve Fischer
In article <35797C1A...@Interaccess.com> RAY <Ki...@Interaccess.com> writes:
>David Friedman wrote:
>
>RR: The NRA would say that "B" is more likely to be
>murdered, because "A" has had the 'foresight' to protect
>himself with an 'equalizer.' Here's the Kellermann quote:
Is it just me, or is anyone else a bit bothered when
Ray starts speaking for the NRA. Would gun-control advocates
believe me when attributing motives to HCI? :-)
>RR> And of course all this "criminal" stuff is irrelevant
>because Kellermann factored out the criminal element and
>STILL came up with guns being 2.7 times more likely to
>increase the risk of homicide in the home.
How'd he do that exactly? He never told us. How never
even gave us an indication what crimes they committed? What
were all those criminals doing with guns in the home in the
first place? That, in an of itself was probably a felony.
Steve Fischer
How do you factor out criminal intent? Was he a mind reader?
Steve Fischer
Did you read what Kellerman said about proxies in his paper?
Let me quote - NEJM 329(15), 7-Oct-93, pg 1085:
"In addition to recording the details of the incident
for law-enforcement purposes, investigators obtained
the names of persons *CLOSE* to the victim who might
provide us with an interview at a later date, thereby
serving as proxies for the victim. These lists were
supplementeed with names obtained from newspaper
accounts, obituaries and calls to funeral homes."
[NEJM 329(15), 7 Oct 93, pg 1085, col 1, para 2]
First, proxies are not necessarily witnesses. Second, they may
or may not reside in the same residence as the victim. We have
no idea what "close" means because Kellerman didn't present us
with any relevant information on them. The closeness of the
proxy to the victim has direct bearing on how much they knew
about him and his guns. Why no table such as: (numbers made up)
======================================================================
TABLE XX: Relationship of Proxies to Victims
Number of Proxies Relationship to Victim Witnesses?
----------------- ---------------------- ----------
44 Current Spouse, living together 10
25 Current Spouse, separated 1
33 Ex-Spouse 0
35 Current girl/boyfriend resident 13
29 Current girl/boyfriend not resident 0
27 Ex-girl/boyfriend 0
15 Sister or Brother of victim 4
10 Parent of victim 1
18 First degree relative of victim 2
5 Roommate of victim 3
35 Friend, living in area 0
10 Friend, not living in area 0
34 Neighbor 0
3 Other 0
=====================================================================
> What you might discover, for instance, is that when
>hospitalized patients are allowed to go home, they STILL are
>at a higher risk for death than the general population.
> In fact, this is the very conclusion of a study in the
>June 3rd Journal of the American Medical Association. It
>looked at hospital trauma cases, and found that patients
>who'd presumably been treated and 'cured' still died at a
>higher rate when they were back at home than the average
>person.
> This may sound common sensical to many of us, but it
>has profound implications for many of the models used by
>health professionals, government agencies and insurance
>companies.
> Getting back to Kellermann, he DID ask far more
>questions than merely whether there was a gun in the home.
>As a result, he was able to factor out the variables that
>you complain about.
Says you! I don't see any questions in the paper and I
don't see how he made the correction. What I do see is that
the more variables he tries to fit, the smaller the number
of case/control pairs he uses for obvious reasons.
>JL> "The relationship between homicides and gun ownership is
>no different. The
>> finding that those who are more likely to own guns suffer a higher homicide
>> rate makes us ask: why were they more likely to own guns? Could it be that
>> they were at greater risk of being attacked? Is it possible that this
>> difference arose because of a higher rate of illegal activities by those in the
>> case study group than in the control group? Owning a gun could lower the
>> probability of attack but still leave it higher than the probability faced by
>> those who never felt the need to buy a gun to begin with. The fact that all or
>> virtually all the homicide victims died from a weapon brought into their home
>> by an intruder makes this all the more plausible.
>>
>> "Unfortunately, the case studies method was not designed to study these types
>> of social issues.
>
>RR: Alcohol, illicit drugs, physical fights, arrests..
>don't you think these factors might just have something to
>do with the 'risk of being attacked?' Don't you think that
>people whose environment contains high scores in these
>indices are more likely to become both perpetrators and
>victims of such attacks? And yet you say that Kellermann
>has not taken such risks into account?
I don't see any examples of how he did it, do you?
>JL> Compare these endogeneity concerns with a laboratory
>> experiment to test the effectiveness of a new drug. Some patients with the
>> disease are provided with the drug, while others are given a placebo. The
>> random assignment of who gets the drug and who receives the placebo is
>> extremely important.
>
>RR: A retrospective case-control study cannot be
>randomized. Why do you insist that it should have been?
I notice you ducked the part about endogeneity within the
Kellerman sampling regions. I should call them microsamples,
since only 3 counties were involved - two for 60 months and
one for 32 months.
Steve
John A. Stovall
On Sat, 6 Jun 1998 19:56:57 GMT, str...@netcom.com (Steve Fischer)
wrote:
>our minds. Of course, my friends are not spouse-beating,
>drug-dealing ex-convicts.
>
> [NOTE to readers: If your friends ARE such
> people, I counsel you to get new friends]
>
Sort of remains me of some advice given me by a old Deputy Sheriff
where I grew up in West Texas. I paraphrase, "Sonny, if you don't want
to get in trouble with the law don't run with them that's always are."
Not my provider’s views.
John Alex Stovall
XVOXVOXVOXVOXVOXVOXVOXVOXVOXVOXVOXVOXVOXVOXVO
"....Long live Freedom and damn the ideologies,"
Said the gamey old back-maned wild boar
Tusking the turf on Mal Paso Mountain.
XVOXVOXVOXVOXVOXVOXVOXVOXVOXVOXVOXVOXVOXVOXVO
David Friedman
In article <stratosE...@netcom.com>, str...@netcom.com (Steve
Fischer) wrote:
> I've always enjoyed watching people squirm when trying to
>justify the "correctness" of a study using such a limited population
>when compared with a study which encompasses the entire country.
>If you want to study GUNS IN THE UNITED STATES, you can't have a
>sample database better than the whole country.
There may be many things wrong with Kellerman's study, but I don't think
this is a valid complaint. He didn't have the data to replicate his study
for the whole U.S., since he got it via questionaires. No doubt if someone
had offered him a billion dollars in funding ... .
David Friedman
In article <35798CD5...@Interaccess.com>, RAY <Ki...@Interaccess.com>
wrote:
>RR: Alcohol, illicit drugs, physical fights, arrests..
>don't you think these factors might just have something to
>do with the 'risk of being attacked?' Don't you think that
>people whose environment contains high scores in these
>indices are more likely to become both perpetrators and
>victims of such attacks? And yet you say that Kellermann
>has not taken such risks into account?
The question is not whether such measures give us some information on
whether people are likely to be attacked, but whether they give us
complete information.
Suppose you have two people who get the same scores on all those factors.
You discover an additional fact--one of them has purchased a gun. Don't
you think you will revise upward your estimate both that the is a criminal
and that he has reason to fear for his life? If so, Kellerman's method
will show a false positive even if there is really no causal relation
between gun ownership and being killed.
>RR: A retrospective case-control study cannot be
>randomized. Why do you insist that it should have been?
Because the inference Kellerman draws only works well if you have either
randomized, or are confident that you have picked up all important factors
that are causal for both your dependent and independent variables. It does
not sound as though Kellerman did, or could do, the latter.
I am curious, by the way, since you are defending Kellerman, what your
view is of the issue of releasing data. The assertion we have seen here,
and have not yet seen contradicted, is that Kellerman kept his data secret
for several years, and only released it under pressure. Do you have
evidence that either half of that is false? If not, would you take such
behavior as in itself suspicious?
John R. Lott, Jr.
> RR: Tell you what. Your book will be next on my plate.
> The new UC Business school building is downtown relatively
> near where I work.. and it has a bookstore. I'll see if
> they carry it. In the meantime I had a number of questions
> which I wrote in the margins of "Crime, Deterrence, and
> Right to Carry Concealed Handguns." If you don't mind I'll
> ask you a few in the next post.
> Regarding the gun debate.. how did a U-C economist like
> yourself get interested in it in the first place?
I have done a lot of research on crime over the years and I was chief
economist at the United States Sentencing Commission during 1988 and 1989.
About 5 years ago I was teaching a class that dealt with crime at the
University of Pennsylvania, and I thought that it would be interesting to
provide some readings on gun control. There were probably a couple of
hundred academic papers on gun control at that time, but I was amazed how
poorly done the papers on both sides of the debate were.
1) All the papes were either purely cross sectional or time series studies.
There are important weaknesses with both approaches. For example, with a
single time series it is difficult to exclude many factors that are occurring
at about the same time. With purely cross sectional data it is very
difficult to deal with causality issues because you don't know what things
looked like before and after the experiment. (See Chapter 2 in my book for a
more complete discussion.)
2) All the previous studies involved very small samples. The largest
previous gun control study looked at 170 cities within just one year (1980).
Small samples by themselves make it difficult to control for many alternative
factors that may also explain changes in the crime rate.
3) The vast majority of previous studies did not control for any other
factors that could effect crime rates, and no previous study attempted to
control for things like arrest rates or conviction rates or prison sentence
lengths.
Academics frequently decide to do studies either because something hasn't
been done before or because they think that they can do a better job. For me
it was this project arose because of this second reason
I know Borders on Michigan Ave carries the book and most other major
bookstores do also.
RAY
Steve Fischer wrote:
RR> >From the study itself we know
> >that only about half of the studied 420 cases were committed
> >with guns. While it would more clear if we knew which of
> >these 209 gun homicides were committed with the guns in the
> >home, I don't think this knowledge is essential in order to
> >show that guns in the home raise the risk of homicide in the
> >home.
>
> > For instance, those who own guns may be more amenable to
> >their friends, family or acquaintences bringing their own
> >guns into the home. Should a deadly argument ensue, the
> >friend's gun might have been used. Had the homeowner not
> >owned a gun, friends would be less likely to bring one with
> >them for a visit, and such a shooting would not occur.
>
SF> You're grasping at straws now, Robert. Let's take
the
> last, most preposterous part of that statement first. When
> people feel they have to bring a gun with them when they
> visit, these people are not called friends, they're called
> ENEMIES. Drug dealers approach each other this way - neither
> trusting the other.
RR: I imagine (in other homes) that friends quite
frequently bring their guns along. When Billie-Bob goes to
Bubba's house, they probably sit up half the night playing
with their guns. Were Billie-Bob to get riled at Bubba for
burning the grits, it's quite possible that Billie-Bob would
use his own gun to kill Bubba.
SF> I often get into spirited arguments with friends as
part
> of political or religious discussions, etc. It would never
> cross my mind, no matter how angry I got, to bring a gun with
> me because my friends also own guns (most of them do). The
> idea of shooting someone over an argument would never enter
> our minds. Of course, my friends are not spouse-beating,
> drug-dealing ex-convicts.
RR: I'll bet you would have invited the Hartmans over for
tea, though. What if she had brought her gun with, and then
went nuts? Friends and acquaintences can wind up doing the
Strangest Things, don't you think?
SF> Finally, in the immediate heat of a killing rage, the
idea
> that taking the final step of taking a life would depend upon
> whether the resident had a gun in another room is laughable.
> If you're mad enough to kill, you're going to kill right then
> and there regardless of other circumstances.
RR: There are, perhaps, five second rages and five hour
rages (The Chicago Bear's Alonzo Spellman comes to mind).
No one would be safe in either case with a gun anywhere in
the home.
RR> > Furthermore, should an enemy wish to do you harm, or
even
> >wish to engage you in what may become an ugly argument, and
> >should that enemy have knowledge that you own a gun, he/she
> >may be more inclined to bring a gun themselves for
> >protection. In the course of the fight, that gun from
> >outside may be the murder weapon, although it was the
> >homeowner's possession of a gun which resulted in the murder
> >weapon originally being brought into the home.
>
SF> Now you're talking about people who are likely to
be
> criminals, not your ordinary whitebread, middle-class,
> suburbanites. Drug dealers are good examples, as I mentioned
> above. They know that other drug dealers carry weapons and
> have large wads of cash around. People like me buy guns to
> protect myself from people like that when we meet them on the
> street. I don't visit their homes.
RR: What about that carpet layer or refridgerator repairman
that likes to do a little extra business at night by
revisiting his customers' homes to burglarize? He might
notice your NRA wallpaper and figure he ought to come back
packin'
Then you confront him, and of course he gets the draw
on you, and it's good bye Steve! If he hadn't thought you
had a gun, perhaps he would have snuck in unarmed, and run
away when you woke up. That's just one example. I'm sure
there are plenty of others.
(snip)
SF> Doesn't it bother you that Kellerman refused to
> discuss it? Why would an honest academic refuse to
> answer the criticisms of a colleague? I deal with
> experimentalists all the time in my work. The good
> ones will discuss their work in detail - especially
> the tests they did to rule out certain complications.
RR: Except for your letter, and Lott's phone calls, I'm not
sure that Kellermann HAS refused to discuss it. Or perhaps
he's just one of those people who likes to have his work
speak for itself. Or perhaps he's been hounded by gun
advocates so much, he's disinclined to answer a call from
another one. Let me tell you, with so-called 'colleagues'
like Kates and Shaeffer you don't need enemies. If someone
called you names enough times, you'd be disinclined to
respond as well!
SF> Kellerman sampled a small fraction of all homicides,
> namely 3 counties out of the entire United States, and
> studied two of them over a 60 month period and another
> over a 32 month period. His most extensive regressions
> used only 316 matched pairs out of a total of 430 homicides.
> The case samples were skewed in various ways with respect
> to the controls, and the way he "corrected" for these
> discrepancies is not clear to the reader. But most
> scandalously, he failed to demonstrate that gun in the
> home = homicide gun for the range of possible homicide
> circumstances.
RR: Most 'studies' are samples. That's not a valid
argument. There is no evidence that his 316 matched pairs
were skewed toward one view. And I think his correction for
the social variabilities of controls is quite clear (his
conditional logistic-regression analysis).
SF> True. The case that convinced me to buy a handgun
occurred
> outside my apartment. The two instances in which I had to defend
> myself with a gun were both outside the home. It is precisely
> this venue that most anti-gunners want to ban guns, however.
RR: Not me. I've said before that I'd have no problem with
CCW if the background checks were better and if we had
tougher laws on gun responsibility.
SF> If only we could coax Gary Kleck to post here. <wish>
> If you have any influence, Mr Lott??
RR: Sometimes Neil Schulman posts here and he's a friend of
Kleck's. Maybe he can get the 2.5 million DGU Man online
:-)
RAY
Steve Fischer wrote:
> >RR> And of course all this "criminal" stuff is irrelevant
> >because Kellermann factored out the criminal element and
> >STILL came up with guns being 2.7 times more likely to
> >increase the risk of homicide in the home.
>
SF> How'd he do that exactly? He never told us.
RR: Hmmm... Had your deceased lover ever used illicit
drugs? Any family member needed hospitalization because of a
fight in the home? And most telling of all: Has any
household member been arrested? If that doesn't factor out
the criminal element, I don't know what would.
SF> What
> were all those criminals doing with guns in the home in the
> first place?
RR: They weren't all criminals. That's why Kellermann was
able to factor out criminals and still arrive at a 2.7 added
risk of homicide in homes with guns.
David Friedman
In article <357A276C...@Interaccess.com>, RAY <Ki...@Interaccess.com> wrote:
>RR: I imagine (in other homes) that friends quite
>frequently bring their guns along. When Billie-Bob goes to
>Bubba's house, they probably sit up half the night playing
>with their guns. Were Billie-Bob to get riled at Bubba for
>burning the grits, it's quite possible that Billie-Bob would
>use his own gun to kill Bubba.
But you are now talking about consequences of Billie-Bob and Bubba's
attitude to guns, not consequences of Bubba's owning a gun. If Bubba
hadn't bought a gun, Billie-Bob would still have brought his over to show
it off. The relevant question is not "are people who share with their
friends (Billie-Bob in this hypothetical) an interest in guns likely to
get shot" but "does the decision to keep a gun in your house make you more
likely to be murdered?"
>RR: What about that carpet layer or refridgerator repairman
>that likes to do a little extra business at night by
>revisiting his customers' homes to burglarize? He might
>notice your NRA wallpaper and figure he ought to come back
>packin'
Or not to come back at all--which reduces the chance of a violent encounter.
>RR: Except for your letter, and Lott's phone calls, I'm not
>sure that Kellermann HAS refused to discuss it.
It sounded from the discussion here as though his prolonged refusal to
make his data available was a matter of public record, although I have not
checked it--if I remember the posts correctly, it resulted in
congressional pressure to get data collected in publicly funded research
made public. Presumably you can check whether that is correct. If it is,
don't you think that is suspicious?
> Or perhaps
>he's just one of those people who likes to have his work
>speak for itself.
Except that the work can't speak for itself if you don't release your
data, because people have no way of telling whether what you say about the
results of your regressions is true.
David Friedman
In article <35797C1A...@Interaccess.com>, RAY <Ki...@Interaccess.com>
wrote:
>> Consider two people, otherwise similar, save that A has been threatened
>> with murder by someone and B has been not. A, not surprisingly, obtains a
>> gun. Which of them is more likely to be a murder victim?
>
>RR: The NRA would say that "B" is more likely to be
>murdered, because "A" has had the 'foresight' to protect
>himself with an 'equalizer.'
I don't think so. They would say that if we knew A and B were otherwise
identical then B would be more likely to be murdered--but the point of my
argument is that the fact that B has gotten a gun and A has not is
evidence that they are not otherwise identical.
> Here's the Kellermann quote:
> "Third, it is possible that reverse caustion accounted
>for some of the association we observed between gun
>ownership and homicide (ie in a limited number of cases,
>people may have acquired a gun in response to a specific
>threat.) If the source of that threat subsequently caused
>the homicide, the link between guns in the home and homicide
>may be due at least in part to the failure of these weapons
>to provide adequate protection from the assailants."
A reasonable statement--except for "limited number" and "some of the
association." How does he limit it? How does he eliminate the possibility
that all of the association they observed was due to such situations.
>DF> Again, consider two people, otherwise similar, one of
>whom is a
>> professional criminal and the other not. Which is more likely to own a
>> gun? To be a murder victim?
>
>RR: Hmmm... the NRA would say that the non-criminal is more
>likely to own a gun. (200 million guns are out there.. there
>can't be THAT many criminals).
There don't have to be. If all criminals have guns and half the
non-criminals, with an average ownership of between 1 and 2 guns per
non-criminal, that can give you 200 million guns, and criminals twice as
likely as non-criminals to own a gun. It sounds as though you are
confusing the fraction of guns owned by criminals with the relative
probability that a criminal or non-criminal will own a gun.
> And of course all this "criminal" stuff is irrelevant
>because Kellermann factored out the criminal element and
>STILL came up with guns being 2.7 times more likely to
>increase the risk of homicide in the home.
How could he factor out the criminal element? He can factor out people who
have a criminal record (did he?), but that isn't the same thing.
>RR: Sigh... Just goes to show that even the most
>knowledgable among us still can suffer from human bias and
>hubris.
Bias, yes. Hubris, no--in this particular case (my judgement). It is hard
to treat people fairly when you are sure they are wrong, and that their
errors are being used to justify what you are sure are bad policies. And
after all, most of us get most of our judgements at second hand, since
nobody knows enough to have an informed judgement on everything based on
his own expertise.
David Friedman
In article <357A28DD...@Interaccess.com>, RAY <Ki...@Interaccess.com>
wrote:
>Steve Fischer wrote:
>
>> >RR> And of course all this "criminal" stuff is irrelevant
>> >because Kellermann factored out the criminal element and
>> >STILL came up with guns being 2.7 times more likely to
>> >increase the risk of homicide in the home.
>>
>SF> How'd he do that exactly? He never told us.
>
>RR: Hmmm... Had your deceased lover ever used illicit
>drugs? Any family member needed hospitalization because of a
>fight in the home? And most telling of all: Has any
>household member been arrested? If that doesn't factor out
>the criminal element, I don't know what would.
1. That doesn't factor out all of the criminal element. The proxy may be
lying--not unlikely if the murder victim was (say) a drug dealer. And
there are lots of crimes that do not result in the perpetrator being
arrested.
2. I don't know what would either--that's the point. The methodology
Kellerman used has problems if:
a. There are characteristics that are likely to have a strong correlation
with both your independent and dependent variable (having a gun and being
murdered--the characteristics are being a criminal and/or having had your
life threatened), and ...
b. There is no adequate way of observing those characteristics in order to
control for them.
David Friedman
>>What Kellerman did was ignore his own findings. He wrote:
>>
>> "The ...characteristics of the victims and controls were similar
>>except that the case subjects were more likely to have rented their
>>homes (70.4% v 47.3%).. lived alone (26.8 v 11.9%) ...
>>
>> Alcohol was more commonly consumed by one or more members of
>>household of case subjects... (and)..by case subjects themselves....
>>Illegal drug-use by case subject....was also reported more commonly by
>>case households than control...
>>
>> Previous episodes of violence were reported more frequently by
>>..case households.. "
...
I don't think he is ignoring his finding. If I understand the study
correctly, he is controlling for these variables.
In other words, the question he is asking, is not "what fraction of the
murder victims and what fraction of the control group had a gun in the
home" but rather "after using a regression to measure the influence of
other variables on the chance of being killed, what was the remaining
influence of having a gun in the home?" I don't know if he did it
correctly, or if he did it honestly, but it seems pretty clear that that
is what he claimed to do.
jackson dryden
--
jackson
David Friedman <DD...@best.com> wrote in article
<DDFr-06069...@ddfr.vip.best.com>...
> In article <stratosE...@netcom.com>, str...@netcom.com (Steve
> Fischer) wrote:
>
>
> > I've always enjoyed watching people squirm when trying to
> >justify the "correctness" of a study using such a limited population
> >when compared with a study which encompasses the entire country.
> >If you want to study GUNS IN THE UNITED STATES, you can't have a
> >sample database better than the whole country.
>
> There may be many things wrong with Kellerman's study, but I don't think
> this is a valid complaint. He didn't have the data to replicate his study
> for the whole U.S., since he got it via questionaires. No doubt if
someone
> had offered him a billion dollars in funding ... .
David,
They tried to get it from the US government, that is to say *us*. They
wanted to get tax money to fund a campaign to justify their goal of
disarming the American people! One of the reasons that they failed is
because of Kellerman's sloppy efforts at cobbling up some anti-gun research
and his equally sloppy efforts at keeping the data secret, even from
congress! In fact the only reason that he provided any data at all was when
congress compelled him to. And that's why he don't work for the CDC
anymore.
jackson
jackson dryden
--
jackson
RAY <Ki...@Interaccess.com> wrote in article
<357A276C...@Interaccess.com>...
> Steve Fischer wrote:
> RR: I imagine (in other homes) that friends quite
> frequently bring their guns along. When Billie-Bob goes to
> Bubba's house, they probably sit up half the night playing
> with their guns. Were Billie-Bob to get riled at Bubba for
> burning the grits, it's quite possible that Billie-Bob would
> use his own gun to kill Bubba.
Not over grits. Get real!
>
> SF> I often get into spirited arguments with friends as
> part
> > of political or religious discussions, etc. It would never
> > cross my mind, no matter how angry I got, to bring a gun with
> > me because my friends also own guns (most of them do). The
> > idea of shooting someone over an argument would never enter
> > our minds. Of course, my friends are not spouse-beating,
> > drug-dealing ex-convicts.
>
> RR: I'll bet you would have invited the Hartmans over for
> tea, though. What if she had brought her gun with, and then
> went nuts? Friends and acquaintences can wind up doing the
> Strangest Things, don't you think?
If that had happened at my house, the second shot would have killed her. If
I couldn't take the shot, my wife would have and Mr. Hartman would be alive
today.
A much more likely scenario is that he will see the wall paper and the
handwriting on it! He will visit someone elses house that night.
>
> SF> Doesn't it bother you that Kellerman refused to
> > discuss it? Why would an honest academic refuse to
> > answer the criticisms of a colleague? I deal with
> > experimentalists all the time in my work. The good
> > ones will discuss their work in detail - especially
> > the tests they did to rule out certain complications.
>
> RR: Except for your letter, and Lott's phone calls, I'm not
> sure that Kellermann HAS refused to discuss it. Or perhaps
> he's just one of those people who likes to have his work
> speak for itself. Or perhaps he's been hounded by gun
> advocates so much, he's disinclined to answer a call from
> another one. Let me tell you, with so-called 'colleagues'
> like Kates and Shaeffer you don't need enemies. If someone
> called you names enough times, you'd be disinclined to
> respond as well!
If ya can't stand the heat, stay the hell out of the kitchen!
>
> RR: Not me. I've said before that I'd have no problem with
> CCW if the background checks were better and if we had
> tougher laws on gun responsibility.
Oh the laws are there now. If you do the crime you ought to do the time,
but unfortunately, our current judicial and corrections systems don't quite
see it that way.
>
jackson
RAY
"Crime, Deterrence, and Right to Carry Concealed Handguns"
6/26/96
I understand some published critiques of your study have
appeared, but I have not seen them. The questions I have
were scribbled into the margins while I was first reading
your study. Perhaps you can comment:
1. Apples and Oranges
The first thing up, you compare defensive gun use with
criminal gun use, and conclude "... defensive gun use on net
saved lives."
However, your comparison is between firearms deaths from
homicides and accidents versus both deaths and non-deaths in
the defensive gun category. Since we know that there are
far more non-death incidents with guns, haven't you skewed
the comparison to arrive at the conclusion you desire?
A more fair comparison, in my mind, would have been
between DGU deaths and criminal deaths, or between both
deaths and non-deaths for both categories. Since we have no
"lived saved" studies on defensive gun uses, we must use the
statistics we have on total incidents (both deaths and
non-deaths).
When we do this, we find that the NCVS gives us 82,000
incidents a year of total defensive gun uses. Using both
UCR and National Safety Council data, however, we find that
there are nearly one-million times a year that guns are used
in criminal, accidental or suicidal incidents. Thus, in
reality, we have more than a ten-to-one ratio of
harm-versus-good for gun usage.
(Kleck's study is omitted here because it's just too
suspect. If you want to defend him, then I would suggest a
whole new thread)
2. Hot Burglaries
You compare the rates of hot burglaries between the U.S.
and Canada/Britain, and attribute the difference to the
defensive nature of guns in the home. But the gun laws in
Canada are not the same as in Britain. In fact, there are
far more long guns in Canada per capita than in either
Britain or the U.S. I doubt a burglar would be particularly
happy to go up against a shotgun as opposed to a pistol.
And there may be other factors besides DGUs that prompt a
smaller percentage of hot burglaries in the U.S. You
haven't looked into those variables. I could just as easily
claim that Canada's tough handgun laws are working because
Canada has only 1/5 the murder rate of the U.S. But without
accounting for the variables, I would be making a mistake.
Also, I wonder if for Britain, you are using police
reports, or the British Crime Survey (BCS)? Police have a
tendency in Britain, I would think, to over-report
confrontational crime (where an eyewitness might provide a
better likelihood of solving the case), and under-report
crime where the offender is never seen (and the police have
less likelihood of solving it) ("The Economist" 10/15/94).
The BCS is more likely to be accurate, I would think, on the
question of hot vs. cold burglaries. Furthermore, the
rising cost of insurance premiums in Britain's inner cities
reportedly caused a drop in all reporting of crime (from
London School of Economics' Robert Reiner.. in "The
Economist" 10/15/94), but would most likely have caused more
of a drop for burglaries committed without any confrontation
with the criminal.
Next, you cite Wright/Rossi's convict fears of armed
victims. But Wright himself admitted a major flaw in that
work. He says he never asked the convicts if they were
talking about being afraid of law-abiding citizens or if
they were talking about their fears of going up against
other armed criminals. Since criminal on criminal crime is
common and more deadly, it's possible the Wright/Rossi
interviews are hopelessly irrelevant for your purpose.
On the positive side, I drew a smiley face next to your
citation of Archie Bunker.... :-)
3. A Criminal's Mind
You said ".. the expected penalty affects the prospective
criminal's desire to commit a crime." Yet I wonder if the
studies you cite incorporated 'heat of passion' crimes or
crazy-people crimes into their equations? I would think
that crazy persons, or those with temporary hatreds, would
be blinded to all thought of penalty (including the
extremely rare possibility that he'd get shot because the
victim might have a concealed carry weapons permit).
You also say ".. accused individuals appear to suffer
large reputional penalities simply from being arrested."
This may be true for otherwise law-abiding citizens, but in
many Gangs, and for non-gang career criminals, arrest is a
way of life, is it not?
How many times are these individuals arrested and then
released with no prosecution? Arrest rates may tell us
something about the nature of the person arrested, but not
necessarily about the rate of crime in the community. For
that, you need convictions.
And it appears that you have time-series arrest,
conviction and sentencing rates at the county level for only
three States. Isn't that a problem?
4. I also see you've obtained time-series data on the
number of right to carry permits for each county in only
three States. Isn't that also a problem?
5. Now here's a real layman's question for you: What is a
'county dummy?' Other than the local idiot, of course :-)
6. Despite passage of state CCW laws, a relatively small
percentage of the population avails themselves of the
permits. And you've mentioned the apparent anomaly of a big
effect for Rapes, although very few women got CCW permits.
From this, you perceive the effect is largely one of
perception in the criminal mind. He 'thinks' he's going to
get shot, even though the likelihood is something else. And
since the biggest change in perception is toward female
victims, you see a bigger benefit in CCW for women than for
men.
We've seen the same pattern, haven't we, in smaller
communities which emphasized gun training for women for
awhile? The citation eludes me, but perhaps it was Kleck or
Kates who mentioned well-publicised gun training for women
in Kansas City and perhaps Orlando. As I recall, rapes
went down there too, but only for awhile... because the
reality of self-defense did not match the perception of
defense in the criminal mind.
Couldn't this 'temporary' effect also be giving you your
CCW conclusions? When criminals realize that few people
actually are taking advantage of CCW laws, won't they resume
their more violent activities?
And furthermore, all government programs, in my mind,
tend toward atrophy or at least change over time. As the CCW
programs continue, I see several things happening. 1)
People will eventually realize that carrying a gun
everywhere is more a nuisance than a necessity (since most
people's chance of being a crime victim is still very
small). 2) States will grow lax in requirements for getting
a CCW permit, and more permit holders will be the type of
people who will commit crimes. or 3) So many people will
have permits that inevitably some will use their guns (even
with current requirements in tact) to commit well publicized
very horrible crimes, and the political landscape will blame
CCW, and perhaps also blame people whose scientific studies
have trumpeted CCW permits (g).
I've got plenty of other questions about the study...
but perhaps I'll reserve them until I see your book.
RAY
David Friedman wrote:
> >RR: I imagine (in other homes) that friends quite
> >frequently bring their guns along. When Billie-Bob goes to
> >Bubba's house, they probably sit up half the night playing
> >with their guns. Were Billie-Bob to get riled at Bubba for
> >burning the grits, it's quite possible that Billie-Bob would
> >use his own gun to kill Bubba.
>
DF> But you are now talking about consequences of Billie-Bob
and Bubba's
> attitude to guns, not consequences of Bubba's owning a gun. If Bubba
> hadn't bought a gun, Billie-Bob would still have brought his over to show
> it off. The relevant question is not "are people who share with their
> friends (Billie-Bob in this hypothetical) an interest in guns likely to
> get shot" but "does the decision to keep a gun in your house make you more
> likely to be murdered?"
RR: Attitude and ownership often go hand in hand. Those
who don't own guns are much more likely to have the attitude
AGAINST having their friends bring guns into their homes.
> >RR: What about that carpet layer or refridgerator
> >repairman that likes to do a little extra business
> >at night by revisiting his customers' homes to
> >burglarize? He might notice your NRA wallpaper
> >and figure he ought to come back packin'
>
DF> Or not to come back at all--which reduces the
>chance of a violent encounter.
RR: We'll never know, now will we? Unless, of course, the
NCVS includes an NRA question. :-)
> >RR: Except for your letter, and Lott's phone calls, I'm not
> >sure that Kellermann HAS refused to discuss it.
>
DF> It sounded from the discussion here as though his
prolonged refusal to
> make his data available was a matter of public record, although I have not
> checked it--if I remember the posts correctly, it resulted in
> congressional pressure to get data collected in publicly funded research
> made public. Presumably you can check whether that is correct. If it is,
> don't you think that is suspicious?
RR: Gunners would like us to believe that Congresscritter
pressure forced Kellermann to release his data, just because
a few NRA lackies asked the CDC for it in a Congressional
hearing. But there is no evidence that Kellermann acted
because of that particular request.
RR> > Or perhaps
> >he's just one of those people who likes to have his work
> >speak for itself.
>
DF> Except that the work can't speak for itself if you don't
release your
> data, because people have no way of telling whether what you say about the
> results of your regressions is true.
RR: Let's see. His data's been released now for about a
year, and no gunner has used it to discredit him. It looks
like Kellermann's published work COULD speak for itself.
RAY
David Friedman wrote:
> >> Consider two people, otherwise similar, save that A has been threatened
> >> with murder by someone and B has been not. A, not surprisingly, obtains a
> >> gun. Which of them is more likely to be a murder victim?
> >
> >RR: The NRA would say that "B" is more likely to be
> >murdered, because "A" has had the 'foresight' to protect
> >himself with an 'equalizer.'
>
DF> I don't think so. They would say that if we knew A and B
were otherwise
> identical then B would be more likely to be murdered--but the point of my
> argument is that the fact that B has gotten a gun and A has not is
> evidence that they are not otherwise identical.
RR: Of course not. The one who got the gun's been
brainwashed by the NRA! :-)
> > Here's the Kellermann quote:
> > "Third, it is possible that reverse caustion accounted
> >for some of the association we observed between gun
> >ownership and homicide (ie in a limited number of cases,
> >people may have acquired a gun in response to a specific
> >threat.) If the source of that threat subsequently caused
> >the homicide, the link between guns in the home and homicide
> >may be due at least in part to the failure of these weapons
> >to provide adequate protection from the assailants."
>
DF> A reasonable statement--except for "limited number" and
"some of the
> association." How does he limit it? How does he eliminate the possibility
> that all of the association they observed was due to such situations.
RR: Well, he's released his data, yet no gunner genius has
found an error in Kellermann's use of "limited number." So
I guess he didn't speak with forked tongue.
> >DF> Again, consider two people, otherwise similar, one of
> >whom is a
> >> professional criminal and the other not. Which is more likely to own a
> >> gun? To be a murder victim?
> >
> >RR: Hmmm... the NRA would say that the non-criminal is more
> >likely to own a gun. (200 million guns are out there.. there
> >can't be THAT many criminals).
>
DF> There don't have to be. If all criminals have guns and
half the
> non-criminals, with an average ownership of between 1 and 2 guns per
> non-criminal, that can give you 200 million guns, and criminals twice as
> likely as non-criminals to own a gun. It sounds as though you are
> confusing the fraction of guns owned by criminals with the relative
> probability that a criminal or non-criminal will own a gun.
RR: You assume that a majority of professional criminals
own guns. How do you define a 'professional criminal'
anyway? Bjerregaard & Lizotte "Gun Ownership and Gang
Membership" suggests that even the oldest Gang members have
guns for 'protection' at only a 36% rate. Anybody got the
Wright/Rossi prison survey handy?
Dan Z
In <DDFr-06069...@ddfr.vip.best.com> DD...@best.com (David
>>justify the "correctness" of a study using such a limited population
>>when compared with a study which encompasses the entire country.
>>If you want to study GUNS IN THE UNITED STATES, you can't have a
>>sample database better than the whole country.
>
think
>this is a valid complaint. He didn't have the data to replicate his
study
>for the whole U.S., since he got it via questionaires. No doubt if
someone
>had offered him a billion dollars in funding ... .
I'm afraid I don't understand your point. It seems that you are saying
that, since Kellermann couldn't afford to study a larger base, that
errors introduced by using a small base are therefore OK.
--
antispam address list, spammers listed will be added to other spammer's lists; poetic justice....
spid...@pompano.net , ab...@myownmail.com , 34...@pcpostal.com , ja...@post.omnitel.net , h...@clark.net ,
netcom...@ns1.gonepostal.com , bonvoya...@worldnet.att.net , force...@msn.com , mar...@aware.nl ,
eth...@spiritnetwork.com , lor...@global-homebiz.com , web...@azstarnet.com , play...@xoom.com ,
webm...@globaltekgrp.com , ad...@sysna.com , the...@biznessweb.net , dal...@hoststar.com ,
da...@linkus.com , sa...@123bizweb.com , webm...@123bizweb.com , re...@123bizweb.com , t...@wolsi.com ,
sysa...@titaniccollectibles.com , sun...@safari.net , ad...@pornacopia.com , sig...@mwci.net ,
rm0...@replinets-max.com , ever...@TheOffice.net , labl...@wekan.com , s...@ras.nexgate.net ,
dan...@aol.com , da...@linkus.com , webm...@nucredit.com , dom...@website-hostings.com , st...@woz.org ,
st...@woz.org , apry...@usa.net , hoo...@xoom.com , sco...@hothotmail.com , micr...@doctorpc.com ,
ke...@hudsonet.com , va...@clubvault.com , gw...@yahoo.com , rich...@webt.com , mic...@netparty.com ,
cha...@aol.com , mike...@aol.com , remo...@aol.com , nappm...@yahoo.com , webm...@imail4u.com ,
sysa...@imail4u.com , cmd...@www.netcom.com , cmd...@ix.netcom.com , cmd...@netcom.com ,
mbente...@hotmail.com ,
David Friedman
In article <357AA485...@Interaccess.com>, RAY <Ki...@Interaccess.com>
wrote:
>"Crime, Deterrence, and Right to Carry Concealed Handguns"
>by Lott and Mustard
>6/26/96
>
> I understand some published critiques of your study have
>appeared, but I have not seen them.
I'm not John, but may be able to help on this. I have a web page up with
links to several pages attacking the Lott/Mustard study, John Lott's
replies to two of them, and some things I have written on the controversy,
including a summary and critique of the one published critique I know of.
http://www.best.com/~ddfr/Lott_v_Teret/Lott_Mustard_Controversy.html
> The questions I have
>were scribbled into the margins while I was first reading
>your study. Perhaps you can comment:
>1. Apples and Oranges
> The first thing up, you compare defensive gun use with
>criminal gun use, and conclude "... defensive gun use on net
>saved lives."
> However, your comparison is between firearms deaths from
>homicides and accidents versus both deaths and non-deaths in
>the defensive gun category. Since we know that there are
>far more non-death incidents with guns, haven't you skewed
>the comparison to arrive at the conclusion you desire?
But the conclusion of the article was based on regressions, in which John
(among other things) estimated the effect of shall issue laws on
homicides--and found that the reduction in homicides was much larger than
the (negligable) increase in accidental deaths. That's a comparison of
deaths to deaths.
> A more fair comparison, in my mind, would have been
>between DGU deaths and criminal deaths, or between both
>deaths and non-deaths for both categories. Since we have no
>"lived saved" studies on defensive gun uses, we must use the
>statistics we have on total incidents (both deaths and
>non-deaths).
I don't follow this. We have statistics on total deaths--the murder rate,
and the rate of lethal gun accidents. What we want to know is not how many
criminal were killed by defensive gun use but how many murders failed to
occur because of defensive gun use (and how many extra ones did occur
because people had handguns around). A defensive gun use can prevent a
murder without killing anyone, and the fact that potential victims might
have handguns might deter a murder even with no use of the gun. After all,
if you were evaluating the use of traffic cops to enforce speed limits,
you wouldn't compare "number of tickets given out" to "number of
speeders," and conclude that if the former number was lower then traffic
tickets didn't work. The relevant test would be to see how the number of
speeders was affected by the number of tickets given out.
> When we do this, we find that the NCVS gives us 82,000
>incidents a year of total defensive gun uses. Using both
>UCR and National Safety Council data, however, we find that
>there are nearly one-million times a year that guns are used
>in criminal, accidental or suicidal incidents. Thus, in
>reality, we have more than a ten-to-one ratio of
>harm-versus-good for gun usage.
1. There seems to be a lot of disagreement over the number of incidents of
defensive gun use.
2. Suppose, however, that your figure is correct--the conclusion does not
follow. The relevant count is not "uses vs uses" but consequences of uses.
If, for example, the fact that some people carried concealed handguns
resulted in many potential crimes not occurring, because the potential
offenders thought it was too risky, that would show up in neither of your
numbers--yet would be highly relevant to the conclusion. John investigated
that question, at great length, by looking at how laws requiring the issue
of concealed carry permits affected crime rates, using county level data
and controlling for a wide variety of other variables that might affect
crmie rates.
Consider your numbers. Eliminate the accideents and suicides and suppose
you end up with ten times as many criminal uses as defensive use. Surely a
lot of criminals would be deterred by the belief that they had one chance
in ten of facing an armed victim. A priori we don't know how many--which
is what John's statistics are trying to determine.
>4. I also see you've obtained time-series data on the
>number of right to carry permits for each county in only
>three States. Isn't that also a problem?
Does "a problem" mean "would it be better if the data were available for
all states?" Obviously yes, but it isn't. Given that it was only available
for three, are you suggesting it would have been better for John not to
use it?
>5. Now here's a real layman's question for you: What is a
>'county dummy?' Other than the local idiot, of course :-)
A dummy variable, typically, is a variable that is 1 if some condition is
true, 0 if it is not.
Suppose we are doing a very large sample regression, and we suspect that
there are county specific affects--some counties have higher crime rates
for reasons not picked up in our control variables, reasons that persist
over time. We add an additional set of county dummies. Thus there is a
"cook county dummy variable" in the regression which is 1 for cook county,
0 for all other counties, and a similar variable for every other county.
The coefficient of the Cook County dummy gives us the county-specific
effect for Cook, and similarly for each other county. The regression is
fitting the data, allowing one extra constant term for each country to
represents any special characteristics of that county that are persistent
over time and not picked up by the other variables.
>6. Despite passage of state CCW laws, a relatively small
>percentage of the population avails themselves of the
>permits. And you've mentioned the apparent anomaly of a big
>effect for Rapes, although very few women got CCW permits.
>From this, you perceive the effect is largely one of
>perception in the criminal mind. He 'thinks' he's going to
>get shot, even though the likelihood is something else.
I think you are misunderstanding the argument. John isn't assuming that
criminals overestimate the odds--merely that a fairly low probability of
being shot is still too high a price to pay for the benefits the criminal
expects from his crime. If only one potential victim in twenty is armed,
that still makes mugging a very unprofitable business.
Steve Fischer
In article <DDFr-06069...@ddfr.vip.best.com> DD...@best.com (David Friedman) writes:
>In article <stratosE...@netcom.com>, str...@netcom.com (Steve
>Fischer) wrote:
>
>
>>justify the "correctness" of a study using such a limited population
>>when compared with a study which encompasses the entire country.
>>If you want to study GUNS IN THE UNITED STATES, you can't have a
>>sample database better than the whole country.
>There may be many things wrong with Kellerman's study, but I don't think
>this is a valid complaint. He didn't have the data to replicate his study
>for the whole U.S., since he got it via questionaires. No doubt if someone
>had offered him a billion dollars in funding ... .
I don't criticize it in terms of research out per dollars spent.
What I dislike is what he's done with the result in terms of making
far too much out of too little. He's done great damage to honest
gun owners by lending his name to the worst anti-gun elements in
society. This paper and his "43/1" paper are MOST QUOTED papers
in the anti-gun literature. If he disagrees with the way his work
is being "used" then let him call the newspapers and take HCI and
the Violence Policy Center to task. The fact that he has chosen not
to proves to me that he is simply part of the plan to ban guns.
Steve
David Friedman
In article <357AA845...@Interaccess.com>, RAY <Ki...@Interaccess.com>
wrote:
>David Friedman wrote:
>> >RR: What about that carpet layer or refridgerator
>> >repairman that likes to do a little extra business
>> >at night by revisiting his customers' homes to
>> >burglarize? He might notice your NRA wallpaper
>> >and figure he ought to come back packin'
>>
>DF> Or not to come back at all--which reduces the
>>chance of a violent encounter.
>
>RR: We'll never know, now will we? Unless, of course, the
>NCVS includes an NRA question. :-)
Or unless someone does the equivalent of the Lott and Mustard study for
guns in the home rather than concealed carry.
In any case, "we'll never know" is equivalent to "Kellerman's conclusion
cannot be relied on," no?
>> >RR: Except for your letter, and Lott's phone calls, I'm not
>> >sure that Kellermann HAS refused to discuss it.
>DF> It sounded from the discussion here as though his
>prolonged refusal to
>> make his data available was a matter of public record, although I have not
>> checked it--if I remember the posts correctly, it resulted in
>> congressional pressure to get data collected in publicly funded research
>> made public. Presumably you can check whether that is correct. If it is,
>> don't you think that is suspicious?
>
>RR: Gunners would like us to believe that Congresscritter
>pressure forced Kellermann to release his data, just because
>a few NRA lackies asked the CDC for it in a Congressional
>hearing. But there is no evidence that Kellermann acted
>because of that particular request.
Is it a fact that he refused to release his data for several
years--whatever the reason he eventually changed his mind? If so, that is
evidence--not proof, but evidence--of dishonesty. A researcher who has
honestly described the implications of his data should be willing to make
it available to others, so that they can check his results. Replicability
is a standard criterion for statistical studies.
>RR: Let's see. His data's been released now for about a
>year, and no gunner has used it to discredit him. It looks
>like Kellermann's published work COULD speak for itself.
I thought Kleck had used some of Kellerman's data to discredit
him--whether successfully or not is one of the things people are arguing
about. In any case, have you looked at the data to determine how complete
what Kellerman released is?
I still don't see what your explanation is for the long refusal to release
the data.
Steve Fischer
>wrote:
>
>
>>don't you think these factors might just have something to
>>do with the 'risk of being attacked?' Don't you think that
>>people whose environment contains high scores in these
>>indices are more likely to become both perpetrators and
>>victims of such attacks? And yet you say that Kellermann
>>has not taken such risks into account?
>
>whether people are likely to be attacked, but whether they give us
>complete information.
>
>Suppose you have two people who get the same scores on all those factors.
>You discover an additional fact--one of them has purchased a gun. Don't
>you think you will revise upward your estimate both that the is a criminal
>and that he has reason to fear for his life? If so, Kellerman's method
>will show a false positive even if there is really no causal relation
>between gun ownership and being killed.
>
>>randomized. Why do you insist that it should have been?
>
>randomized, or are confident that you have picked up all important factors
>that are causal for both your dependent and independent variables. It does
>not sound as though Kellerman did, or could do, the latter.
>
>I am curious, by the way, since you are defending Kellerman, what your
>view is of the issue of releasing data. The assertion we have seen here,
>and have not yet seen contradicted, is that Kellerman kept his data secret
>for several years, and only released it under pressure. Do you have
>evidence that either half of that is false? If not, would you take such
>behavior as in itself suspicious?
I know Ray. I've been reading his stuff for some time. The only
people whose work he questions are those whose research lessens
the link between guns and violence.
He demands the highest standards of proof against people like
Kellerman, yet accepts the flimsiest excuses to criticize
people like Lott. It took him FOREVER to quit talking about
the Olin connection. Point out to him that Stephen Teret is
funded by the Joyce Foundation of Chicago - a self-stated
left wing think tank, and he pooh-poohs the idea. He is the
worst opponent one can face - a cunning, intelligent hypocrite.
David Friedman
In article <357AACF7...@Interaccess.com>, RAY <Ki...@Interaccess.com>
wrote:
>David Friedman wrote:
>DF> A reasonable statement--except for "limited number" and
>"some of the
>> association." How does he limit it? How does he eliminate the possibility
>> that all of the association they observed was due to such situations.
>
>RR: Well, he's released his data, yet no gunner genius has
>found an error in Kellermann's use of "limited number." So
>I guess he didn't speak with forked tongue.
I don't follow that. How could his data tell him that only part of the
observed association was explained by the unobserved variables? My point
isn't about what the numbers are, but about what the inferences are.
>> >RR: Hmmm... the NRA would say that the non-criminal is more
>> >likely to own a gun. (200 million guns are out there.. there
>> >can't be THAT many criminals).
>>
>DF> There don't have to be. If all criminals have guns and
>half the
>> non-criminals, with an average ownership of between 1 and 2 guns per
>> non-criminal, that can give you 200 million guns, and criminals twice as
>> likely as non-criminals to own a gun. It sounds as though you are
>> confusing the fraction of guns owned by criminals with the relative
>> probability that a criminal or non-criminal will own a gun.
>
>RR: You assume that a majority of professional criminals
>own guns. How do you define a 'professional criminal'
>anyway? Bjerregaard & Lizotte "Gun Ownership and Gang
>Membership" suggests that even the oldest Gang members have
>guns for 'protection' at only a 36% rate. Anybody got the
>Wright/Rossi prison survey handy?
No--I made no assumption about the real world rate, since I don't have the
data. I merely pointed out that your conclusion didn't follow from your
facts.
David Friedman
In article <6ledn2$4...@sjx-ixn4.ix.netcom.com>, dan...@ix.netcom.com(Dan
Z) wrote:
>In <DDFr-06069...@ddfr.vip.best.com> DD...@best.com (David
>Friedman) writes:
>>
>>In article <stratosE...@netcom.com>, str...@netcom.com (Steve
>>Fischer) wrote:
>>
>>
>>> I've always enjoyed watching people squirm when trying to
>>>justify the "correctness" of a study using such a limited population
>>>when compared with a study which encompasses the entire country.
>>>If you want to study GUNS IN THE UNITED STATES, you can't have a
>>>sample database better than the whole country.
>>
>>There may be many things wrong with Kellerman's study, but I don't
>think
>>this is a valid complaint. He didn't have the data to replicate his
>study
>>for the whole U.S., since he got it via questionaires. No doubt if
>someone
>>had offered him a billion dollars in funding ... .
>>--
>
>I'm afraid I don't understand your point. It seems that you are saying
>that, since Kellermann couldn't afford to study a larger base, that
>errors introduced by using a small base are therefore OK.
I am saying that his use of a small base is not "incorrect"--it might be
the best available way of doing the study. Of course we would have more
confidence in his results if he had had more data--but in the real world,
one frequently has to do the best one can with limited amounts of
information.
Steve Fischer
You're not addressing the issue I criticized. You suggested
that because one person has a gun in the home, that a visiting
friend would "feel the need." I suggest that this was absurd.
FRIENDS don't think of each other that way. Enemies do. Your
answer did not address that issue. Rather, you shifted focus
back to the other part of your argument, which I address below.
>SF> I often get into spirited arguments with friends as part
>> of political or religious discussions, etc. It would never
>> cross my mind, no matter how angry I got, to bring a gun with
>> me because my friends also own guns (most of them do). The
>> idea of shooting someone over an argument would never enter
>> our minds. Of course, my friends are not spouse-beating,
>> drug-dealing ex-convicts.
>RR: I'll bet you would have invited the Hartmans over for
>tea, though. What if she had brought her gun with, and then
>went nuts? Friends and acquaintences can wind up doing the
>Strangest Things, don't you think?
What if, what if, what if? That isn't the way it played
itself out. From what I've read she planned this at least a
day in advance (she told a friend) and was alone with her family
and under the influence of mind-altering drugs when she did it.
>SF> Finally, in the immediate heat of a killing rage, the idea
>> that taking the final step of taking a life would depend upon
>> whether the resident had a gun in another room is laughable.
>> If you're mad enough to kill, you're going to kill right then
>> and there regardless of other circumstances.
>RR: There are, perhaps, five second rages and five hour
>rages (The Chicago Bear's Alonzo Spellman comes to mind).
>No one would be safe in either case with a gun anywhere in
>the home.
I have only one question for you. Was he drunk/drinking
or under the influence of other drugs (anabolic steroids in
particular, which make you mean) the whole time? Bet he was.
Picking a burly, agressive athlete whose job is basically
organized mayhem, is not impressive if you're trying to make
a general case.
>RR> > Furthermore, should an enemy wish to do you harm, or even
>> >wish to engage you in what may become an ugly argument, and
>> >should that enemy have knowledge that you own a gun, he/she
>> >may be more inclined to bring a gun themselves for
>> >protection. In the course of the fight, that gun from
>> >outside may be the murder weapon, although it was the
>> >homeowner's possession of a gun which resulted in the murder
>> >weapon originally being brought into the home.
>>
>SF> Now you're talking about people who are likely to
>be
>> criminals, not your ordinary whitebread, middle-class,
>> suburbanites. Drug dealers are good examples, as I mentioned
>> above. They know that other drug dealers carry weapons and
>> have large wads of cash around. People like me buy guns to
>> protect myself from people like that when we meet them on the
>> street. I don't visit their homes.
>
>RR: What about that carpet layer or refridgerator repairman
>that likes to do a little extra business at night by
>revisiting his customers' homes to burglarize? He might
>notice your NRA wallpaper and figure he ought to come back
>packin'. Then you confront him, and of course he gets the draw
>on you, and it's good bye Steve! If he hadn't thought you
>had a gun, perhaps he would have snuck in unarmed, and run
>away when you woke up. That's just one example. I'm sure
>there are plenty of others.
He won't get the drop on me. I won't be precise, but
let me say there is only ONE way to get to me, and in order
to do that you have to get by a device that produces loud,
piercing noises. Not only that, but there are a number of
large obstacles between me and an intruder which I can
navigate in the dark while he can't. I keep a loaded gun
next to my bed once I've gone to sleep. I thank you for
your concern, but I'm pretty safe, Robert.
>SF> Doesn't it bother you that Kellerman refused to
>> discuss it? Why would an honest academic refuse to
>> answer the criticisms of a colleague? I deal with
>> experimentalists all the time in my work. The good
>> ones will discuss their work in detail - especially
>> the tests they did to rule out certain complications.
>RR: Except for your letter, and Lott's phone calls, I'm not
>sure that Kellermann HAS refused to discuss it. Or perhaps
>he's just one of those people who likes to have his work
>speak for itself.
In that case, he's not much of an academic. Discussion
is part of the process of scientific inquiry.
>Or perhaps he's been hounded by gun
>advocates so much, he's disinclined to answer a call from
>another one.
I would believe that if he got calls from someone whose
name he didn't know - a crackpot whose language makes it clear
he's just blowing off steam. He knew Lott was working on a book
yet refused to give him data. Not very helpful, in scientific
terms.
>Let me tell you, with so-called 'colleagues'
>like Kates and Shaeffer you don't need enemies. If someone
>called you names enough times, you'd be disinclined to
>respond as well!
He has it coming. He's let anti-gun activists like
Sarah Brady and Josh Sugarmann uncritically use his work.
These people call people much nastier names than Kates and
Schaeffer. If Kellerman thinks too much is being made out
of his work, given the uncertainties, then it is his respon-
sibility to refute any extraordinary claims being made for
it in a public forum, otherwise he's just part of their
anti-gun movement.
>SF> Kellerman sampled a small fraction of all homicides,
>> namely 3 counties out of the entire United States, and
>> studied two of them over a 60 month period and another
>> over a 32 month period. His most extensive regressions
>> used only 316 matched pairs out of a total of 430 homicides.
>> The case samples were skewed in various ways with respect
>> to the controls, and the way he "corrected" for these
>> discrepancies is not clear to the reader. But most
>> scandalously, he failed to demonstrate that gun in the
>> home = homicide gun for the range of possible homicide
>> circumstances.
>
>RR: Most 'studies' are samples. That's not a valid
>argument.
It is when you're comparing a sample (Kellerman) to
a comprehensive piece of work (Lott).
>There is no evidence that his 316 matched pairs
>were skewed toward one view. And I think his correction for
>the social variabilities of controls is quite clear (his
>conditional logistic-regression analysis).
Oh yeah. Pick a few numbers and show me. Oops, no
raw data.
>SF> True. The case that convinced me to buy a handgun occurred
>> outside my apartment. The two instances in which I had to defend
>> myself with a gun were both outside the home. It is precisely
>> this venue that most anti-gunners want to ban guns, however.
>RR: Not me. I've said before that I'd have no problem with
>CCW if the background checks were better and if we had
>tougher laws on gun responsibility.
Cool so long as they are "shall issue" type permits and not
at the "discretion" of law enforcement officers. There should be
a very clear set of criteria (felony conviction, mental instability,
etc) used to deny a permit. "Need" is irrelevant. Whenever need
is used, it turns out that friends of the CLEO seem to ge all the
permits.
>SF> If only we could coax Gary Kleck to post here. <wish>
>> If you have any influence, Mr Lott??
>RR: Sometimes Neil Schulman posts here and he's a friend of
>Kleck's. Maybe he can get the 2.5 million DGU Man online
>:-)
As far as I am concerned it can only enhance the debate.
I have a number of specific questions I'd like to ask him.
Steve
Steve Fischer
In article <357A28DD...@Interaccess.com> RAY <Ki...@Interaccess.com> writes:
>Steve Fischer wrote:
>
>> >because Kellermann factored out the criminal element and
>> >STILL came up with guns being 2.7 times more likely to
>> >increase the risk of homicide in the home.
>>
>SF> How'd he do that exactly? He never told us.
>
>RR: Hmmm... Had your deceased lover ever used illicit
>drugs? Any family member needed hospitalization because of a
>fight in the home? And most telling of all: Has any
>household member been arrested? If that doesn't factor out
>the criminal element, I don't know what would.
How does it "factor out?" He didn't simply drop the
case from what I can see.
>SF> What
>> were all those criminals doing with guns in the home in the
>> first place?
>RR: They weren't all criminals. That's why Kellermann was
>able to factor out criminals and still arrive at a 2.7 added
>risk of homicide in homes with guns.
I sure would like to see the details of that calculation.
Like my teaches used to say on tests: "Be sure to show your
work."
Sam A. Kersh
DD...@best.com (David Friedman) wrote:
That's what Prof Kellermann claimed. But, as Prof John Lott has pointed
out, Kellermann to this day has refused to release ALL the data.
Selected parts are available only. The question of "What is he
hiding?" is valid.
Sam A. Kersh
NRA Life Member
TSRA Life Member
L.E.A.A., JPFO
http://www.flash.net/~csmkersh/csmkersh.htm
=======================================================
Read the book UNINTENDED CONSEQUENCES.
Reviews are at
http://www.amazon.com/exec/obidos/ISBN%3D1888118040
or
http://www.nguworld.com/vindex/95/123195vs.htm
Sam A. Kersh
RAY <Ki...@Interaccess.com> wrote:
>RR: I imagine (in other homes) that friends quite
>frequently bring their guns along. When Billie-Bob goes to
>Bubba's house, they probably sit up half the night playing
>with their guns. Were Billie-Bob to get riled at Bubba for
>burning the grits, it's quite possible that Billie-Bob would
>use his own gun to kill Bubba.
That's your problem... You imagine too much and reject facts... Your
bigoted attitude toward Southerners is showing again...
Rather than asking Prof Lott how is became interested; why don't you
explain how you became interested and just how germane your background
is to the discussion....
Sam A. Kersh
str...@netcom.com (Steve Fischer) wrote:
>In article <35797D78...@Interaccess.com> RAY <Ki...@Interaccess.com> writes:
>
>> Kellermann has factored out criminal intent, and
>>commented on the protective purchase of guns. What other
>>'rational choices' are you interested in?
>
> How do you factor out criminal intent? Was he a mind reader?
No, but Ray considers himself to be a mind-reader...
Sam A. Kersh
RAY <Ki...@Interaccess.com> wrote:
>Steve Fischer wrote:
>
>> >RR> And of course all this "criminal" stuff is irrelevant
>> >because Kellermann factored out the criminal element and
>> >STILL came up with guns being 2.7 times more likely to
>> >increase the risk of homicide in the home.
>>
>SF> How'd he do that exactly? He never told us.
>
>RR: Hmmm... Had your deceased lover ever used illicit
>drugs? Any family member needed hospitalization because of a
>fight in the home? And most telling of all: Has any
>household member been arrested? If that doesn't factor out
>the criminal element, I don't know what would.
>
When he still uses that case in the study?? You really need to take
another look at Tables 3 & 4..... Remember he started with 444 cases.
By table 3, his study group was down to 388 matched pairs... Table 4
316...
But from Table 3 we find
if removed from 444
case drank alcohol 238 206
case used drugs 74 370
non-case arrested 193 251
case arrested 132 312
It is obvious these "problem people" were used in his analysis... You
can't factor out behavior if the actor himself is used...
>SF> What
>> were all those criminals doing with guns in the home in the
>> first place?
>
>RR: They weren't all criminals. That's why Kellermann was
>able to factor out criminals and still arrive at a 2.7 added
>risk of homicide in homes with guns.
Cite from his data the number of criminals and non-criminals....
Also cite from his data the number of household firearms actually used
in the homicides..... Bet you can't verify more than 8.
RAY
Steve Fischer wrote:
> I know Ray. I've been reading his stuff for some time. The only
> people whose work he questions are those whose research lessens
> the link between guns and violence.
> He demands the highest standards of proof against people like
> Kellerman, yet accepts the flimsiest excuses to criticize
> people like Lott. It took him FOREVER to quit talking about
> the Olin connection. Point out to him that Stephen Teret is
> funded by the Joyce Foundation of Chicago - a self-stated
> left wing think tank, and he pooh-poohs the idea. He is the
> worst opponent one can face - a cunning, intelligent hypocrite.
RR: Your stupid name calling is what gives tpg a flaming
bad name. The fact is that I have NEVER brought up the
"Olin connection." Get your facts straight, Fischer, or
stay out of the way.
RAY
Steve Fischer wrote:
RR> >> > For instance, those who own guns may be more
amenable to
> >> >their friends, family or acquaintences bringing their own
> >> >guns into the home. Should a deadly argument ensue, the
> >> >friend's gun might have been used. Had the homeowner not
> >> >owned a gun, friends would be less likely to bring one with
> >> >them for a visit, and such a shooting would not occur.
> >>
> >SF> You're grasping at straws now, Robert. Let's take the last,
> >> most preposterous part of that statement first. When
> >> people feel they have to bring a gun with them when they
> >> visit, these people are not called friends, they're called
> >> ENEMIES. Drug dealers approach each other this way - neither
> >> trusting the other.
>
> >RR: I imagine (in other homes) that friends quite
> >frequently bring their guns along. When Billie-Bob goes to
> >Bubba's house, they probably sit up half the night playing
> >with their guns. Were Billie-Bob to get riled at Bubba for
> >burning the grits, it's quite possible that Billie-Bob would
> >use his own gun to kill Bubba.
>
SF> You're not addressing the issue I criticized. You
suggested
> that because one person has a gun in the home, that a visiting
> friend would "feel the need." I suggest that this was absurd.
> FRIENDS don't think of each other that way. Enemies do. Your
> answer did not address that issue. Rather, you shifted focus
> back to the other part of your argument, which I address below.
RR: No shift at all. I addressed both scenarios in my
original post. It is you who mixed them together. To
refresh.. one scenario is the above Billie-Bob Bubba
situation, where friends gather to show their guns to each
other, then have a fatal falling out.
The other scenario I mentioned was if an acquaintance
came to your home meaning to do you harm, or have an
arguement.. and brought a gun for self-defense because he
knew that you owned a gun.
> >RR: What about that carpet layer or refridgerator repairman
> >that likes to do a little extra business at night by
> >revisiting his customers' homes to burglarize? He might
> >notice your NRA wallpaper and figure he ought to come back
> >packin'. Then you confront him, and of course he gets the draw
> >on you, and it's good bye Steve! If he hadn't thought you
> >had a gun, perhaps he would have snuck in unarmed, and run
> >away when you woke up. That's just one example. I'm sure
> >there are plenty of others.
>
SF> He won't get the drop on me. I won't be precise, but
> let me say there is only ONE way to get to me, and in order
> to do that you have to get by a device that produces loud,
> piercing noises. Not only that, but there are a number of
> large obstacles between me and an intruder which I can
> navigate in the dark while he can't. I keep a loaded gun
> next to my bed once I've gone to sleep. I thank you for
> your concern, but I'm pretty safe, Robert.
RR: Nobody's perfect, and a gun in the home just raises the
stakes that when you do mess up, the result will be far more
serious than without a gun.
> >RR: Except for your letter, and Lott's phone calls, I'm not
> >sure that Kellermann HAS refused to discuss it. Or perhaps
> >he's just one of those people who likes to have his work
> >speak for itself.
>
SF> In that case, he's not much of an academic. Discussion
> is part of the process of scientific inquiry.
RR: Discussion yes. Namecalling, no.
RR> >Or perhaps he's been hounded by gun
> >advocates so much, he's disinclined to answer a call from
> >another one.
>
SF> I would believe that if he got calls from someone
whose
> name he didn't know - a crackpot whose language makes it clear
> he's just blowing off steam. He knew Lott was working on a book
> yet refused to give him data. Not very helpful, in scientific
> terms.
RR: Kates and Schaeffer were working on a Tenn. Law Review
article. You see how they slandered Kellermann and others
in
that.
SF> He has it coming. He's let anti-gun activists like
> Sarah Brady and Josh Sugarmann uncritically use his work.
> These people call people much nastier names than Kates and
> Schaeffer.
RR: Kellermann never called his opponants nasty names.. at
least that I know of in print. Funny that in one breath you
say that academics should be nice to each other.. and in the
other breath you say that Kellermann "has it coming."
> >RR: Not me. I've said before that I'd have no problem with
> >CCW if the background checks were better and if we had
> >tougher laws on gun responsibility.
>
SF> Cool so long as they are "shall issue" type permits
and not
> at the "discretion" of law enforcement officers. There should be
> a very clear set of criteria (felony conviction, mental instability,
> etc) used to deny a permit. "Need" is irrelevant. Whenever need
> is used, it turns out that friends of the CLEO seem to ge all the
> permits.
RR: I agree with your parameters, but your criteria need
some
firming up.
> >SF> If only we could coax Gary Kleck to post here. <wish>
> >> If you have any influence, Mr Lott??
>
> >RR: Sometimes Neil Schulman posts here and he's a friend of
> >Kleck's. Maybe he can get the 2.5 million DGU Man online
> >:-)
>
SF> As far as I am concerned it can only enhance the
debate.
> I have a number of specific questions I'd like to ask him.
RR: So Gary Kleck won't talk to you also?
Oracle
--
xxxxxxxxxxxxxxx
HANDGUN CONTROL
xxxxxxxxxxxxxxx
xxxxxxxxxxxxxxx
Steve Fischer <str...@netcom.com> wrote in article
<stratosE...@netcom.com>...
> In article <01bd90e6$5db89e60$188653d1@xbwerrsn> "Oracle"
<aa1...@tseinc.com> writes:
> >Kellermann is not my "favorite expert." I know little of his
> >work. In fact, because I'm not conversant in the details of
any
> >statistical analysis, I really don't have a favorite.
Whatever
> >opinion I've developed of the validity of any such work is
based
> >solely on the reputation and credentials of its author.
>
> I seriously doubt that you know anything about the
"credentials"
> of any of the researchers involved here. It was just a day or
two
> ago that John Lott was kind enough to allow us to look at HIS
> Curriculum Vitae?
I know enought about them to develop a reasonable opinion.
Lott's work is tainted by his association with "Law and
Economics" and the Olin Foundation.
In certain professions, a candiate for certification must
demonstrate a familiarity with the ethics of the profession
before it will be granted. Included among them, for good
reason, is a caution to avoid the appearance of impropriety.
When that appearance exists, professional opinions are suspect
because we, the general public, can't be sure if they are
intended to serve the improper relationship or "truth." Taking
money from an organization which clearly has a political agenda
and which may benefit directly from the work being funded
clearly is a substantial reason to question the results of the
work.
James F. Mayer
In <357B1D57...@Interaccess.com> RAY <Ki...@Interaccess.com>
writes:
Ray can't stand the truth that contradicts his assertions or that
he has to admit that he is wrong. He can't stand that certain people
have found out that he has also written a post that supports neo-nazi
historical revisionism.
Did someone hit a raw nerve, Ray?
David Friedman
In article <357B5D92...@Interaccess.com>, RAY <Ki...@Interaccess.com>
wrote:
>Steve Fischer wrote:
>SF> I would believe that if he got calls from someone
>whose
>> name he didn't know - a crackpot whose language makes it clear
>> he's just blowing off steam. He knew Lott was working on a book
>> yet refused to give him data. Not very helpful, in scientific
>> terms.
>
>RR: Kates and Schaeffer were working on a Tenn. Law Review
>article. You see how they slandered Kellermann and others
>in
>that.
Black and Nagin were working on a JLS article; you can see how they
attacked Lott in that (for details, see my discussion at
http://www.best.com/~ddfr/Lott_v_Teret/Friedman_on_B_and_N.html). Lott
gave them his data. Do you really thing that a policy of "refuse to talk
with people, or give them your data, if you think they are going to end up
attacking you" is proper behavior for a scientist.
David Friedman
In article <6lfmo9$b...@sjx-ixn6.ix.netcom.com>, jf...@ix.netcom.com(James
F. Mayer) wrote:
>In <357B1D57...@Interaccess.com> RAY <Ki...@Interaccess.com>
>writes:
>>
>>Steve Fischer wrote:
>>
>>> I know Ray. I've been reading his stuff for some time. The only
>>> people whose work he questions are those whose research lessens
>>> the link between guns and violence.
>>> He demands the highest standards of proof against people like
>>> Kellerman, yet accepts the flimsiest excuses to criticize
>>> people like Lott. It took him FOREVER to quit talking about
>>> the Olin connection. Point out to him that Stephen Teret is
>>> funded by the Joyce Foundation of Chicago - a self-stated
>>> left wing think tank, and he pooh-poohs the idea. He is the
>>> worst opponent one can face - a cunning, intelligent hypocrite.
>>
>>RR: Your stupid name calling is what gives tpg a flaming
>>bad name. The fact is that I have NEVER brought up the
>>"Olin connection." Get your facts straight, Fischer, or
>>stay out of the way.
>
> Ray can't stand the truth that contradicts his assertions or that
>he has to admit that he is wrong.
I have just done an Altavista search, and am unable to find any post where
Ray asserts a connection between Lott and Winchester. Can you quote one?
julia.c...@iint.com
In article <mcculloch-060...@dial-47-96.ots.utexas.edu>,
mccu...@mail.utexas.edu (Jim McCulloch) wrote:
>
> In article <6l9na6$r9b$1...@nnrp1.dejanews.com>, julia.c...@iint.com wrote:
>
> > Okay, this is where you've got a problem in your reasoning.
> > The object of scientific research is to prove things. Your study
> > only proves things if your assumptions are statistically valid.
> > If you presume a population is "random" in your study, and someone
> > criticizing your study points out that that sample "may" not be
> > random, whammo, all your statistics become invalid
>
> That's nonsense. The criticism has to be, well, plausible, to perform
> this magic. The tobacco industry for years could think of a multitide of
> *possible* weaknesses, including possible sampling non-randomness --which
> however were only *remotely* probable-- in the various studies indicating
> links between smoking and several illnesses, and they did indeed hope that
> the magic you mention, "whammo, all your statistics become invalid", would
> occur. But it didn't.
>
> My question to Mr. Lott had to do with the plausibility of his criticism
> concerning nonrandomness. It seems to me from what I have read that many
> gun owners do not buy guns because they are objectively in danger.
I believe I mentioned plausibility or something of the sort in some of
that "more of the same"---In any case, Dr. Lott has very kindly expanded
upon his original remarks and referred you to his book which I also need
to go order, myself.
It does not matter if "many" gun owners do not buy guns because of some
danger. If the percentage of those in danger who buy guns is greater than
the percentage of those not in danger who buy guns, then your underlying
sample population is skewed and you have to find some way of quantifying
that and compensating for it or you don't get accurate results.
It looks as though Kellerman not only didn't address important confounding
factors in his study, but is going out of his way to avoid addressing
them now. He's not demonstrating very much of that scientific integrity
they taught me about in school, where you do well-designed research to
find things out, and accept the results even if it means you sometimes
come up empty in terms of those results not supporting your pet theory.
As one of my professors said about chemicals in a test tube, what it does
do is a lot more important than what it's supposed to do.
Julie
-----== Posted via Deja News, The Leader in Internet Discussion ==-----
http://www.dejanews.com/ Now offering spam-free web-based newsreading
julia.c...@iint.com
In article <stratosE...@netcom.com>,
str...@netcom.com (Steve Fischer) wrote:
>
> In article <mcculloch-060...@dial-47-96.ots.utexas.edu>
mccu...@mail.utexas.edu (Jim McCulloch) writes:
> >In article <6l9na6$r9b$1...@nnrp1.dejanews.com>, julia.c...@iint.com wrote:
> >
> >> Okay, this is where you've got a problem in your reasoning.
> >> The object of scientific research is to prove things. Your study
> >> only proves things if your assumptions are statistically valid.
> >> If you presume a population is "random" in your study, and someone
> >> criticizing your study points out that that sample "may" not be
> >> random, whammo, all your statistics become invalid
> >
> >That's nonsense. The criticism has to be, well, plausible, to perform
> >this magic. The tobacco industry for years could think of a multitide of
> >*possible* weaknesses, including possible sampling non-randomness --which
> >however were only *remotely* probable-- in the various studies indicating
> >links between smoking and several illnesses, and they did indeed hope that
> >the magic you mention, "whammo, all your statistics become invalid", would
> >occur. But it didn't.
> >
> >My question to Mr. Lott had to do with the plausibility of his criticism
> >concerning nonrandomness. It seems to me from what I have read that many
> >gun owners do not buy guns because they are objectively in danger.
>
> >On the other hand, it *may not*. I was bringing to his attention some
> >possible evidence against the plausibility of his claim. He may have an
> >answer. I don't know. If he doesn't, and if this potential nonrandomness
> >is essential to his claim that Kellermann's study is wrong, then he
> >certainly needs to show us why his criticism in not in the same league
> >with the criticisms of smoking studies by the tobacco industry, which
> >everyone agreed identified possible weaknesses, but very few people really
> >thought to be very likely.
>
> I bought my first handgun in response to an attempted rape that
> occurred in the parking lot of my apartment complex. That was way
> back in 1984 or 85. Chris C. whom you know, bought his first gun in
> response to being attacked one night at a Marta Station. Glenn you-
> know-who was attacked on his way home when he lived near Ga Tech.
> He started collecting guns then.
Is Jim McCulloch part of the GT crowd? I don't remember him.....
Julie
>
> Steve
>
> --
>
> /Steve D. Fischer/Atlanta, Georgia/str...@netcom.com/
>
>
uspc...@my-dejanews.com
In article <35797710...@Interaccess.com>,
RAY <Ki...@Interaccess.com> wrote:
> Next time I get a chance to write to Kellermann I shall
> ask him about your phone bill, and about the guns used in
> the remaining 430 homicides. From the study itself we know
> that only about half of the studied 420 cases were committed
> with guns. While it would more clear if we knew which of
> these 209 gun homicides were committed with the guns in the
> home, I don't think this knowledge is essential in order to
> show that guns in the home raise the risk of homicide in the
> home.
Oh, come now Ray. The whole purpose of Kellermann's study was to show that
guns IN THE HOME increase the likelyhood of a gun homicide. To say it
doesn't matter if a gun was in the home of a gun homicide, or if the gun in
the home was the gun used goes against Kellermann's entire arguement.
Your two fictional annecdotes as support are at best a stretch.
Jim
julia.c...@iint.com
In article <DDFr-06069...@ddfr.vip.best.com>,
DD...@best.com (David Friedman) wrote:
>
> In article <stratosE...@netcom.com>, str...@netcom.com (Steve
> Fischer) wrote:
>
> >justify the "correctness" of a study using such a limited population
> >when compared with a study which encompasses the entire country.
> >If you want to study GUNS IN THE UNITED STATES, you can't have a
> >sample database better than the whole country.
>
> this is a valid complaint. He didn't have the data to replicate his study
Of course it's a valid complaint about the study. There have been landmark
studies that have been very small, but statistically significant. But
when you have studies that are small and not terribly well done and along
comes another study with a much larger sample that's also very well done,
it's fair to point out that the size and breadth of the sample strenghthens
the study.
It's not about being poor but pure in heart, it's about finding out
accurate things about the world. Being large and spanning the entire
population one is drawing conclusions about is better than being small
and localized, even if the studies were otherwise equally well designed.
(They aren't, but that's beside the point in this case.)
Perhaps I'm misunderstanding you, but it sounds like you're saying
scientific studies should get sympathy points based on how little money
the researcher has? It doesn't work that way. A better study is a better
study, no matter how much money you spent or didn't spend on it, and having
a larger, broader sample population is one of the elements of "better"
in a study.
(Granted, you can spend a lot of money on a lot of subjects and still
screw it all up with bad design, but Lott didn't do that, so that's not
an issue here.)
Julie
> for the whole U.S., since he got it via questionaires. No doubt if someone
> had offered him a billion dollars in funding ... .
> --
> David Friedman
> DD...@Best.com
> http://www.best.com/~ddfr/
>
julia.c...@iint.com
In article <357A276C...@Interaccess.com>,
RAY <Ki...@Interaccess.com> wrote:
>
> RR: I imagine (in other homes) that friends quite
> frequently bring their guns along. When Billie-Bob goes to
> Bubba's house, they probably sit up half the night playing
> with their guns. Were Billie-Bob to get riled at Bubba for
> burning the grits, it's quite possible that Billie-Bob would
> use his own gun to kill Bubba.
>
I just want to let this stand clear and alone as a sample of
what lives in Robert Ray's head. If it isn't clear enough, change
Billie-Bob to Rastus, Bubba to Tyrone, and grits to chitlins, then
read it again.
Or use Abram, Joseph, and lox. Or Juan, Julio, and chimichangas.
Or Victor, Yuri, and borscht.
Ray, sometimes you disgust me too much for words.
Julie
julia.c...@iint.com
In article <DDFr-07069...@ddfr.vip.best.com>,
DD...@best.com (David Friedman) wrote:
<snip>
> I am saying that his use of a small base is not "incorrect"--it might be
> the best available way of doing the study. Of course we would have more
> confidence in his results if he had had more data--but in the real world,
> one frequently has to do the best one can with limited amounts of
> information.
Ah. I see your point better than I did in an earlier post.
I still don't agree with it though.
Small or unrepresentative sample sizes are raised as a concern with
or weakness of studies all the time. Those concerns are valid and
need to be taken within the context of the quality of the rest of
the study's design and methodology, and the size or significance of
what you find.
In a well-designed study, if what you find is important and interesting
to others, your having a small or possibly unrepresentative sample is
usually good grounds for someone interested by what you found to cough
up the money to try to replicate your work with a better sample.
In Kellerman's case, the study design has enough problems that I
wouldn't think it would be worth doing on a larger sample---better to
start fresh with a better design. Now maybe someone will disagree with
me and follow his research with a better version that addresses many
of the concerns raised, like doing a better job of tracing which gun
was the murder weapon and how it got there and doing a better job of
matching cases with controls.
I think it would be more interesting if someone found a group of
people who looked a lot like his cases, statistically speaking,
questioned them periodically on their perceived level of danger to
themselves and gun owning habits, and followed them longitudinally.
But again, ex-pen-sive---and tough not to have a high drop-out rate,
too. Not to mention whether they'd be truthful about their gun
ownership.
Yes, everybody knows research funds are usually tight and samples are
smaller than we'd like and researchers sometimes use study designs that
have some inherent weaknesses but still can provide good directions for
future follow-up research. This means that choosing to do a study on
a budget isn't a sign of being a bad researcher or being intellectually
dishonest. It's no *personal* shame to the researcher. But it *is*
a valid weakness to question in the study itself.
Julie
> --
> David Friedman
> DD...@Best.com
> http://www.best.com/~ddfr/
>