John Lott-
Stephen J. Fromm
http://markarkleiman.blogspot.com/2003_01_01_markarkleiman_archive.html#90181328
Tim Lambert
> Did John Lott commit academic fraud?
>
> http://markarkleiman.blogspot.com/2003_01_01_markarkleiman_archive.html#90181328
I think he did.
Tim
David Lloyd-Jones
I think it's pretty clear that he did -- and congratulations to you,
Tim, on your part in exposing the creep.
The interesting bit of by-play in all of this is the tendency of some
gun nuts to say Hey, Lott is just the victim of lefties trying to get
back at us for the Bellesiles affair.
Me, I'm one lefty that's covered: I sent Clayton Cramer a
congratulations and Bellesiles-should-be-boiled-in-oil note at the time
of *his* good work.
Cheers,
-dlj.
Grinch
wrote:
>Tim Lambert wrote:
>> On Tue, 14 Jan 2003 14:01:49 +0000, Stephen J. Fromm wrote:
>>>Did John Lott commit academic fraud?
>>>http://markarkleiman.blogspot.com/2003_01_01_markarkleiman_archive.html#90181328
Blogs are nice and enjoyable and sometimes informative, but also
superficial and opinionated to the point of making the tabloids look
researched and judicious.
Anyhow, from which:
"Glenn makes the point that the claim in question -- that only 2% of
self-reported 'defensive gun uses' actually involve firing the weapon,
as opposed to merely brandishing it -- is quite peripheral to the
claim for which Lott is most famous: that laws permitting anyone
allowed to own a gun to get a permit to carry a concealed weapon
reduce the incidence of crime."
Quite correct.
"Fair enough. But what basis is there for believing that more
important claim?"
Oh, say, nine peer-reviewed articles in the Journal of Law and
Economics, to begin with...
http://www.journals.uchicago.edu/JLE/journal/contents/v44nS2.html
If one wants to debate the merits of Lott's argument, debate them for
starters.
As to Lott's "peripheral" purported fraud, the results were readily
duplicated by Lott and others. Which is the essence of scientific work
-- putting out the results so others can duplicate them. Lott
certainly passes that test.
(Which makes harping on "fraud" here fit right into the familiar
usenet pattern of ignoring all substance to indulge in name calling).
>> I think he did.
>
>I think it's pretty clear that he did -- and congratulations to you,
>Tim, on your part in exposing the creep.
>
>The interesting bit of by-play in all of this is the tendency of some
>gun nuts to say Hey, Lott is just the victim of lefties trying to get
>back at us for the Bellesiles affair.
If others had been able to reproduce Bellesiles' data he'd still have
his job and his prize and the publisher wouldn't be killing his book.
He'd still be collecting honors.
Equating Lott to Bellesiles is assinine on its face.
susupply
"Tim Lambert" <lam...@cse.unsw.edu.au> wrote in message
news:pan.2003.01.17....@cse.unsw.edu.au...
In the past you've told the group a few things you "thought" that didn't
hold up to scrutiny, and it looks like this doesn't either:
http://www.juliansanchez.com/2003_01_01_notesarch.html#90191241
<<--------------------quote--------------------------
I decided to do a little more digging on the question of Lott's survey.
University of Chicago professor William Landes was working on a paper with
Lott around the time of the crash: he confirms that Lott lost quite a bit of
data in July 1997, and emphasizes his general impression of Lott's
honesty -- though he also cannot recall any discussion of a lost survey.
There is something slightly more promising, however, from the former U of C
Press editor who worked on the first edition of More Guns, Less Crime.
Recall that Lott says he had intended to include a chapter on the survey in
the book, but opted not to do so when the data was lost. Here's what the
editor wrote me:
"I have a vague recollection of a chapter or a section or sections of a
chapter that had to be scrapped because of the computer crash, but I don't
at this stage remember the subject of it (or them). At the time, we were
talking about a variety of things John could do (e.g., including a chapter
on mass public shootings). As to my e-mail archives, there are a couple of
brief mentions in John's and my exchanges about the crash and loss of data,
though I have found nothing explicitly about the defensive use of handguns
in them (which doesn't mean anything in itself, since we were mostly talking
on the phone and there must have been all kinds of things that were lost in
the crash that we didn't discuss in the archived e-mails I still have)"
Still nothing explicit about the survey, but this does seem to confirm that
at least some chapter had to be scrapped as a result of the crash, whereas
the claim would have been falsified had no changes in the tentative plan of
the book been made as a result of the crash.
-------------------------endquote------------------------------->>
All this over a trivial part of "More Guns, Less Crime".
Stephen J. Fromm
My partisan disposition is to think he did, but I'm not willing to
come to that conclusion until I've looked more at the issues.
What do you know that makes you think such?
Best,
sjfromm
David Lloyd-Jones
>
> As to Lott's "peripheral" purported fraud, the results were readily
> duplicated by Lott and others. Which is the essence of scientific work
> -- putting out the results so others can duplicate them. Lott
> certainly passes that test.
>
I don't know what you mean by "peripheral" fraud, Grinch. Inventing a
survey that didn't take place seems rather more than peripheral, seems
to me, and in any event the staandard for academic work is Caesar's
wife. There ain't no periphery when it comes to getting, handling and
showing data. Anything to do with actual fact is central.
His claim, that in 98% of cases all people did was "brandish" the
weapon, has emphatically *not* been duplicated, contrary to what you
claim here -- assuming that that is what you mean by the peripheral
claim. Of the 12~18 genuine surveys floating around, none show any
percentage lower than 26%, I think it is, of weapons being fired.
In other words, not only was Lott's survey imaginary, but the numbers he
made up to go with it are wildly at variance with reality.
-dlj.
Stephen J. Fromm
> On Fri, 17 Jan 2003 13:52:30 GMT, David Lloyd-Jones <d...@rogers.com>
> wrote:
>
> >Tim Lambert wrote:
> >> On Tue, 14 Jan 2003 14:01:49 +0000, Stephen J. Fromm wrote:
> >>>Did John Lott commit academic fraud?
> >>>http://markarkleiman.blogspot.com/2003_01_01_markarkleiman_archive.html#90181328
>
> Blogs are nice and enjoyable and sometimes informative, but also
> superficial and opinionated to the point of making the tabloids look
> researched and judicious.
Nice ad hominem argument. Doesn't sound like you read MK's blog very
often.
> Anyhow, from which:
>
> "Glenn makes the point that the claim in question -- that only 2% of
> self-reported 'defensive gun uses' actually involve firing the weapon,
> as opposed to merely brandishing it -- is quite peripheral to the
> claim for which Lott is most famous: that laws permitting anyone
> allowed to own a gun to get a permit to carry a concealed weapon
> reduce the incidence of crime."
>
> Quite correct.
>
> "Fair enough. But what basis is there for believing that more
> important claim?"
>
> Oh, say, nine peer-reviewed articles in the Journal of Law and
> Economics, to begin with...
> http://www.journals.uchicago.edu/JLE/journal/contents/v44nS2.html
Right. If a bunch of people peer review each other's publications,
then they *have* to be scientific.
Somehow I doubt you'd be so charitable when it come to journals in
education or environmental science. Just a hunch.
> If one wants to debate the merits of Lott's argument, debate them for
> starters.
I already did---I pointed out in an old post that independent folks
like Ehrlich (a physicist) found Lott's methodology to be sloppy, at
best.
> As to Lott's "peripheral" purported fraud, the results were readily
> duplicated by Lott and others. Which is the essence of scientific work
> -- putting out the results so others can duplicate them. Lott
> certainly passes that test.
No, many have questioned his methodology.
> (Which makes harping on "fraud" here fit right into the familiar
> usenet pattern of ignoring all substance to indulge in name calling).
There's no name calling whatsoever. If he did what people OFF usenet
are claiming he did, he'll never get another government
grant-deservedly so.
> >> I think he did.
> >
> >I think it's pretty clear that he did -- and congratulations to you,
> >Tim, on your part in exposing the creep.
> >
> >The interesting bit of by-play in all of this is the tendency of some
> >gun nuts to say Hey, Lott is just the victim of lefties trying to get
> >back at us for the Bellesiles affair.
>
> If others had been able to reproduce Bellesiles' data he'd still have
> his job and his prize and the publisher wouldn't be killing his book.
> He'd still be collecting honors.
>
> Equating Lott to Bellesiles is assinine on its face.
Really? If Lott fabricated a survey, it's just a petty crime?
Stephen J. Fromm
Maybe trivial. The so-called non-trivial part has been critiqued for
its sloppy methodology.
Clayton E. Cramer
This is no surprise, considering where Tim is on the gun control
issue.
At this point, there is no proof that Dr. Lott fabricated the 1997
survey. There is also no proof that the 1997 survey took place.
The sequence of events that others have brought together is certainly
a little strange, and certainly should cause some headscratching about
this question.
On the other hand, the 2002 survey (which is documented) gives
approximately the same numbers as the 1997 survey. This is evidence
that the 1997 survey could well have taken place. (Not DID but
COULD.)
I really hope that Dr. Lott manages to locate the students that did
the survey in 1997, so that we can clear this matter up. Otherwise,
this is going to be a dark cloud hanging over his head (and his work)
forever.
Keep in mind that this was one sentence out of his book. Until some
clear evidence of fraud is established, I don't think this is a valid
reason to discount the rest of his work--which while some economists
try very hard to discredit it, seems to be in the realm of which
economists do you believe.
I know just enough about statistics to know that in evaluating Lott's
work I am having to take quite a bit on faith in the people who do
know enough about statistics.
David Lloyd-Jones
>
> I really hope that Dr. Lott manages to locate the students that did
> the survey in 1997, so that we can clear this matter up. Otherwise,
> this is going to be a dark cloud hanging over his head (and his work)
> forever.
Clayton,
I shure hope he finds them, too. Unfortunately they may be too busy to
phone in. Didn't that whole bunch get hired by O.J. to help him find the
real killer?
> Keep in mind that this was one sentence out of his book.
And a good solid part of his talk radio and TV schtick, and his magazine
articles, over a couple of years of his moving head career.
-dlj.
Grinch
wrote:
>Grinch wrote:
>>
>> As to Lott's "peripheral" purported fraud, the results were readily
>> duplicated by Lott and others. Which is the essence of scientific work
>> -- putting out the results so others can duplicate them. Lott
>> certainly passes that test.
>
>I don't know what you mean by "peripheral" fraud,
"Peripheral" means, as per the dictionary, "Of minor relevance or
importance". That is, the issue is of little importance to Lott's
analysis -- the discussion of which, pro and con, would remain
unchanged without it.
(And it was the cited critic of Lott who used the word, you will
note.)
But you are correct this far: since the issue is "quite peripheral" --
i.e., of quite minor relevance -- then even if one assumes the very
worst that is charged about it is true (as you are happy to do) it is
hard to see how it could be any kind of "fraud", which by the essence
of the word indicates something substantive and meaningful.
"We have a heinous deceit here that doesn't matter! A fraud that is
irrelevant to substance!!"
BTW, as to the merits of the real argument I don't know or have a
position, or particularly care. I have no dog in that fight. But in my
old age I am getting really cranky and annoyed by people who rush into
name calling "liar! thief! resign you scum!" supported only by an
underdose of facts and perspective, and an overdose of
self-rightouesness. Too long reading usenet and the NY Times editorial
pages, perhaps.
>Grinch. Inventing a
>survey that didn't take place seems rather more than peripheral,
You also deleted the word "purported" -- as in "put forth as true on
inconclusive grounds" -- in your rush to name calling.
That is, you are happy to assume guilt while conveniently ignoring
the witnesses who say they recall Lott losing a survey at the time,
and the replication of the data, and so on.
>to me, and in any event the staandard for academic work is Caesar's
>wife. There ain't no periphery when it comes to getting, handling and
>showing data. Anything to do with actual fact is central.
Which is why scientific method calls for making data public, so others
can examine and seek to replicate it -- exactly as Lott did.
Beyond that it's "Caesar's wife" my ass. Scientists are just as much
lying, selfish, manipulators as any other group of people. At least.
But that doesn't reduce the value of their published, reproducible
analysis one bit more than the value of Shakespeare's plays will be
reduced when we find out he was a molester of children and sheep. If
the advance of science depended on scientists having the character of
Caesar's wife we'd all still be living in mud puddles.
To follow the logic of that blog entry....
"Here's something dubious about a little bit of Lott -- though it
is 'quite peripheral' to his main published findings and argument
'fair enough' ... I am going to assume it is fraud ... therefore, I
have no reason to give any respect to his main published findings and
argument and can dismiss them, because they come from a committer of
fraud."
... is beneath stupid. It's just name calling.
Or put it this way: the blogger can refuse to take Lott's main
findings and argument seriously because he doesn't like him -- but he
can't avoid taking seriously the people who take them seriously,
unless he wants to believe that the likes of the contributors, editors
and referees of the JofL&E have all become gun nuts. Which would just
be name calling about a peer-reviewed journal.
Lott's book has been out five years and been examined up down and
sideways -- and for all the argument about it nobody has detected a
whit of "fraud" as to its substance.
Which makes the comparison to Bellesiles assinine.
>His claim, that in 98% of cases all people did was "brandish" the
>weapon, has emphatically *not* been duplicated, contrary to what you
>claim here -- assuming that that is what you mean by the peripheral
>claim. Of the 12~18 genuine surveys floating around, none show any
>percentage lower than 26%, I think it is, of weapons being fired.
That is not my understanding. ;-)
But even if it were so it would be barely relevant to any substantive
issue.
>In other words, not only was Lott's survey imaginary, but the numbers he
>made up to go with it are wildly at variance with reality.
Let's put it this way.
If all of Bellesiles' missing data was equally peripheral to his main
findings and argument, so they continued to stand on their own as
before, and the barely relevant missing data was reproduced with equal
speed and quality as in Lott's case, then *without doubt* Bellesiles
would still have his job and his prize, his book would be going into
extra editions, and he'd still be collecting accolades.
Nobody would be accusing Bellesiles of "fraud" or of anything more
than sloppiness on a tangential point -- and nobody would be
dismissing the major substance of his work on the basis of his not
showing the character of Caesar's wife on a peripheral, tangential
point.
And I'd speculate that that "nobody" would include you.
> -dlj.
>
>
David Lloyd-Jones
>
> Which is why scientific method calls for making data public, so others
> can examine and seek to replicate it -- exactly as Lott did.
Sure. "Here's the data: 2%. It comes from my survey."
> Lott's book has been out five years and been examined up down and
> sideways -- and for all the argument about it nobody has detected a
> whit of "fraud" as to its substance.
Nobody but a few gun nuts has ever taken Lott's book seriously because
the actual experiment has been carried out on a large scale for a
century: Canada vs. The United States side by side.
> Which makes the comparison to Bellesiles assinine.
As you saw in my earlier post, I think Bellesiles should be boiled in
oil, and said so at the time it came up on the blogs in a note to Cramer
congratulating him on his excellent work.
The comparison is not at all asinine: they are two guys who were so sure
of their conclusions that they didn't think mere facts were needed. The
prototype of these guys was Sir Cyril Burt who just *knew* that
intelligence was normally distributed, and it was so obvious that it
wasn't worth the trouble of testing. So he just pencilled in the
invented data on his charts and published it that way. Lott, like
Bellesiles, is cut from the same cloth.
-dlj.
David Lloyd-Jones
>
> That is, you are happy to assume guilt while conveniently ignoring
> the witnesses who say they recall Lott losing a survey at the time,
> and the replication of the data, and so on.
>
Grinch,
There were no such witnesses. There were witnesses who recall, somewhat
vaguely, that well after the event Lott went around demonstratively
saying something or other about his computer having earlier crashed.
Nobody saw the dog eating his homework. Some people think he may, later,
have claimed that there had been a dog in the neighborhood.
Not only did the computer crash, but also he can't recall the names of
the students who did thousands of phone calls to get to 2,000++ people;
he can't find the work sheets; he doesn't recall the study design. Oh,
yeah, and he has no expense records for all these deductible long
distance calls...
>>His claim, that in 98% of cases all people did was "brandish" the
>>weapon, has emphatically *not* been duplicated, contrary to what you
>>claim here -- assuming that that is what you mean by the peripheral
>>claim. Of the 12~18 genuine surveys floating around, none show any
>>percentage lower than 26%, I think it is, of weapons being fired.
>That is not my understanding. ;-)
Your understanding is wrong. A survey of 1,000 people, of whom at most
130 or so have ever pulled out a gun, is not enough to say anything at
all about 2%, i.e. 2.6 people, out of that sample of a sample. (One of
the criticisms of the earlier fantasurvey was that even if it had
existed it would not have been able to prove what he had claimed it showed.)
So there you have it: Lousy research supporting research that has
vanished with an implausible thoroughness. Quack.
-dlj.
susupply
"Stephen J. Fromm" <stephe...@verizon.net> wrote in message
news:b4cc5e7c.0301...@posting.google.com...
> "susupply" <susu...@mindspring.com> wrote in message
news:<b098vj$ch5$1...@slb9.atl.mindspring.net>...
> > All this over a trivial part of "More Guns, Less Crime".
>
> Maybe trivial. The so-called non-trivial part has been critiqued for
> its sloppy methodology.
So what, you yourself on this forum have been criticized for being sloppy in
your assessment of Lott's work. Remember:
From: Christopher Auld (au...@acs.ucalgary.ca)
Subject: Re: Did the Nobel Peace Prize come with a money back guarantee?
View: Complete Thread (19 articles)
Original Format
Newsgroups: sci.econ
Date: 2002-11-01 11:00:11 PST
Stephen J. Fromm <stephe...@verizon.net> wrote:
>But if you read what Ehrlich is saying, it puts Lott in an extremely
>bad light. Ehrlich alleges that "Lott neglects to tell the reader
>that all his plots are not the actual FBI data (downloadable here),
>but merely his fits to the data" and "Lott doesn't deny that he
>misleads the reader by neglecting to mention that his plots are fits
>to the data, because he can't. His graphs are in fact labelled "number
>of violent crimes" per 100,000 population and I find no statement in
>his book that the graphs are fits, rather than actual data.
This accusation is incorrect. For the most part (but not
always), Lott does plot estimates from econometric models,
but it should not be at all unclear that that is what he
is doing. It is very obvious from the text which graphs
refer to actual data and which graphs refer to model
estimates: Lott displays the regression estimates, then
refers the graphs to show them in a more easily digestible
manner than tables of numbers.
Since his book is aimed a lay audience, possibly he could
have been clearer by referring to which sets of estimates
are displayed in each graph, but nonetheless `he could have
been clearer' is not tantamount to `this is academic fraud.'
Other authors have pointed out some caveats to Lott's work
which mitigate or possibly reverse Lott's conclusions (e.g.
Mark Duggan), but the charge that Lott is deliberately
misleading readers in the manner suggested above is unfounded.
--
Chris Auld
Department of Economics
University of Calgary
au...@ucalgary.ca
susupply
> > Oh, say, nine peer-reviewed articles in the Journal of Law and
> > Economics, to begin with...
> > http://www.journals.uchicago.edu/JLE/journal/contents/v44nS2.html
>
> Right. If a bunch of people peer review each other's publications,
> then they *have* to be scientific.
susupply
"David Lloyd-Jones" <d...@rogers.com> wrote in message
news:3E286EBB...@rogers.com...
> Clayton,
>
> I shure hope he finds them, too. Unfortunately they may be too busy to
> phone in. Didn't that whole bunch get hired by O.J. to help him find the
> real killer?
John Lott has never hidden from his critics. Usually he just eats their
lunches, as he did Ehrlich's in the Reason debate at:
http://reason.com/hod/debate1.1.shtml
David Lloyd-Jones
Patrick,
I guess beauty is in the eye of the beholder, because I don't see Robert
Ehlich getting taken here. What strikes me is Lott trying to bluster his
way a bunch of statistical blunders that he himself doesn't understand.
The way he got caught on the study was that he invented a number -- 2%
-- that was wildly at variance with the numbers found by a doaen or so
people who actually had done studies. One of his problems is that he
seems to think that since he doesn't understand statistics -- twice in
the blog reports he is quoted asking people why it mattered whether he
reported numbers of guns pulled per family or by individual -- since he
doesn't understand statistics, therefore nobody else will understand
when he makes something up.
Now the ugly little fact is that a lot of people do understand sampling
and reporting of statistics, and they spotted the guy for what he is, a
phoney, and apparently a not very bright one.
-dlj.
Mary Rosh
It doesn't appear as if you all are reporting both sides of this
debate. The following was sent to me from Marie Gryphon who has most
of this up on her web site at:
http://www.mariegryphon.com/archives/003393.html#003393<p>
In a message dated 1/17/03 11:19:34 AM, mgry...@cato.org writes:
<p>
<< -----Original Message-----
From: John Lott Sent: Thursday, January 16, 2003 1:49 PM
To: Jim Henley; jul...@juliansanchez.com; Marie Gryphon
Subject: I don't know if you all got this previously
<p>
You can republish what is below if you desire. It was sent out a
couple of days ago, though I can't remember if I sent this to you.
<p>
Here is a response to some of what has been going on over the web. I
have already sent much of this information to people who have already
contacted me in person. If Eugene would like to post this on his web
site, I must ask that all the e-mail addresses and telephone numbers
be removed. If you all don't trust the leg work done by Dan Polsby on
this issue since Christmas, you can nominate someone else to go and do
it, but I don't think that it is appropriate for everyone from Lambert
on to go and harass these people.
<p>
Regnery (the publisher of my new book due the middle or end of March)
wants me not to release the results from the poll last year. They
want me to keep quite about the book until it comes out. As has been
reported previously, the survey was done with similar questions in a
very similar way to what was done earlier and the results were
essentially the same. I will check with the publisher about releasing
this data early, but it is still two months before the book is due to
be published. In the interim, I am sure that I could arrange it so
that interested parties could question the person who keep the survey
results as they came in to confirm that we only got one person who
said that they had actually fired a gun and that the rest were
brandishings.
<p>
Here are some of the things that I have done to try to establish a
record of events. 1) My wife contacted the bank that we had in
Chicago and tried to get copies of bank statements and checks from the
period of time. Unfortunately, the bank does not keep copies of
statements or checks longer than five years. (If you would like to
verify, we talked to Yvonne Macias in the book keeping department at
University National Bank.) Lindgren does not accurately report my
conversation with him about how I paid people (in that I said that I
possibly paid by check), but this information makes that point
irrelevant. 2) I asked Sam Peltzman last year about whether the
Alumni Association has the e-mail of past students. Sam, who seems to
know virtually everything that is going on at the University, told me
that they have the e-mail addresses for at most 10 percent of the
former students. 3) I had a former alumni and several time co-author,
John Whitley, placed in an ad in the Alumni magazine in the December
issue to track down the students. I don't know if the ad has appeared
but thus far I have gotten no response.
<p>
I have given out massive amounts of data to people on the guns and
other issues, and I will be happy to do so on the new survey. Data
has been given to critics as well as people who have been unwilling to
share their own data on other projects. I have given out county,
state, and city level crime data to academics at dozens of
universities, with data sets ranging from 36MB to over 300MB. I have
given out data on multiple victim public shootings as well as safe
storage laws. These different data have often been given out before
the research is published and sometimes even before it has been
accepted for publications. We are not talking about recent events or
conversations and there is a question about what is a reasonable time
period for people to keep records. There is also a question as to why
people have waited so long to ask for this additional information when
people have known about the lost data for years.
<p>
As to the claims about "apparently changing positions," I disagree. I
have told people directly (including Otis Duncan) from the beginning
that the data were lost. Op-ed pieces and other public statements
where I mention these numbers briefly usually do not lend themselves
to discussions of the sources of numbers. The fact that David Mustard
does not remember exactly when we discussed the survey 6+ years ago
does not surprise me given how long ago this was.
<p>
Unfortunately, there are many problems with Lindgren's write up. He
gives essentially uncritical acceptance of Otis Duncan's discussion of
events in 1999. Yet, while Lindgren writes that "Otis Dudley Duncan
raised questions about the 98% figure . . . after exchanges between
Lott and Duncan," Duncan's write-up in the Criminologist news letter
failed to mention any such possible discussions. In fact his
newsletter piece leaves the opposite impression as he endlessly
speculates about what I may have meant about certain statements. My
response in the Criminologist also discussed other incorrect claims by
Duncan.
<p>
As to the attribution of sources, look at the complete context of the
quote Lindgren mentions:
<p>
Polls by the Los Angeles Times, Gallup and Peter Hart Research
Associates show that there are at least 760,000, and possibly as many
as 3.6 million, defensive uses of guns per year. In 98 percent of the
cases, such polls show, people simply brandish the weapon to stop an
attack. -- August 6, 1998, Chicago Tribune and August 14, 1998,
Washington Times
<p>
References by Lindgren to things like the Linnet Myers piece in the
Chicago Tribune to provide evidence that I didn't do a survey or that
I have changed my statements over time are simply bizarre. Attached
below is an edited down version of the letter that was published by me
in the Tribune. Myers used her article to refloat claims such as my
Olin Funding, inaccurately reported exactly what the concealed handgun
research covered, and claimed that "others haven't confirmed (my)
findings." I no longer have the original letter to the editor, but as
I recall this is just a partial listing of her inaccurate statements.
The Tribune was not willing to run a longer letter, though the letter
that they ran was quite long.
<p>
As to so-called technical problems, I am have always acknowledged that
these are small samples, especially when one breaks down the
composition of those who use guns defensively. Even the largest of
the surveys have few observations in this category. The attached
e-mail that I sent to Glenn Reynolds goes into this more in depth.
<p>
"No direct evidence of survey" – discussing Lindgren's point-by-point
discussion of our conversation
<p>
1) "No funding for the project" I regularly have paid for
research myself. Sometimes large amounts of money have been spent,
but it is not uncommon for me to spend several thousand dollars. On
the paper on multiple victim public shootings, I know that one payment
that I made to Kevin, a research assistant to Landes and Posner, was
$750. I paid for the special issue of the JLE in 1999 on sentencing
myself, and the special issue and part of the conference cost me
around $30,000. I have not applied for funds from outside sources
over the years.
<p>
2) "No financial employee records" This is not unrelated to the
first point. Incidentally, I told Jim that there were "two" Chicago
students. Those students had also gotten others that they knew from
other campuses from places such as I think the University of Illinois
at Chicago circle (but I am not sure that I remember this accurately).
What I told him was that I remembered that one of the two University
of Chicago students was a senior.
<p>
3) "calling was done by the undergraduates from their own phones."
most of this next statement is correct except the point about the
"possible" use of checks. But as noted earlier this point is
irrelevant in terms of evidence.
<p>
4) "does not remember names" I have had 12 interns and RAs just since
I arrived at AEI. This excludes people whose only work was on the
survey. I am horrible at names and I couldn't even give you the names
for all of these folks let alone people who did something six years
ago. All my names and addresses for everything were on my computer
when the hard disk crashed.
<p>
5) "no discussions with any samplers"
<p>
I had lunch Tom Smith during the fall of 1996. However, while I asked
him many questions about surveys, I did not tell him what I was
planning on doing because Tom works very closely with gun control
organizations.
<p>
6) weighting the sample
<p>
I did not weight the sample by household size but used the state level
age, race, and sex data that I had used in the rest of my book. There
where 36 categories by state. Lindgren hypotheses why you can get
such small weights for some people and I think that this fine of a
breakdown easily explains it. I don't remember who answered what
after all these years, but suppose someone who fired a gun was a
elderly black in Utah or Vermont.
<p>
7) "commercially available CD-ROM with names on it. He does not
remember where he got it from."
<p>
It is true that I don't have the original CD-ROM. I have a telephone
number CD from the end of 1997, but it is not the one that we used. I
only picked up the other one on the off chance that I was going to
have the time and resources to redo the lost data. The CD did have
the features that the earlier one had and was not very useable. I was
so rapped up in trying to replace my lost data on so many other
projects that I had no thought of going back to what I regarded as a
minor project. I had revise and resubmits at the JPE and other
journals that had much greater importance and the data for the book
had to be replaced.
<p>
8) "Lott does not remember how he drew his sample from the CD-ROM"
<p>
Not true. I told Jim that one of the students had a program to
randomly sample the telephone numbers by state. My guess is that it
was part of the CD, but on that point I can't be sure.
<p>
9) "doesn't remember the wording of the questions."
<p>
It is also not quite correct to say that "doesn't remember the wording
of the questions." I told Jim that I don't remember the "exact
wording" of the questions, but I gave him the general outline of the
questions.
<p>
10) more on weighting
<p>
See point 6 above.
<p>
11) "A chapter he had not yet written"
<p>
This is not correct. What I had done is write up the section, but I
only had a computer file of it. When the hard disk crashed, I only
had a hard copy of the book and I had to spend considerable time
scanning in the book and correcting the new file. I was unable to
replace the lost polling section that I had recently added. I didn't
think that it was worthwhile relying solely on memory for different
things and I had too much else to do to concern myself with something
that wasn't central to the book.
<p>
12) "did not retain any of the tally sheets"
<p>
I have looked through some things but I haven't found anything. As
Lindgren correctly notes, I have moved three times in the last six
years.
<p>
13) Sheets versus entry of data into computers
<p>
Lindgren has the "impression" that the students entered the data on
sheets. I do not directly recall this part of our conversation, but I
would have said that both were done.
<p>
I sent Lindgren two e-mails on December 26th. Just so no one accuses
me of adding new things in now, one of my e-mails to Lindgren noted:
"I did not take the time to correct or respond to all the issues
raised, but I wanted to mention a few points." Recent e-mails to
Lindgren have also already responded to some of these points beyond
the e-mail that he apparently posted.
<p>
I have not participated in the firearms discussion group nor in the
apparent online newsgroup discussions, but what I have done is respond
to e-mails. (The one exception are those from Lambert whose e-mail
address was placed on my blocked list.) If you all have questions, I
will be happy to discuss them, but I am not going be involved in these
online groups. My response to Glenn below goes through some of the
history of what I heard on this and when I heard it. The bottom line
is that you all should not assume that everyone participates in these
discussions.
<p>
Appendi
<p>
xChicago Tribune June 20, 1999 Sunday, CHICAGOLAND FINAL EDITION
<p>
SECTION: MAGAZINE; Pg. 4; ZONE: C; LETTERS TO THE EDITOR.
<p>
LENGTH: 684 words
<p>
HEADLINE: GUNS AND CRIME
<p>
BODY: The article accompanying "Anne, Get Your Gun" (May 2),
discussing my book "More Guns, Less Crime" (University of Chicago
Press, 1998), made several inaccurate claims. Despite the claims in
the article, my research looked at much more than just the "impact of
laws that allow guns to be carried outdoors." My book analyzed FBI
crime statistics for all 3,054 American counties from 1977 to 1994 as
well as extensive cross-county information on accidental gun deaths
and suicides. This is by far the largest study ever conducted on
crime, accidental gun deaths or suicide. I examined not only
concealed-handgun laws, but also other gun-control laws such as state
waiting periods, the length of waiting periods, the Brady law,
criminal background checks, penalties for using guns in commission of
crime and the impact of increasing gun ownership. The only gun laws
that produced benefits were those allowing concealed handguns. The
evidence also strongly indicates that increased gun ownership on net
saves lives. More disappointing were inaccurate references to the
funding of my research. The claims previously floated by gun-control
groups like Handgun Control were found by the Tribune's own Steve
Chapman to be false (Aug. 15, 1996). Chapman pointed out that not only
was the Olin Foundation "independent" of the ties the Sunday Magazine
article discussed, but also that the "foundation didn't (1) choose
Lott as a fellow, (2) give him money or (3) approve his topic." The
article's claim that "others haven't confirmed (my) findings" is
bizarre. To date, I have made the data available to academics at 37
universities, from Harvard to Berkeley. Everyone who has tried has
been able to replicate my findings, and only three have written pieces
critical of my general approach. Although the vast majority of
researchers concur that concealed weapons deter crime, not even those
three critics have argued that more guns cost lives or increase crime.
-- John R. Lott Jr., University of Chicago Editor's note: Reporter
Linnet Myers responds: Various researchers have praised John Lott's
thorough research, although some disagree with his results, which
indicate that crime drops when laws allow citizens to carry concealed
guns. Whether his findings have been "confirmed" may depend on exactly
what that means. Three professors interviewed at separate
universities said Lott's data and computations were mathematically
correct. But because each professor's analysis differed, one didn't
find significant drops in crime while another found more dramatic
decreases than Lott did. The third said Lott's results have been
"confirmed in the sense that they've been replicated." Yet the
findings remain hotly debated. Some researchers, as well as many
gun-control advocates, flat-out reject them. Others say only time will
tell. In the midst of this controversy, my statement that Lott's
results haven't been "confirmed" was one of caution. And the article
did not suggest that he hasn't studied anything beyond those laws.
Most researchers interviewed did agree on one point: Despite the fears
of gun-control groups, there is currently little evidence that the
laws have caused any rise in crime. Lastly, though I don't think the
reference to Lott's funding was "inaccurate," it may have been
unclear. The original version of my article quoted a researcher who
said that while Lott's fellowship had a link to an ammunition company,
"Lott's findings weren't swayed by the somewhat remote connection."
The researcher said that though gun-control advocates have focused on
it, the funding foundation "isn't reputed to be an arm of the gun
industry any more than the Rockefeller Foundation is a tool of the oil
companies." Because of limited space, however, the story was cut and
that quote never made it into print. I apologize to Mr. Lott for that
trim.
<p>
Most of an e-mail that I sent recently to Glenn Reynolds (cutting
some personal comments at the end)
<p>
Dear Glenn:
<p>
First, I have responded to people. I responded to the e-mail from you
that had been forwarded via Clayton Cramer last year (you did not send
it directly to me for some reason). I have responded extensively to
Polsby when he wrote me after Christmas and I responded again to
Lindgren (twice) when he e-mailed me on December 24th. During the
last week, I have also corresponded with Dave Kopel. The data on the
original survey was lost and I will go into it later. First, here is
a similar survey that I did as well as some comments on it. This
survey is NOT for public dissemination as it is for a book that I have
that will shortly becoming out. My publisher would be very upset if
the results of the survey or the survey itself were released.
<p>
Survey questions:
<p>
Hello, my name is _______, and I am a student at ________ working on a
very brief survey on crime. The survey should take about one minute.
Could I please ask you a few questions?
<p>
1) During the last year, were you ever felt threatened with physical
violence or harmed by another person or were you present when someone
else faced such a situation?
<p>
(Threats do not have to be spoken threats. Includes physically
menacing. Attacks include an assault, robbery or rape.)
<p>
a) Yes b) No c) Uncertain d) Declined to answer
<p>
(Just ask people "YES" or "NO." If they answer "NO" or "Decline to
answer," go directly to demographic questions. The vast majority of
people will answer "NO." If people say "Uncertain," continue on as if
they answered "YES." 90 plus percent will probably "NO." )
<p>
2) How many times did these threats of violence or crimes occur?
_____
<p>
3) Which of the following best describe how you responded to the
threat(s) or crime(s)? Pick one from the following list that best
described your behavior or the person who you were with for each case
faced.
<p>
a) behaved passively b) used your fists c) ran away d) screamed or
called for help e) used a gun f) used a knife g) used mace h) used a
baseball bat or club
<p>
i) other
<p>
(Rotate these answers (a) through (h), place a number for 0 to
whatever for each option. Stop going through list if they volunteer
answer(s) that account for the number of threats that they faced.)
<p>
4) This is only done if answer "e" (a gun) to question 3
<p>
If a gun was used, did you
<p>
a) brandish it b) fire a warning shot c) fire at the attacker d)
injure the attacker e) kill the attacker
<p>
(Again, place a number for 0 to whatever for each option. Rotate
answers.)
<p>
5) Were you harmed by the attack(s)?
<p>
a) Yes b) No c) Refused to answer
<p>
(We obviously have the area code, write down sex from the voice if
possible.)
<p>
I have two demographic questions for the survey.
<p>
What is your race? black, white, Hispanic, Asian, Other.
<p>
What is your age by decade? 20s, 30s, 40s, so on.
<p>
Question for surveyor: Is there any reason for you to believe that
the person was not being honest with you?
<p>
a) Didn't believe respondent at all b) Had some concerns c) Had no
serious concerns
<p>
<p>
Write up by James Knowles of the discussion of the survey:
<p>
We had a small army of interns and AEI staff making phone calls. The
callers for any given night varied according to who was
available/willing to make phone calls. I was here every night
supervising from my office at AEI. The survey was conducted over eight
nights. Calls were made between 7pm and 9pm local time. Here are the
list of callers and their email addresses, I can try to track down
phone numbers if need be.
<p>
Susan Follett- Carl Westine- Andrei Zlate- Radek Muron- Arnaud
Bonraison- Craig Morehead- Melissa Robe- Salma Sallam- Jill
Mitchell- Matt Trager- and Myself-
<p>
. . .
<p>
We used a phonebook program from a company called infoUSA the program
was called Select Phone Pro version 2.4. The program has a random
function. First, we calculated how many numbers should be drawn from
each state, we decided that we would pull 4,000 numbers (based on how
much the PhonePro program gives us for free). Then we took the
populations of each state from the Census and assigned the quantity of
numbers that we indended to get from the program. Then, in PhonePro,
we picked a state, then sorted the state's list by area code, then
randomly generated a number (using excel's analysis pack) as a
starting point, then the Phone Pro program would export a number every
so often from the list until it reached the desired number of listings
exported from the state. This may be something that is easier
explained in a conversation . . ..
<p>
___________________
<p>
End discussion on survey. James' E-MAIL address is: jkno...@aei.org.
<p>
<p>
If you have questions or doubts about my discussion of what happened,
you might want to ask me about what happened. If you want to talk
during the day on Monday, you can call me up at (610) 543-2409.
<p>
1) I have lots of people who can say that I lost my hard disk in July
of 1997: for example, David Mustard, Geoff Huck (the editor who
handled my book at the University of Chicago Press), John Whitley,
William Landes. If you want to you can feel free to contact
especially David and John. Russell Roberts is someone that I bounced
the survey questions off of and he can possibly talk to you about it,
though he hasn't kept e-mails from 1997.
<p>
2) I lost ALL the data for my paper with David in the JLS and for my
book. David and I reconstructed the county and state level data from
our paper and I got the rest together that was in my book. I have
consistently provided the county, state, and city level data as well
as the multiple victim shooting data and the safe storage data to
people when they have asked. That constitutes almost all the numbers
mentioned in the book. Months were spent redoing this data so that it
could be given out to people. (Just a note, the critics to whom we
had given out the data up to that point were unwilling to return the
data to us so we had to put the JLS data together all over again.) If
I had the other data, I would have been happy to give it to who ever
wanted it.
<p>
The survey data could not be replaced by going to things like the UCR
or the census data. It could only be replaced by doing another
survey. I ended up only briefly mentioning the survey in the past and
worked on replacing all the other data that I lost for not just the
book but for all the other projects that I was working on. I spent a
good portion of the next two years trying to replace data for other
projects that I had been working on. I had important papers for the
Journal of Political Economy and other journal for which replacing the
data was my first priority after replacing the data for the JLS and
the book. Thinking about the survey was well down on my list. Its
importance was not particularly high given that I had only one
sentence on the issue in my book and have never written the survey
into a research paper because the data was lost.
<p>
3) This is something that was done six years ago. During the
intervening time I have moved three times. Usually I pay students in
cash. When I am at universities I don't apply for grants and the
money is mine so there is no record of universities paying for the
students. My records of the students names and contact information
were lost. You can get an idea of how much total time was involved
from the survey discussed above. An ad was taken out in the fall in
the University of Chicago Alumni magazine to try to contact the two
University of Chicago students who organized some other people from
different places to work on it. I have gone through well over 12 RAs
and interns independent of those used on the survey since I got to AEI
and I can't say that I remember most of their names (I am really
horrible at names).
<p>
4) Issues about the significance difference in results. Given the
very small sample sizes, the differences in results are not
statistically significant and are really trivial. About one percent
of people in a survey note that they have used a gun defensively.
Whether one is talking about 2 percent or 20 percent simply
brandishing a gun, you are talking about the different of 2 percent of
1 percent or 20 percent of 1 percent. Depending upon who answers the
questions and what weighted group they are in, you are literally
talking about the answers of a few people out of 1,000 that is the
difference between these two results.
<p>
5) If you have doubts about anything in specific, you should ask me
about it. Last year I worked extremely hard. I was the only
affirmative numbers expert for the Senator Mitch McConnell's side in
the campaign finance case. Once that was done, I had to deal with
something that Ian Ayres and John Donohue wrote attacking my work and
finishing things for my book. Only in the last couple of weeks have I
gotten a breath on things, but I have responded when people e-mailed
me. I am not a member of the firearms discussion groups and I have
not been following them. I read your site once in a while just to
keep up with the news (and because of that I have sent you some money
from time to time e.g., just on Saturday), but otherwise I have been
too busy to follow a lot of things. (When I recently accidentally
sent you an e-mail it was at the end of one of many all nighters.)
<p>
So as to state things clearly, the bottom line is that I have
provided data on county, state, and city level crime data when I have
been asked as well as the data on the multiple victim public shootings
and the data on safe storage laws, even before the papers have been
published. We are talking about one number in one sentence in the
book, a claim that I have also used in some op-eds and in some talks.
I know of no one who has given out his data as quickly and
consistently before even papers are published as I have and over the
years. Finally, let me note the most important bottom line: the
survey that was done last fall produced very similar results. The
earlier results were replicated. This survey was done more recently
and I will release the data when the book is released in March. To
keep the publisher happy, I will not release it before hand unless you
can give me a very good reason.
<p>
[cut] There are errors in Lindgren's write up (at least the one that
he sent me) and if you have specific questions about it, I will
respond. But instead of claiming that I haven't responded to people
you should talk to people like Dan Polsby who raised claims voiced to
him by others that I had fabricated this second survey. He spent a
good deal of time verifying that the survey did indeed take place.
Polsby can be reached at XXX. I am sure that he would be happy to
talk to you.
<p>
Best.
<p>
John
<p>
David Lloyd-Jones
>
> It doesn't appear as if you all are reporting both sides of this
> debate. The following was sent to me from Marie Gryphon who has most
> of this up on her web site at:
> http://www.mariegryphon.com/archives/003393.html#003393<p>
>
Mary,
Aren't you embarassed by how often on Marie Gryphon's site the guy, when
asked for evidence for something, points to the fact that he had said it
before as being that evidence?
-dlj.
Mary Rosh
The Lindgren piece mentions that David Mustard remembers talking to
Lott about a survey in 1997.
"Mustard recalls hearing about the 1997 study, though when Lott told
him about it is a little unclear from Mustardï¼´ email to me: John told
me that he had conducted a survey in 1997. I did not participate in
the survey--it was after our concealed carry paper had been published
(Jan 1997) and was after I was on the job market and while I was
finishing my dissertation and then moving to Georgia (Aug 1997)."
One of the bloggers (http://www.juliansanchez.com/2003_01_01_notesarch.html#90191241)
notes that an editor at the University of Chicago Press says:
I have a vague recollection of a chapter or a section or sections of a
chapter that had to be scrapped because of the computer crash, but I
don't at this stage remember the subject of it (or them). At the time,
we were talking about a variety of things John could do (e.g.,
including a chapter on mass public shootings). As to my e-mail
archives, there are a couple of brief mentions in John's and my
exchanges about the crash and loss of data, though I have found
nothing explicitly about the defensive use of handguns in them (which
doesn't mean anything in itself, since we were mostly talking on the
phone and there must have been all kinds of things that were lost in
the crash that we didn't discuss in the archived e-mails I still have)
Lott's piece that I put up earlier also notes that his critics have
known about the lost data for years. Why wait so many years to deal
with this? Memories fade. In any case, I think that Cramer is wrong
to say that there is no evidence.
Malcolm Kirkpatrick
> susupply wrote:...
> > "David Lloyd-Jones" wrote:...
>
MK. Discusson deleted (sarcasm),,,
>
> > John Lott has never hidden from his critics. Usually he just eats their
> > lunches, as he did Ehrlich's in the Reason debate at:
> >
> > http://reason.com/hod/debate1.1.shtml
>
> Patrick,
>
> I guess beauty is in the eye of the beholder, because I don't see Robert
> Ehlich getting taken here. What strikes me is Lott trying to bluster his
> way a bunch of statistical blunders that he himself doesn't understand.
>
> The way he got caught on the study was that he invented a number -- 2%
> seems to think that since he doesn't understand statistics -- twice in
> the blog reports he is quoted asking people why it mattered whether he
> reported numbers of guns pulled per family or by individual -- since he
> doesn't understand statistics, therefore nobody else will understand
> when he makes something up.
>
--not--rhetorical. What difference do you suppose it will make whether
you count non-discharge use by household or by individual? Sounds to
me that Lott here attempts to educate the listener.
>
If you are reporting the frequency with which a firearm is "used" to
discourage a criminal, but not discharged, the question "how many
people live in the house of the gun owner?" is as relevant as "what
color was that house?" Even if the question somehow involved people
related to the gun owner, it is a reasonable question to ask: how much
would counting those others matter? I have not read Lott's book,so if
I misunderstand the issue, correct me here.
>
> Now the ugly little fact is that a lot of people do understand sampling
> and reporting of statistics, and they spotted the guy for what he is, a
> phoney, and apparently a not very bright one.
>
degree. One does not need a PhD in math to use basic statistics to
answer questions such as "What is the impact of concealed carry laws
on violent crime?"
>
MK. 2+2 = 4. I do not have a PhD in math. Is my equation --therefore--
incorrect?
>
David Lloyd-Jones
>
> MK. 2+2 = 4. I do not have a PhD in math. Is my equation --therefore--
> incorrect?
>
First, your addition is correct. Second, its correctness has no
connection with whether you have a PhD or not.
Third, I have mercifully spared you the ignominy of having the rest of
your stupid post repeated.
-dlj.
Tim Lambert
> Subject: Re: Did the Nobel Peace Prize come with a money back guarantee?
> Stephen J. Fromm <stephe...@verizon.net> wrote:
>
>>But if you read what Ehrlich is saying, it puts Lott in an extremely
>>bad light. Ehrlich alleges that "Lott neglects to tell the reader
>>that all his plots are not the actual FBI data (downloadable here),
>>but merely his fits to the data" and "Lott doesn't deny that he
>>misleads the reader by neglecting to mention that his plots are fits
>>to the data, because he can't. His graphs are in fact labelled "number
>>of violent crimes" per 100,000 population and I find no statement in
>>his book that the graphs are fits, rather than actual data.
>
> This accusation is incorrect.
You are mistaken. The graphs are labelled as Ehrlich says,
and if you can find where Lott informs his readers that the
graphs are fits, rather than data, please share it with us.
Furthermore, if you look at Appendix 1, where Lott explains
multiple regression, you will see one example graph of
regression results. That graph shows a fit (a straight line)
AND the data (as dots). Compare that with the graphs in
the body of the book, which show a fit (two parabolic arcs)
and the same fit again (dots lying on the arcs).
> For the most part (but not
> always), Lott does plot estimates from econometric models,
> but it should not be at all unclear that that is what he
> is doing. It is very obvious from the text which graphs
> refer to actual data and which graphs refer to model
> estimates: Lott displays the regression estimates, then
> refers the graphs to show them in a more easily digestible
> manner than tables of numbers.
It is very obvious to you, but you are probably one of
the few people that it is obvious to. For example,
I had a discussion with James Donald where it took multiple
posts before I was able to persuade him that Lott's graphs were
fits. See:
http://groups.google.com/groups?selm=tnwuwgtvem.fsf%40oolong.orchestra.cse.unsw.EDU.AU
> Since his book is aimed a lay audience, possibly he could
> have been clearer by referring to which sets of estimates
> are displayed in each graph, but nonetheless `he could have
> been clearer' is not tantamount to `this is academic fraud.'
> Other authors have pointed out some caveats to Lott's work
> which mitigate or possibly reverse Lott's conclusions (e.g.
> Mark Duggan), but the charge that Lott is deliberately
> misleading readers in the manner suggested above is unfounded.
I don't think Lott is being deliberately misleading with the
graphs. In fact he has misled himself:
"To illustrate that the results are not merely due to the
'normal' ups and downs for crime, we can look again at the
diagrams in chapter 4 showing crime patterns before and
after the adoption of the non-discretionary laws. The
declines not only begin right when the concealed handgun
laws pass, but the crime rates end up well below their
levels prior to the law." -- page 131 MGLC
It is no surprise that there is a peak when the laws were
passed--this is one of the few places where it is possible
for the fitted curves to peak. Even if the crime rate started
to decline before the laws passed, Lott's diagram could still
show a peak coinciding with the law.
I ran some experiments by fitting a similar pair of quadratic
curves to a sequence of random numbers. Almost always the
curves seemed to show that something had happened at the
junction of the two curves, even though nothing had.
Lott has confused the fitted curves with the actual data.
Tim
susupply
"Tim Lambert" <lam...@cse.unsw.edu.au>
spectacularly reversing course,
wrote in message news:pan.2003.01.18....@cse.unsw.edu.au...
> I don't think Lott is being deliberately misleading with the
> graphs. In fact he has misled himself:
Which is, even if true, not at all the same thing as "committing academic
fraud". As Chris Auld correctly pointed out.
Mary Rosh
I don't at all understand what you are saying. I thought that it
seemed like a very complete answer. One of the blog sites also
mentioned that he has given out the data to a second survey that he
did last year and that this survey also got similar results. So what
is the issue here? We have two surveys, both get similar results, but
one is six years ago and the banks don't even keep records that long
ago. So he had a computer hard disk crash that everyone seems to
agree happened. Do you deny that a hard disk crash occurred? THat
crash was six years ago. Give me a break. There is some evidence
that the survey data was lost with the hard disk (Mustard and on a
closely related point Huck).
I am curious how long do you or Lambert keep your cancelled checks?
How many years back do you keep your e-mails? Suppose that someone
did yard work for you six years ago, could you prove that they did it?
Do you have cancelled checks from then? If it was a company that did
it for just one summer, do you remember its name?
Mary Rosh
> > From: Christopher Auld (au...@acs.ucalgary.ca)
> > Subject: Re: Did the Nobel Peace Prize come with a money back guarantee?
> > Date: 2002-11-01 11:00:11 PST
> > Stephen J. Fromm <stephe...@verizon.net> wrote:
> >
> >>But if you read what Ehrlich is saying, it puts Lott in an extremely
> >>bad light. Ehrlich alleges that "Lott neglects to tell the reader
> >>that all his plots are not the actual FBI data (downloadable here),
> >>but merely his fits to the data" and "Lott doesn't deny that he
> >>misleads the reader by neglecting to mention that his plots are fits
> >>to the data, because he can't. His graphs are in fact labelled "number
> >>of violent crimes" per 100,000 population and I find no statement in
> >>his book that the graphs are fits, rather than actual data.
> >
> > This accusation is incorrect.
>
> You are mistaken. The graphs are labelled as Ehrlich says,
> and if you can find where Lott informs his readers that the
> graphs are fits, rather than data, please share it with us.
> Furthermore, if you look at Appendix 1, where Lott explains
> multiple regression, you will see one example graph of
> regression results. That graph shows a fit (a straight line)
> AND the data (as dots). Compare that with the graphs in
> the body of the book, which show a fit (two parabolic arcs)
> and the same fit again (dots lying on the arcs).
I don't have Lott's book in front of me, but I have Lott's paper with
Mustard in the Journal of Legal Studies which uses the same types of
graphs and it is very clear. Look at pages 34 and 35 for a discussion
of how they replaced the simple dummy variable with nonlinear time
trends both before and after the law.
Give it a break Tim, you are again making incorrect statements. Lott
is fitting regressions lines before and after the law and seeing
whether there is a change in patterns. He tests for whether there is
a significant cnanges and finds that the effect is indeed
statistically significant. Tim you might not understand this, but
even with straight lines there are a lot of combinations of results
which would not be consistent with right-to-carry laws reducing crime.
When you look at nonlinear estimates as Lott does crime might rise,
first fall then rise, rise then fall, stay flat, etc. and you have
those combinations in both the before and after periods. Many of
these patterns would not be consistent with the shall issue laws
producing a benefit.
Clayton E. Cramer
> cla...@claytoncramer.com (Clayton E. Cramer) wrote in message news:<6a4c48bf.03011...@posting.google.com>...
> known about the lost data for years. Why wait so many years to deal
> with this? Memories fade. In any case, I think that Cramer is wrong
> to say that there is no evidence.
I agree. The evidence is somewhat weak and circumstantial, rather like
the evidence that suggests that he didn't do the 1997 survey. At this
stage, I would be inclined to say that the evidence on both sides is so
weak that there is no point in arguing that this is evidence of fraud.
Christopher Auld
>> From: Christopher Auld (au...@acs.ucalgary.ca)
>> This accusation is incorrect.
>You are mistaken. The graphs are labelled as Ehrlich says,
The accusation was that Lott deliberately misleads
the reader. The accusation is incorrect because
in context it is obvious what Lott is graphing, so
there is no intent to deceive.
>It is very obvious to you, but you are probably one of
>the few people that it is obvious to.
Anyone familiar with multivariate statistical inference
would have no difficulty interpreting Lott's results.
Anyone who lacks that familiarity should probably not
be arguing about Lott's quantitative methods or results
in any case.
>I don't think Lott is being deliberately misleading with the
>graphs. In fact he has misled himself:
I don't think so, and I don't think your quote proves
your point. Most of the results in the book were
previously published in prestigious journals, and Lott
is a capable econometrician. I don't have the book
handy, but if I recall correctly the key results were
based on dummies for years before and after the
introduction of laws. I don't think your comments on
the statistics are correct, but perhaps you could
present in more detail what you think the problem is,
None of this has anything to do with the recent
allegations that Lott simply fabricated data.
Mary Rosh
> mary...@aol.com (Mary Rosh) wrote in message news:<23fa92fe.03011...@posting.google.com>...
> > cla...@claytoncramer.com (Clayton E. Cramer) wrote in message news:<6a4c48bf.03011...@posting.google.com>...
> > Lott's piece that I put up earlier also notes that his critics have
> > known about the lost data for years. Why wait so many years to deal
> > with this? Memories fade. In any case, I think that Cramer is wrong
> > to say that there is no evidence.
>
> I agree. The evidence is somewhat weak and circumstantial, rather like
> the evidence that suggests that he didn't do the 1997 survey. At this
> stage, I would be inclined to say that the evidence on both sides is so
> weak that there is no point in arguing that this is evidence of fraud.
So what is wrong with Mustard's comments. Why isn't he viewed as
sufficiently credible? Lindgren doesn't seem to put much weight on
them (I guess because he gives a range of time during 1997 between
January and August when he says that he learned about the survey),
but, give me a break, is he supposed to remember down to the exact
date? At least he has the move to Georgia to help pin down the time
when he learned about it. Mustard confirms not only that a survey was
done but that the data was lost in the hard disk crash and that he
knew this by August of 1997. Doesn't this August date fit in well
with Lott's claim about losing the hard disk in July? Julian's web
site's quote from Geoff Huck also provides consistent evidence. Are
these two guys lying? Apparently no one contests that he lost the
hard disk. What is the evidence on the other side?
Lott's comments apparently claim that he has given out lots of other
data. Do you have any evidence that any of these other statements are
not true?
Robert Vienneau
(Christopher Auld) wrote:
> Patrick Sullivan writes:
> > > Tim Lambert <lam...@cse.unsw.edu.au> wrote:
> > >I don't think Lott is being deliberately misleading with the
> > >graphs. In fact he has misled himself:
> > Which is, even if true, not at all the same thing as "committing
> > academic
> > fraud". As Chris Auld correctly pointed out.
> None of this has anything to do with the recent
> allegations that Lott simply fabricated data.
I thank Chris Auld for his honest expression of his disagreement
with Patrick Sullivan.
--
Try http://csf.colorado.edu/pkt/pktauthors/Vienneau.Robert/Bukharin.html
To solve Linear Programs: .../LPSolver.html
r c A game: .../Keynes.html
v s a Whether strength of body or of mind, or wisdom, or
i m p virtue, are found in proportion to the power or wealth
e a e of a man is a question fit perhaps to be discussed by
n e . slaves in the hearing of their masters, but highly
@ r c m unbecoming to reasonable and free men in search of
d o the truth. -- Rousseau
Tim Lambert
> So what is wrong with Mustard's comments. Why isn't he viewed as
> sufficiently credible? Lindgren doesn't seem to put much weight on
> them (I guess because he gives a range of time during 1997 between
> January and August when he says that he learned about the survey),
> but, give me a break, is he supposed to remember down to the exact
> date?
You seem to have misunderstood. Those are the dates when he
thought the survey was, not when Lott told him about it.
> At least he has the move to Georgia to help pin down the time
> when he learned about it. Mustard confirms not only that a survey was
> done but that the data was lost in the hard disk crash and that he
> knew this by August of 1997.
No, he is just reporting what Lott told him. He has no direct
knowledge of the survey.
> Doesn't this August date fit in well
> with Lott's claim about losing the hard disk in July? Julian's web
> site's quote from Geoff Huck also provides consistent evidence. Are
> these two guys lying? Apparently no one contests that he lost the
> hard disk. What is the evidence on the other side?
If a student proves he has a dog, it does not prove his claim that
the dog ate his homework.
> Lott's comments apparently claim that he has given out lots of other
> data. Do you have any evidence that any of these other statements are
> not true?
You should read section 4 of Lindgren's report:
http://www.cse.unsw.edu.au/~lambert/guns/lindgren.html#section4
Tim
Tim Lambert
> Tim Lambert <lam...@cse.unsw.edu.au> wrote:
>>> From: Christopher Auld (au...@acs.ucalgary.ca)
>
>>> This accusation is incorrect.
>
>>You are mistaken. The graphs are labelled as Ehrlich says,
>
> The accusation was that Lott deliberately misleads
> the reader.
No the accusation was that Lott misleads the reader. The
"deliberately" part was your addition.
> The accusation is incorrect because
> in context it is obvious what Lott is graphing, so
> there is no intent to deceive.
I have already demonstrated how people were misled, so
your assertion that it is "obvious" has been proven false.
>>It is very obvious to you, but you are probably one of
>>the few people that it is obvious to.
>
> Anyone familiar with multivariate statistical inference
> would have no difficulty interpreting Lott's results.
> Anyone who lacks that familiarity should probably not
> be arguing about Lott's quantitative methods or results
> in any case.
The intended audience of Lott's book was not people
"familiar with multivariate statistical inference".
The graphs are misleading to almost all of his readers.
>>I don't think Lott is being deliberately misleading with the
>>graphs. In fact he has misled himself:
>
> I don't think so, and I don't think your quote proves
> your point.
You will need to explain why. Here it is again:
"To illustrate that the results are not merely due to the
'normal' ups and downs for crime, we can look again at the
diagrams in chapter 4 showing crime patterns before and
after the adoption of the non-discretionary laws. The
declines not only begin right when the concealed handgun
laws pass, but the crime rates end up well below their
levels prior to the law." -- page 131 MGLC
It is no surprise that there is a peak when the laws were
passed--this is one of the few places where it is possible
for the fitted curves to peak. Even if the crime rate started
to decline before the laws passed, Lott's diagram could still
show a peak coinciding with the law.
I ran some experiments by fitting a similar pair of quadratic
curves to a sequence of random numbers. Almost always the
curves seemed to show that something had happened at the
junction of the two curves, even though nothing had.
Lott has confused the fitted curves with the actual data.
> Most of the results in the book were
> previously published in prestigious journals,
The claim about the timing did not appear in a journal.
Christopher Auld
>Christopher Auld wrote:
>> The accusation was that Lott deliberately misleads
>> the reader.
>No the accusation was that Lott misleads the reader. The
>"deliberately" part was your addition.
I don't even know how to begin to make sense of this. Could
a statistically illiterate person be confused by a discussion
of regression results? Sure. Does Lott present his results
in a confusing or non-standard manner? No. Upon looking at
the book again, I don't even know how one could conceivably
confuse the graphs for real data. The data, for example, is
at the county level. What data would the single curves
presented on the graphs represent?
>> The accusation is incorrect because
>> in context it is obvious what Lott is graphing, so
>> there is no intent to deceive.
>
>I have already demonstrated how people were misled, so
>your assertion that it is "obvious" has been proven false.
I find your quarrelsome tone irritating. I have said, and
will say again, that someone unfamiliar with statistics may
be confused by Lott's presentation. But that is true of
all writing drawing on mathetmatical statistics, and -- again --
I do not think Lott's presentation is confusing, intentionally
or otherwise.
>The intended audience of Lott's book was not people
>"familiar with multivariate statistical inference".
>The graphs are misleading to almost all of his readers.
You will have to provide some justification for this remark.
Notice that pointing to a usenet idealogue who couldn't
understand the book does not constitute such a justification.
>> I don't think so, and I don't think your quote proves
>> your point.
Tim then repeats verbatim his previous comments, after snipping
my request he provide more detail on what he thinks the problem
is.
>I ran some experiments by fitting a similar pair of quadratic
>curves to a sequence of random numbers. Almost always the
>curves seemed to show that something had happened at the
>junction of the two curves, even though nothing had.
>
>Lott has confused the fitted curves with the actual data.
Lott's models are perfectly standard, as is his interpretation.
Ignoring quadratic terms, the models in chapter 4 take the form
y = Xb + ad + b(dt) + noise,
where d is a dummy indicating a policy change and t is a
time trend. If one generates a "sequence of random numbers"
with the property that nothing happens in response to the
policy, one will not "almost always [find] the curves show
that something happened at the junction." One will in fact
find, with reasonably large samples, that the coefficient on
the interaction term is statistically insignificant at a
rate equal to the size of the test, conventionally 5%. What
is more, most of Lott's models are less restrictive, treating
the impact of a policy change in a semiparametric manner by
using sets of dummies capturing time to/since the policy
change. Again, this is all very standard and uncontroversial.
Mary Rosh
> On Sun, 19 Jan 2003 14:46:49 +0000, Mary Rosh wrote:
>
> > So what is wrong with Mustard's comments. Why isn't he viewed as
> > sufficiently credible? Lindgren doesn't seem to put much weight on
> > them (I guess because he gives a range of time during 1997 between
> > January and August when he says that he learned about the survey),
> > but, give me a break, is he supposed to remember down to the exact
> > date?
>
> You seem to have misunderstood. Those are the dates when he
> thought the survey was, not when Lott told him about it.
Mustard's statements seem clear to me.
>
> > At least he has the move to Georgia to help pin down the time
> > when he learned about it. Mustard confirms not only that a survey was
> > done but that the data was lost in the hard disk crash and that he
> > knew this by August of 1997.
>
> No, he is just reporting what Lott told him. He has no direct
> knowledge of the survey.
I read the Mustard quote. If you have other information from
Professor Mustard directly, let us know about it. Have you contacted
Professor Mustard? What additional clarification can you add?
>
> > Doesn't this August date fit in well
> > with Lott's claim about losing the hard disk in July? Julian's web
> > site's quote from Geoff Huck also provides consistent evidence. Are
> > these two guys lying? Apparently no one contests that he lost the
> > hard disk. What is the evidence on the other side?
>
> If a student proves he has a dog, it does not prove his claim that
> the dog ate his homework.
Huck is saying that it thinks that part of the book was lost because
it was only available on the hard disk (consistent with Lott) and that
the hard disk crashed around the time that Lott said that it did.
What evidence do you have that the data wasn't on the hard disk? Are
you just asserting that you have additional information that the data
was not on the hard disk? If you do, please let us all know about it.
>
> > Lott's comments apparently claim that he has given out lots of other
> > data. Do you have any evidence that any of these other statements are
> > not true?
>
> You should read section 4 of Lindgren's report:
You are going to have egg all over your face on this one Lambert. You
are already starting to look desparate. Six years is a really long
time to ask people for records for, and yet different people have
confirmed different parts of Lott's discussion. But I will tell you
what. If you get Professor Mustard to agree with your interpretation
of what he said, I will agree with you. For some reason, I bet that
you are not even going to try asking him. You are a professor
(right?), so just ask him one professor to another about whether your
interpretation is correct.
Mary Rosh
> On Sun, 19 Jan 2003 14:46:49 +0000, Mary Rosh wrote:
>
By the way TIm, what about the fact that Lott's 2002 survey apparently
produces the same results? Do you think that survey was fabricated?
Was it just dumb luck that Lott guessed correctly what the results
would be?
susupply
> You should read section 4 of Lindgren's report:
> http://www.cse.unsw.edu.au/~lambert/guns/lindgren.html#section4
Here's more from Lindgren, and he's accepting more evidence that the survey
was in fact conducted:
http://www.juliansanchez.com/2003_01_01_notesarch.html#90211591
Clayton E. Cramer
> Grinch wrote:
> >
> > Which is why scientific method calls for making data public, so others
> > can examine and seek to replicate it -- exactly as Lott did.
>
> Sure. "Here's the data: 2%. It comes from my survey."
>
> > Lott's book has been out five years and been examined up down and
> > sideways -- and for all the argument about it nobody has detected a
> > whit of "fraud" as to its substance.
>
> Nobody but a few gun nuts has ever taken Lott's book seriously because
> the actual experiment has been carried out on a large scale for a
> century: Canada vs. The United States side by side.
Do you seriously suggest that the only differences between the two
countries over a century are concealed weapon permit availability?
Remember that until 1961, there was NO American state that had non-
discretionary issuance of concealed weapon permits. Nearly all of
this took place in the 1980s and 1990s.
When you compare states and provinces across that boundary line, you
find a lot of interesting differences, and a lot of interesting
similarities. Prairie provinces are often comparable in murder rates
to their American neighbors. New York State, of course, is terribly,
terribly violent--but then again, it's a restrictive state not just about
concealed weapon permits, but even about permission to have a handgun
in your own home.
Seattle and Vancouver are an interesting comparison, because murder
rates for whites are about the same; Seattle is very slightly lower
(6.2/100,000 instead of 6.4 in Vancouver). The big difference is
that Seattle has a much larger black and Hispanic population, and
this is most of the difference. Since there is no difference in
handgun ownership laws or carry permit issuance based on race in
Seattle, or I presume, in Vancouver, this suggests that the difference
isn't the laws, but the subcultures that are associated with black
and Hispanic populations in the U.S.
> The comparison is not at all asinine: they are two guys who were so sure
> of their conclusions that they didn't think mere facts were needed. The
> prototype of these guys was Sir Cyril Burt who just *knew* that
> intelligence was normally distributed, and it was so obvious that it
> wasn't worth the trouble of testing. So he just pencilled in the
> invented data on his charts and published it that way. Lott, like
> Bellesiles, is cut from the same cloth.
>
> -dlj.
It is now established that the 1997 survey did take place; at least
one person who was surveyed has come forward. Professor Lindgren,
my fellow inquisitor of Bellesiles, has interviewed the survey
subject (a lawyer, and an assistant district attorney a few years
back), and concluded that he is a credible witness.
David Lloyd-Jones
> Do you seriously suggest that the only differences between the two
> countries over a century are concealed weapon permit availability?
Uh, no, Cramer, I don't seriously suggest that this is the only
difference between the two countries over a century. (and I have no clue
what you think is plural in your sentence above.)
Have you considered the possibility that this is an astonishingly stupid
question to ask? And that you incomprehensible grammar is caused by a
flush of high emotion caused by the montion of guns?
> Remember that until 1961, there was NO American state that had non-
> discretionary issuance of concealed weapon permits. Nearly all of
> this took place in the 1980s and 1990s.
So? Even if you take Lott seriously, there's very little difference
before and after in any of his many counties. The differences between
Canada and the US in both crime and in gun violence are striking -- at
the 98% level of confidence. Lott's claim, by contrast, is that he has
carried out two surveys which are accurate within 1,300% 50% of the time
or 3,000% all of the time within the universe of the twelve, thirteen,
or fifteen surveys on the topic. Gimme a break.
> When you compare states and provinces across that boundary line, you
> find a lot of interesting differences, and a lot of interesting
> similarities. Prairie provinces are often comparable in murder rates
> to their American neighbors. New York State, of course, is terribly,
> terribly violent--but then again, it's a restrictive state not just about
> concealed weapon permits, but even about permission to have a handgun
> in your own home.
The three Prairie Provinces are very different from each other in
culture and in ethnic makeup. I would expect there to be not much
difference between Alberta and the Dakotas simply because the settlers
are the same people -- refugees from Bleeding Kansas in the 1840's.
Saskatchewan and Minnesota, in the same vein are in some ways similar.
Manitoba is a very complicated scene from any point of view.
I reject entirely your facile suggestion -- you didn't dare say it, you
just suggested it -- that New York is violent because it has laws
against guns in the home. From Canada's point of view the most violent
part of New York is all those Al Quaeda guys in the Buffalo suburbs --
and I doubt that they've been there long enough to read the law in English.
> Seattle and Vancouver are an interesting comparison, because murder
> rates for whites are about the same; Seattle is very slightly lower
> (6.2/100,000 instead of 6.4 in Vancouver). The big difference is
> that Seattle has a much larger black and Hispanic population, and
> this is most of the difference. Since there is no difference in
> handgun ownership laws or carry permit issuance based on race in
> Seattle, or I presume, in Vancouver, this suggests that the difference
> isn't the laws, but the subcultures that are associated with black
> and Hispanic populations in the U.S.
I think this is quite plausible. Here in Toronto we have a lively
Jamaican immigrant group. Jamaicans as a group are culturally
identifiable -- like Polish Jews or Lebanese: they are natural
succeeders. (In the States think Colin Powell -- and notice that
Jamaican-American incomes are something like, last time I checked, 9%
higher than those of American white Protestants. "But we have to work at
two jobs to do it," a Jamaican-Canadian friend of mine said.)
But in Toronto this highly motivated, probably above-average in
literacy, and generally successful, group also harbour a hugely
disproportionate amount of gun violence.
Seems to me one useful step is to take away their guns.
>>The comparison is not at all asinine: they are two guys who were so sure
>>of their conclusions that they didn't think mere facts were needed. The
>>prototype of these guys was Sir Cyril Burt who just *knew* that
>>intelligence was normally distributed, and it was so obvious that it
>>wasn't worth the trouble of testing. So he just pencilled in the
>>invented data on his charts and published it that way. Lott, like
>>Bellesiles, is cut from the same cloth.
>>
> It is now established that the 1997 survey did take place; at least
> one person who was surveyed has come forward. Professor Lindgren,
> my fellow inquisitor of Bellesiles, has interviewed the survey
> subject (a lawyer, and an assistant district attorney a few years
> back), and concluded that he is a credible witness.
It is established that one person has phoned in and said they took a
survey some time ago, and it might have been Lott's.
I think far more people have phoned in in support of OJ.
To my mind the thing that Lott needs to explain is this: how come both,
giggle, surveys are so different in their findings from the dozen plus
surveys taken by people who were actually able to keep track of their
tally sheets, their questions, their students?
There's presumably a relatively real objective reality out there, where
people pull guns on burglars, in bar fights, on people who try to rob
them, and so forth. Everybody who surveys these people and asks how
often they actually shoot the gun gets fairly scary numbers, in the 26%
to 67% range (is it?).
Lott, twice by his claim, gets a wildly different view of reality. 2%.
How come?
How, with so very very few people (not the supposed 2,000 and the 1,000,
but the small percentage of these who said they had drawn guns) did he
get such mutually similar results, while everybody else's results are so
different? It's like firing a rifle from a merry-go-round and hitting
the number "2" on two successive tries. And while the numbers everybody
else hits are nowhere near two.
Not credible. Even if "2%" were the real number, on his sample sizes you
would expect to get 1 one time and 4 the other, or something like that.
To get "2" both times defies belief.
Clayton, you are a guy with high standards. Why do you attach your own
credibility to a flako like Lott?
Best wishes,
-dlj.
Clayton E. Cramer
> Clayton E. Cramer asks me:
>
> > Do you seriously suggest that the only differences between the two
> > countries over a century are concealed weapon permit availability?
>
> Uh, no, Cramer, I don't seriously suggest that this is the only
> difference between the two countries over a century. (and I have no clue
> what you think is plural in your sentence above.)
>
> Have you considered the possibility that this is an astonishingly stupid
> question to ask? And that you incomprehensible grammar is caused by a
> flush of high emotion caused by the montion of guns?
Here's what you wrote:
>> Nobody but a few gun nuts has ever taken Lott's book seriously because
>> the actual experiment has been carried out on a large scale for a
>> century: Canada vs. The United States side by side.
We are discussing Lott's claim that concealed weapon permit availability
reduces violent crime. You respond that "the actual experiment" has
been carried out on a large scale with Canada and the U.S. "side by
side." What did you mean? If this is an experiment, that implies that
there are similar circumstances on both sides of the border, with the
controlled variable being concealed weapon permits.
There are LOTS of other differences, and even concealed weapon permit
issuance, for most of the 20th century, was discretionary.
> > Remember that until 1961, there was NO American state that had non-
> > discretionary issuance of concealed weapon permits. Nearly all of
> > this took place in the 1980s and 1990s.
>
> So? Even if you take Lott seriously, there's very little difference
> before and after in any of his many counties. The differences between
Yup. Low violence counties aren't the major beneficiaries of
non-discretionary permit issuance; it's high violence counties that
are the beneficiaries.
> Canada and the US in both crime and in gun violence are striking -- at
> the 98% level of confidence. Lott's claim, by contrast, is that he has
Do you honestly not realize that there are a stack of other differences
besides concealed weapon permit issuance?
> carried out two surveys which are accurate within 1,300% 50% of the time
> or 3,000% all of the time within the universe of the twelve, thirteen,
> or fifteen surveys on the topic. Gimme a break.
He has carried out two surveys that give similar results. Within the
confidence interval appropriate to such a tiny sample size (the subgroup
that used a gun for self-defense), these numbers overlap most of the
other surveys. You might make the argument that the sample size is so
small as to be meaningless, and I would agree that by itself, these
surveys don't tell us much about shots fired in self-defense, but they
do confirm the other surveys about the percentage of Americans who
use a gun each year in self-defense.
> > When you compare states and provinces across that boundary line, you
> > find a lot of interesting differences, and a lot of interesting
> > similarities. Prairie provinces are often comparable in murder rates
> > to their American neighbors. New York State, of course, is terribly,
> > terribly violent--but then again, it's a restrictive state not just about
> > concealed weapon permits, but even about permission to have a handgun
> > in your own home.
>
> The three Prairie Provinces are very different from each other in
> culture and in ethnic makeup. I would expect there to be not much
> difference between Alberta and the Dakotas simply because the settlers
> are the same people -- refugees from Bleeding Kansas in the 1840's.
> Saskatchewan and Minnesota, in the same vein are in some ways similar.
> Manitoba is a very complicated scene from any point of view.
And your point is? Gun control laws are dramatically looser on the
U.S. side of the border, and yet murder rates in states like Montana,
Idaho, North Dakota, aren't dramatically higher than their Canadian
province neighbors. Why do you suppose that is?
> I reject entirely your facile suggestion -- you didn't dare say it, you
> just suggested it -- that New York is violent because it has laws
> against guns in the home. From Canada's point of view the most violent
I didn't say it, because I don't believe it. Except at the extremes
(complete and effective bans on guns, no regulation whatsoever), gun
control makes pretty small differences in violent crime rates--as
_More Guns, Less Crime_ points out, since the reductions caused by
non-discretionary permit are statistically significant, but fairly
small--even though large numbers of people get permits.
> part of New York is all those Al Quaeda guys in the Buffalo suburbs --
> and I doubt that they've been there long enough to read the law in English.
If that's really "Canada's point of view" then Canadians must be very
stupid. The al-Qaeda guys in the Buffalo suburbs haven't been charged
with ANY violent crimes. Most of the violent crime in New York is in
New York City, and has been for decades.
> > Seattle and Vancouver are an interesting comparison, because murder
> > rates for whites are about the same; Seattle is very slightly lower
> > (6.2/100,000 instead of 6.4 in Vancouver). The big difference is
> > that Seattle has a much larger black and Hispanic population, and
> > this is most of the difference. Since there is no difference in
> > handgun ownership laws or carry permit issuance based on race in
> > Seattle, or I presume, in Vancouver, this suggests that the difference
> > isn't the laws, but the subcultures that are associated with black
> > and Hispanic populations in the U.S.
>
> I think this is quite plausible. Here in Toronto we have a lively
> Jamaican immigrant group. Jamaicans as a group are culturally
> identifiable -- like Polish Jews or Lebanese: they are natural
> succeeders. (In the States think Colin Powell -- and notice that
> Jamaican-American incomes are something like, last time I checked, 9%
> higher than those of American white Protestants. "But we have to work at
> two jobs to do it," a Jamaican-Canadian friend of mine said.)
>
> But in Toronto this highly motivated, probably above-average in
> literacy, and generally successful, group also harbour a hugely
> disproportionate amount of gun violence.
>
> Seems to me one useful step is to take away their guns.
Just from Jamaicans? Of course not, that would be racist! So why
disarm the vast majority of Canadians that are not violent, and for
whom gun ownership is not a risk factor? Why not try to correct the
cultural problems that cause Jamaican immigrants to be so violent?
> > It is now established that the 1997 survey did take place; at least
> > one person who was surveyed has come forward. Professor Lindgren,
> > my fellow inquisitor of Bellesiles, has interviewed the survey
> > subject (a lawyer, and an assistant district attorney a few years
> > back), and concluded that he is a credible witness.
>
> It is established that one person has phoned in and said they took a
> survey some time ago, and it might have been Lott's.
1. There were no other surveys of defensive gun use in progress at
the time; at least, no one can find anything published by anyone else
that asked those questions.
2. The lawyer who has come forth remembered details of the questions
and the interviewers that were consistent with Lott's survey.
3. As Professor Lindgren points out, if 100,000 people in the U.S.
were aware of this controversy, one person out of the survey group
coming forward would statistically the most likely situation.
> I think far more people have phoned in in support of OJ.
There is other evidence that suggests that such a survey took place.
It's not terribly persuasive, but in combination with one of the
surveyed people coming forward, I would say that it is enough to
overwhelm the not terribly persuasive evidence the other way.
> To my mind the thing that Lott needs to explain is this: how come both,
> giggle, surveys are so different in their findings from the dozen plus
> surveys taken by people who were actually able to keep track of their
> tally sheets, their questions, their students?
Do you honestly not realize how survey questions, methods, sampling
techniques, can all influence the results of a survey? The NCVS
data shows the lowest number of defensive gun uses of all major
studies, but no one suggests that this is the result of fraud. It is
because one of the questions asks if the respondent was the victim of a
crime, and only if they say yes are they asked how they responded to
that crime. A person who used a gun defensively may not perceive
that they were the victim of a crime--after all, the bad guy stopped
before they were hurt or lost anything of value.
> There's presumably a relatively real objective reality out there, where
> people pull guns on burglars, in bar fights, on people who try to rob
> them, and so forth. Everybody who surveys these people and asks how
> often they actually shoot the gun gets fairly scary numbers, in the 26%
> to 67% range (is it?).
Why are these numbers scary? Because the bad guys might get shot?
By the way, I understand that there are only two surveys besides Lott's
surveys that provide numbers on guns fired; most only ask if the gun
was used defensively.
> Lott, twice by his claim, gets a wildly different view of reality. 2%.
>
> How come?
Confidence intervals are tiny for such small subgroups as those who
fired a gun.
> How, with so very very few people (not the supposed 2,000 and the 1,000,
> but the small percentage of these who said they had drawn guns) did he
> get such mutually similar results, while everybody else's results are so
> different? It's like firing a rifle from a merry-go-round and hitting
> the number "2" on two successive tries. And while the numbers everybody
> else hits are nowhere near two.
See above.
> Not credible. Even if "2%" were the real number, on his sample sizes you
> would expect to get 1 one time and 4 the other, or something like that.
> To get "2" both times defies belief.
That's not my understanding of what he got.
> Clayton, you are a guy with high standards. Why do you attach your own
> credibility to a flako like Lott?
>
> Best wishes,
>
> -dlj.
Because so far you haven't shown that you really understand the issue of
survey methodology well enough to intelligently criticize the results.
Christopher Auld
>Have you considered the possibility that this is an astonishingly stupid
>question to ask? And that you incomprehensible grammar is caused by a
^^^
>flush of high emotion caused by the montion of guns?
^^^^^^^
This is the sort of statement that makes wasting time on
usenet almost worthwhile.
>Clayton, you are a guy with high standards. Why do you attach your own
>credibility to a flako like Lott?
A priori, I would tend to be skeptical of Lott's results.
But when the chatter of his critics consists of statistically
illiterate, ideological nonsense and cheap ad hominem, it
does give pause.
Stephen J Fromm
"susupply" <susu...@mindspring.com> wrote in message
news:b0k1mg$fbc$1...@slb6.atl.mindspring.net...
The issue of fraud is a sideshow to the far more damaging point that Lott's
methodology is, to put it charitably, weak. See Lambert's posts, in
particular where he describes fitting curves to randomly generated data.
susupply
"Stephen J Fromm" <stephe...@verizon.net>
flexible, if not convincing,
wrote in message news:plDX9.570$Hz....@nwrddc02.gnilink.net...
> The issue of fraud is a sideshow to the far more damaging point that
Lott's
> methodology is, to put it charitably, weak. See Lambert's posts, in
> particular where he describes fitting curves to randomly generated data.
Well, we've just seen Lambert made mincemeat of by sci.econ's favorite
econometrician. However, this thread began with YOU posting this:
<<---------------------------------------------
From: Stephen J. Fromm (stephe...@verizon.net)
Subject: John Lott-
Newsgroups: sci.econ
Date: 2003-01-14 14:01:50 PST
Did John Lott commit academic fraud?
http://markarkleiman.blogspot.com/2003_01_01_markarkleiman_archive.html#9018
1328
-------------------------------------------->>
David Friedman
> The
> prototype of these guys was Sir Cyril Burt who just *knew* that
> intelligence was normally distributed, and it was so obvious that it
> wasn't worth the trouble of testing. So he just pencilled in the
> invented data on his charts and published it that way.
You might want to read _Cyril Burt: Fraud or Framed_, which is, I
think, the most recent book summing up the Burt controversy. Your
description isn't even close to accurate. That particular accusation
was rebutted, as best I recall, by two statisticians, one from Chicago
(probably Stephen Stigler, who I know was involved in parts of the
controversy) and one from Harvard, who pointed out that Burt's
normalization of the data was a perfectly normal procedure, and that
he said in his article that he was doing it. No pencilling in of
invented data.
There were a bunch of other accusations, all made after Burt had died
and his papers had been destroyed--by his housekeeper, on the advice
of another professor in the field. The relevant professional
association was sufficiently convinced to issue a public statement
denouncing Burt. Some years later two authors independently wrote
books defending Burt and debunking the attacks, and were sufficiently
convincing so that the association (I think the British Psychology
Association or some similar title) withdrew its denunciation.
The authors of the Fraud or Framed book, which I read a good deal of
recently and which consists of about seven different articles by
different authors, conclude that many of the charges against Burt are
either clearly wrong or very likely wrong, that in some cases there is
evidence of fraud, but not conclusive evidence.
So far as Lott is concerned, I haven't followed the current thread but
as far as I can tell there is no evidence at all of fraud with regard
to the original study. He made his data available to other scholars,
they analyzed it in a variety of different ways, some agreed with his
conclusion, some disagreed. The disagreement wasn't of the form "his
data were wrong" or "he lied about the results" but "we think a
different way of analyzing the data gives a more reliable result." The
accusations of dishonesty or incompetence that I saw, mostly in the
early stage of the dispute when I was paying attention to it, were
themselves pretty clearly dishonest and being made by people with a
clear axe to grind. You can find a discussion of some of them on m web
page, but I haven't tried to keep up with the later controversies.
What bothers me about the whole business is that Lott has ended up
spending a lot of time and energy defending his old work instead of
producing new work on other subjects--if you look at his vita, he was
a very productive scholar writing on a range of subjects. There is
much to be said for a scholarly strategy of fire and forget--publish
your work and let other people fight over it. But of course it's hard
to follow that advice when other people are calling you a crook.
David Lloyd-Jones
> David Lloyd-Jones <d...@rogers.com> wrote
>
>>The
>>prototype of these guys was Sir Cyril Burt who just *knew* that
>>intelligence was normally distributed, and it was so obvious that it
>>wasn't worth the trouble of testing. So he just pencilled in the
>>invented data on his charts and published it that way.
>
>
> You might want to read _Cyril Burt: Fraud or Framed_, which is, I
> think, the most recent book summing up the Burt controversy. Your
> description isn't even close to accurate. That particular accusation
> was rebutted, as best I recall, by two statisticians, one from Chicago
> (probably Stephen Stigler, who I know was involved in parts of the
> controversy) and one from Harvard, who pointed out that Burt's
> normalization of the data was a perfectly normal procedure, and that
> he said in his article that he was doing it. No pencilling in of
> invented data.
David,
I'm not sure I stand corrected here. You say he "normalized the data"
and said openly that he was doing so. And you add that this was a
perfectly, ahem, normal statistical procedure.
Your problem is that it is only a perfectly normal statistical procedue
if normalization makes sense, i.e. if the phenomenon being modelled is
bell-curve distributed. But this is not the case.
Even models of intelligence have to be both fat-tailed and fractal to
approximate the real world. The real world differs from the models in
having, at minimum, random variation on top of this.
Have your authors considered the possibility that by normalizing his
allegedly genuine data Burt postponed the discovery of the ways in which
the models he used delayed the creation of better models?
Nor am I sure that what you say in Lott's defence is true. Certainly he
has been normal in sharing data where that data exists. Are you claiming
that he planned on being equally generous with the rest of his data had
the dog not eaten it?
* * *
An aside: Lott's defenders in the blogs, and the excellent Clayton
Cramer here, have defended "both" Lott's studies by saying that a world
of 98% brandishers is consistent with his paltry data because one
responder is within the range of data such a world could produce at high
confidence levels given the small numberes he produced and claimed to
have produced.
This is the logical fallacy of post hoc or of affirming the consequent.
"If the Moon were made out of green cheese, we would have dairy products
on sale in all our supermarkets. We have dairy products in all our
supermarkets, therefore the Moon is made out of green cheese."
The fact that Lott's responder when a fractional person is expected by
his 2% claim does not in any way support his 2% claim just because it is
non-contradictory of it: it is non-contradictory of a whole lot of other
claims, including those made by the mass of the research. This
non-contradictoryness does not support a propter hoc claim that it is
supportive of his claims being legitimately different from those of all
other research.
-dlj.
Stephen J Fromm
>
> "Stephen J Fromm" <stephe...@verizon.net>
>
> flexible, if not convincing,
>
> wrote in message news:plDX9.570$Hz....@nwrddc02.gnilink.net...
>
> > The issue of fraud is a sideshow to the far more damaging point that
> Lott's
> > methodology is, to put it charitably, weak. See Lambert's posts, in
> > particular where he describes fitting curves to randomly generated data.
>
> Well, we've just seen Lambert made mincemeat of by sci.econ's favorite
> econometrician. However, this thread began with YOU posting this:
And whose sci.econ's favorite econometrician?
You seem to like to appeal to authority ("sci.econ's favorite
econometrician", certain right-wing economists) instead of debating the
issues.
Lambert's point regarding Lott's procedure of fitting curves, which was also
mentioned by Ehrlich in his book, hasn't been convincingly rebutted, AFAICT,
your appeal to authority notwithstanding.
And noticed I posed the *question* of whether Lott committed academic fraud.
I didn't say I thought this was more important than the general shoddiness
of his work.
And why should apologies be given? Anyone using shoddy methodology like
Lott, who moreover doesn't take proper care with archiving his data, is
going to be more vulnerable to accusations of fraud.
Christopher Auld
>I'm not sure I stand corrected here. You say he "normalized the data"
>and said openly that he was doing so. And you add that this was a
>perfectly, ahem, normal statistical procedure.
>
>Your problem is that it is only a perfectly normal statistical procedue
>if normalization makes sense, i.e. if the phenomenon being modelled is
>bell-curve distributed. But this is not the case.
>
>Even models of intelligence have to be both fat-tailed and fractal to
>approximate the real world. The real world differs from the models in
>having, at minimum, random variation on top of this.
This is statistical gibberish. DLJ has no idea what
he's talking about.
David Lloyd-Jones
> David Lloyd-Jones <d...@rogers.com> wrote:
>
>
>>I'm not sure I stand corrected here. You say he "normalized the data"
>>and said openly that he was doing so. And you add that this was a
>>perfectly, ahem, normal statistical procedure.
>>
>>Your problem is that it is only a perfectly normal statistical procedue
>>if normalization makes sense, i.e. if the phenomenon being modelled is
>>bell-curve distributed. But this is not the case.
>>
>>Even models of intelligence have to be both fat-tailed and fractal to
>>approximate the real world. The real world differs from the models in
>>having, at minimum, random variation on top of this.
>
>
> This is statistical gibberish. DLJ has no idea what
> he's talking about.
Guess for poor Chris's sake I shouldn't elide. To add the missing bit in
the second paragraph:
It only makes sense to fix up your data to make it look more like a bell
shaped curve if the model in which you are doing your normalizing is
reflective of a reality in which the facts on which you have data are
really distributed that way.
To fill out the third pargraph for poor Chris:
Intelligence in the real world is not dristributed normally, as Burt
assumed and forcefully propounded. The curve of data reflecting real
world intelligence has more values at the high and low extremes than are
generated in the abstract by a bell curve, and the curve is wildly
irregular as you study intelligence in greater detail: all the
underlying values are distributed in what look like ragged, probably
fractal, curves, not smooth distributions.
If there's anything else you need spelled out in greater detail, Chris,
don't hesitate to ask.
-dlj.
Stephen J. Fromm
Nonsense. Ehrlich, author of _Nine Crazy Ideas in Science_, has no
obvious axe to grind.
Lott's either incompetent or dishonest or both.
> What bothers me about the whole business is that Lott has ended up
> spending a lot of time and energy defending his old work instead of
> producing new work on other subjects--if you look at his vita, he was
> a very productive scholar writing on a range of subjects. There is
> much to be said for a scholarly strategy of fire and forget--publish
> your work and let other people fight over it. But of course it's hard
> to follow that advice when other people are calling you a crook.
It's good that he's spending his time that way, as many people have
pointed out flaws in his work. See my posts on Ehrlich and Lambert's
posts on Lott.
Stephen J. Fromm
> David Lloyd-Jones <d...@rogers.com> wrote:
>
> >Have you considered the possibility that this is an astonishingly stupid
> >question to ask? And that you incomprehensible grammar is caused by a
> ^^^
> >flush of high emotion caused by the montion of guns?
> ^^^^^^^
>
> This is the sort of statement that makes wasting time on
> usenet almost worthwhile.
>
>
> >Clayton, you are a guy with high standards. Why do you attach your own
> >credibility to a flako like Lott?
>
> A priori, I would tend to be skeptical of Lott's results.
> But when the chatter of his critics consists of statistically
> illiterate, ideological nonsense and cheap ad hominem, it
> does give pause.
ROTFL!
Lambert wrote, for example,
"It is no surprise that there is a peak when the laws were
passed--this is one of the few places where it is possible
for the fitted curves to peak. Even if the crime rate started
to decline before the laws passed, Lott's diagram could still
show a peak coinciding with the law.
I ran some experiments by fitting a similar pair of quadratic
curves to a sequence of random numbers. Almost always the
curves seemed to show that something had happened at the
junction of the two curves, even though nothing had."
Ehrlich made similar points in his book.
Is that point "statistically illiterate"?
Robert Vienneau
> au...@acs.ucalgary.ca (Christopher Auld) wrote in message
> news:<b0mmg9$7r...@acs4.acs.ucalgary.ca>...
>> [Silliness deleted.]
> Lambert wrote, for example,
> "It is no surprise that there is a peak when the laws were
> passed--this is one of the few places where it is possible
> for the fitted curves to peak. Even if the crime rate started
> to decline before the laws passed, Lott's diagram could still
> show a peak coinciding with the law.
>
> I ran some experiments by fitting a similar pair of quadratic
> curves to a sequence of random numbers. Almost always the
> curves seemed to show that something had happened at the
> junction of the two curves, even though nothing had."
>
> Ehrlich made similar points in his book.
>
> Is that point "statistically illiterate"?
Poor Chris Auld has been demonstrating for years that he doesn't
know how to raise points without extreme silliness. Here's
what he should be asking or saying. (I haven't read Ehrlich
or Lott.)
When saying that the curves in this simulation peak, are you
relying on a visual impression of the graphs alone? Or do you
conduct a test of statistical significance, such as an F-test?
If the latter, was the simulated data synthesized under the
model described by the null hypothesis? Or was there some
sort of model mismatch?
If there was a formal test of statistical significance and
no model mismatch, many would find these results surprising.
Do you have any explanation of your results that addresses
these points?
Apparently, Lott has lied. Supposedly he had been saying he
does not participate in Usenet discussion groups.
By the way, I like the novel Primary Colors. But I don't know
that Joe Klein gave it a positive review on Amazon.
Christopher Auld
>It only makes sense to fix up your data to make it look more like a bell
>shaped curve if the model in which you are doing your normalizing is
>reflective of a reality in which the facts on which you have data are
>really distributed that way.
There is no natural metric for "intelligence." The proxies
that are used are usually, as in the case at hand, based on
tests. The scores are not comparable across tests and there
is no reason to believe that given differences in test scores
preserve cardinality when mapped into whatever "intelligence"
is. Thus, it is common to rescale these scores. One manner
of rescaling is simply to report the individuals' percentiles.
Another manner of presenting the same information is to
express those percentiles rescaled using the function F(p),
where F() denotes the normal distribution function with mean
100 and standard deviation 15 and p gives percentiles.
That is, as was and is common in psychometrics, Burt (as he
reported) simply rescaled his data such that they were normal.
It makes no sense to allege that the data "aren't really
distributed that way" because this measure _defines_
intelligence such that it is by construction distributed that
way. The letter of Stigler and Rubin in _Science_ points this
out quite clearly.
>generated in the abstract by a bell curve, and the curve is wildly
>irregular as you study intelligence in greater detail: all the
>underlying values are distributed in what look like ragged, probably
>fractal, curves, not smooth distributions.
I can't read this as anything other than gibberish, but I'm
willing to give David the benefit of the doubt. What
"underlying values" are you talking about? If I had a vector
of observations on those values, how would I go about figuring
out if its distribution "looks like ragged, probably fractal,
curves?" Let's be more concrete and suppose we know V's
density, say f(V). Please provide a formal answer in terms of
the properties of f(V).
David Lloyd-Jones
>
> There is no natural metric for "intelligence." The proxies
> that are used are usually, as in the case at hand, based on
> tests. The scores are not comparable across tests and there
> is no reason to believe that given differences in test scores
> preserve cardinality when mapped into whatever "intelligence"
> is.
Thank you. It's nice to know that you can make contact with the real
world from time to time, Chris.
<snips>
> I can't read this as anything other than gibberish, but I'm
> willing to give David the benefit of the doubt. What
> "underlying values" are you talking about? If I had a vector
> of observations on those values, how would I go about figuring
> out if its distribution "looks like ragged, probably fractal,
> curves?" Let's be more concrete and suppose we know V's
> density, say f(V). Please provide a formal answer in terms of
> the properties of f(V).
Not going to play that game, Chris. You've switched from talking about
intelligence or intelligences, qualities of human beings, to talking
about a single vector measuring a restricted set of those in a small set
of people, just once each, at one particular time. Nobody doubts that
you can "normalise" that set of numbers to fit any curve you choose,
bell curve or anything else.
However if you make up an "intelligence test," and frankly say to anyone
who asks that intelligence is defined as being what this test measures,
and you normalise the scoring of the values of the individual questions
on this test over a very large group of people so that they come out in
your bell curve or other curve of choice -- and then you go and apply it
to a different group of people, or the same group of people at a
different time, you won't get the same shape of curve for the second set
of results.
The shape of the data is not the shape of any underlying distribution of
qualities. Ever.
-dlj.
David Friedman
David Lloyd-Jones <d...@rogers.com> wrote:
> David Friedman wrote:
> > David Lloyd-Jones <d...@rogers.com> wrote
> >
> >>The
> >>prototype of these guys was Sir Cyril Burt who just *knew* that
> >>intelligence was normally distributed, and it was so obvious that it
> >>wasn't worth the trouble of testing. So he just pencilled in the
> >>invented data on his charts and published it that way.
> >
> >
> > You might want to read _Cyril Burt: Fraud or Framed_, which is, I
> > think, the most recent book summing up the Burt controversy. Your
> > description isn't even close to accurate. That particular accusation
> > was rebutted, as best I recall, by two statisticians, one from Chicago
> > (probably Stephen Stigler, who I know was involved in parts of the
> > controversy) and one from Harvard, who pointed out that Burt's
> > normalization of the data was a perfectly normal procedure, and that
> > he said in his article that he was doing it. No pencilling in of
> > invented data.
>
> David,
>
> I'm not sure I stand corrected here. You say he "normalized the data"
> and said openly that he was doing so. And you add that this was a
> perfectly, ahem, normal statistical procedure.
Actually, I was repeating from memory the summary of the views of two
reputable statisticians on the subject. I haven't read the Burt article
in question, and I presume from your comments later that you haven't
either, at least assuming it is the article with regard to which I have
seen the normalization charge made.
> Your problem is that it is only a perfectly normal statistical procedue
> if normalization makes sense, i.e. if the phenomenon being modelled is
> bell-curve distributed. But this is not the case.
>
> Even models of intelligence have to be both fat-tailed and fractal to
> approximate the real world. The real world differs from the models in
> having, at minimum, random variation on top of this.
>
> Have your authors considered the possibility that by normalizing his
> allegedly genuine data Burt postponed the discovery of the ways in which
> the models he used delayed the creation of better models?
As best I recall Stigler's comments (which I was seeing cited by one of
the authors in the book), he had reservations about various features of
the Burt article, but none of them had anything to do with fraud and
they were not the features which (if I remember correctly) Burt's
biographer and others offered as evidence of fraud.
> Nor am I sure that what you say in Lott's defence is true. Certainly he
> has been normal in sharing data where that data exists. Are you claiming
> that he planned on being equally generous with the rest of his data had
> the dog not eaten it?
As I wrote in the post you are responding to, and you didn't quote:
"So far as Lott is concerned, I haven't followed the current thread but
as far as I can tell there is no evidence at all of fraud with regard
to the original study."
If the dog ate the data in his original study I haven't heard about it.
So far as the rest of your post is concerned, since I haven't followed
the current controversy I have no opinion on it. I came into this thread
not because you were talking about Lott but because I had just read a
book about Burt and was curious as to whether people were discussing him
online.
I note, however, that you made a factual assertion about Burt's
misdeeds, and so far as I can tell you have no support for it--merely
arguments about how he might have misanalyzed his data, none of which
have anything to do with the claim that he:
"just pencilled in the invented data on his charts and published it
that way."
I suggest that you either come up with evidence of invented data that
Burt pencilled into a chart or concede that you made a nasty allegation
about Burt without knowing anything about its truth beyond vague
memories of second or third hand accusations.
I also suggest that you read the book, since it's an interesting story.
susupply
"Stephen J Fromm" <stephe...@verizon.net>
before saying "never mind" his original contribution,
wrote in message news:m4TX9.123$3J....@nwrddc01.gnilink.net...
> You seem to like to appeal to authority ("sci.econ's favorite
> econometrician", certain right-wing economists) instead of debating the
> issues.
>
> Lambert's point regarding Lott's procedure of fitting curves, which was
also
> mentioned by Ehrlich in his book, hasn't been convincingly rebutted,
AFAICT,
> your appeal to authority notwithstanding.
You seem to have a penchant for not following threads YOU created. Lambert
has been made to look silly (and not for the first time on sci.econ), and
has had no response to the refutations of his claims:
<<---------------quote--------------------
[Tim Lambert had written:]
> >I ran some experiments by fitting a similar pair of quadratic
> >curves to a sequence of random numbers. Almost always the
> >curves seemed to show that something had happened at the
> >junction of the two curves, even though nothing had.
> >
> >Lott has confused the fitted curves with the actual data.
>
> Lott's models are perfectly standard, as is his interpretation.
> Ignoring quadratic terms, the models in chapter 4 take the form
>
> y = Xb + ad + b(dt) + noise,
>
> where d is a dummy indicating a policy change and t is a
> time trend. If one generates a "sequence of random numbers"
> with the property that nothing happens in response to the
> policy, one will not "almost always [find] the curves show
> that something happened at the junction." One will in fact
> find, with reasonably large samples, that the coefficient on
> the interaction term is statistically insignificant at a
> rate equal to the size of the test, conventionally 5%. What
> is more, most of Lott's models are less restrictive, treating
> the impact of a policy change in a semiparametric manner by
> using sets of dummies capturing time to/since the policy
> change. Again, this is all very standard and uncontroversial.
>
> --
> Chris Auld
> Department of Economics
> University of Calgary
> au...@ucalgary.ca
------------------endquote-------------------->>
I also missed YOUR response to Prof. Auld's specifics.
> And noticed I posed the *question* of whether Lott committed academic
fraud.
> I didn't say I thought this was more important than the general shoddiness
> of his work.
>
> And why should apologies be given? Anyone using shoddy methodology like
> Lott, who moreover doesn't take proper care with archiving his data, is
> going to be more vulnerable to accusations of fraud.
And we've just seen that the "shoddy methodology" charge is silly. Just as
we saw Ehrlich's claim refuted by Lott himself.
But if you don't want to look similarly stupid in the future making claims
like:
> > Did John Lott commit academic fraud?
You might want to preface your rhetorical questions with something like,
"Not that it matters".
Stephen J Fromm
>
> "Stephen J Fromm" <stephe...@verizon.net>
>
> before saying "never mind" his original contribution,
>
> wrote in message news:m4TX9.123$3J....@nwrddc01.gnilink.net...
> > You seem to like to appeal to authority ("sci.econ's favorite
> > econometrician", certain right-wing economists) instead of debating the
> > issues.
> >
> > Lambert's point regarding Lott's procedure of fitting curves, which was
> also
> > mentioned by Ehrlich in his book, hasn't been convincingly rebutted,
> AFAICT,
> > your appeal to authority notwithstanding.
>
> You seem to have a penchant for not following threads YOU created.
Lambert
> has been made to look silly (and not for the first time on sci.econ), and
> has had no response to the refutations of his claims:
Balogne. You're just recycling what Auld wrote. Do you actually *know*
anything about math or statistics, or do you just pretend to do so?
See above. Lambert/Ehrlich are specifically referring to fitting two
different curves, one on each side of the point of interest. Nothing in
your clip above refutes that.
Again, what do *you* know about statistics? Seems like you like citing
authority to hide your own ignorance.
> > And noticed I posed the *question* of whether Lott committed academic
> fraud.
> > I didn't say I thought this was more important than the general
shoddiness
> > of his work.
> >
> > And why should apologies be given? Anyone using shoddy methodology like
> > Lott, who moreover doesn't take proper care with archiving his data, is
> > going to be more vulnerable to accusations of fraud.
>
> And we've just seen that the "shoddy methodology" charge is silly. Just
as
> we saw Ehrlich's claim refuted by Lott himself.
Uh huh. And you're able to independently evaluate (as opposed to just
recycling what other people write) this charge because...?
> But if you don't want to look similarly stupid in the future making claims
> like:
>
> > > Did John Lott commit academic fraud?
>
> You might want to preface your rhetorical questions with something like,
> "Not that it matters".
What's silly?
(1) Lott has been shown by several independent people to have shoddy
methodology.
(2) Lott apparently doesn't take care archiving his data.
This kind of thing could easily lead to accusations of possible impropriety.
Christopher Auld
>> curves?" Let's be more concrete and suppose we know V's
>> density, say f(V). Please provide a formal answer in terms of
>> the properties of f(V).
>Not going to play that game, Chris. You've switched from talking about
>intelligence or intelligences, qualities of human beings, to talking
>about a single vector measuring a restricted set of those in a small set
>of people, just once each, at one particular time.
I am not playing a "game," David, I am asking you to clearly explain
the statistical properties you have described. Either you are talking
gibberish or are you are referring to results which I would be interested
in knowing about for professional reasons. Now, I fully expect, given
your history on sci.econ, that the former is the case, but here's your
second chance to surprise everyone. You said:
>> the curve is wildly
>>irregular as you study intelligence in greater detail: all the
>>underlying values are distributed in what look like ragged, probably
>>fractal, curves, not smooth distributions.
I am asking you to be more specific as to what you mean by this. Suppose
I have a set of "underlying values." I want to know whether these
values "are distributed in what look like ragged, probably fractal, curves,
not smooth distributions." Please clearly tell me precisely what this
means in terms of the density function of the "underlying values."
>However if you make up an "intelligence test," and frankly say to anyone
>who asks that intelligence is defined as being what this test measures,
>and you normalise the scoring of the values of the individual questions
>on this test over a very large group of people so that they come out in
>your bell curve or other curve of choice -- and then you go and apply it
>to a different group of people, or the same group of people at a
>different time, you won't get the same shape of curve for the second set
>of results.
Several issues are conflated here, and these issues are largely unrelated
to the topic at hand. You charged that Burt "fixed up his data to make
it look more like a bell curve." What Burt did was a use a perfectly
standard rescaling under which _by definition_ the measure is normally
distributed. Two distinct questions include: (1) If this procedure were
repeated using a different sample, would the distribution look the same?
and (2) if repeated measurements were taken, would the subsequent distributions
look the same? Neither of these questions has anything to do with whether
the "underlying qualities" are normal or whether what Burt did was
fraudulent. The answer to the first question is: If both samples are
drawn from the same process, then both curves will asymptotically "look
the same," by the Fundamental Theorem of Statistics. The answer to the
second question is trickier. But if the second set of measurements are
not Gaussian, it follows that the unscaled values also have a different
distribution. That is, the distribution of the "underlying values" would
have to change as well.
>The shape of the data is not the shape of any underlying distribution of
>qualities. Ever.
Again, using the normal distribution with mean 100 and standard deviation
15 is a rescaling commonly used to describe the results of intelligence tests.
"The shape of the underlying distribution of qualities" depends on the units
in which "qualities" are measured. As I have explained, there is no natural
metric for "intelligence," so reporting results of tests of cognitive ability
on a Gaussian curve is no less valid or meaningful than reporting, say,
percentiles. This is very common procedure in psychometrics, and Burt's use
of it was not nonstandard, much less fraudulent.
The history of this charge against Burt is worth noting. In 1978 a
psychologist named Doffman published a piece in _Science_ making several
allegations, including the one David keeps repeating. This piece was
widely reported in the popular media. Shortly, and amongst other forceful
refutations of Doffman, _Science_ published letters from the eminent
statisticians Stigler and Rubin who pointed out quite clearly that these
particular charges against Burt -- definitely including the one David blithely
repeats -- were completely unfounded. Apparently, these responses were not
repeated in the mass media.
Christopher Auld
>different curves, one on each side of the point of interest. Nothing in
>your clip above refutes that.
Of course it does. Lott used a perfectly standard regression-based
technique to test for what's called "structural change" in the
econometrics literature. This essentially involves fitting "two
curves" on either side of (in this case, known) break point, and
testing statistically to see if there is a significant difference in
the slopes before and after this point. Lambert claimed that his
simulations showed that it almost always "looks" like something
happens at this point, but he does not address the critical point
that Lott isn't merely eyeballing graphs, he's conducting tests to
see if there is a statistically significant change at that point.
If there is no change, then we will not find a statistically significant
effect except if the data are "flukey," in a formally defined sense,
rather than "almost always."
>(1) Lott has been shown by several independent people to have shoddy
>methodology.
I don't think this is the case. As with all else in academia, Lott's
results will be subject to further scrutiny, and may or may not stand
over time. But his methodology is not "shoddy." On the other hand,
if it turns out he fabricated data, that's much worse than mere shoddy
methods (I tend to think this unlikely, though). On my third hand, I
do find it very disquieting that he would post strident defences of his
own work under an assumed name.
susupply
years in a row.
In the category, "These Foolish Things (Remind Me of You)", the nominee is:
"Stephen J Fromm" <stephe...@verizon.net for:
> > > You seem to like to appeal to authority ....
> > > your appeal to authority ....
> ...Seems like you like citing
> authority to hide your own ignorance.
How could we not honor a guy with such total self-unawareness as to write
the above, intermingled with such gems as:
> > I also missed YOUR response to Prof. Auld's specifics.
>
> See above. Lambert/Ehrlich are ....
and
> Lott has been shown by several independent people to ....
and
> Ehrlich, author of _Nine Crazy Ideas in Science_, ....
and
> See my posts on Ehrlich and Lambert's posts on Lott.
and
>> Lambert wrote, for example....Ehrlich made similar points in his book.
and
> I pointed out in an old post that independent folks like Ehrlich (a
physicist) found ....
and
>> Read Ehrlich's book. The chapter shows that Lott makes fundamental
mistakes ....
and
>> But if you read what Ehrlich is saying, it puts Lott in an extremely
>> bad light. Ehrlich alleges....
>> If this is true, Lott's statistical technique is suspect....
>> Of course, I'd have to have Lott's book in front of me to validate
these claims.
Get that tuxedo dry-cleaned, Stephen.
susupply
"Christopher Auld" <au...@acs.ucalgary.ca> wrote in message
news:b0s0pt$7b...@acs1.acs.ucalgary.ca...
> On my third hand, I
> do find it very disquieting that he would post strident defences of his
> own work under an assumed name.
Yes, that does seem very weird. However, I take into consideration that a
number of malicious nincompoops have spared little effort in a campaign to
destroy his professional reputation. I wouldn't do what Lott did (I think),
but Marquis of Queensbury Rules may not apply when you are fighting the
kinds of people Lott is.
tonyp
"susupply" <susu...@mindspring.com> wrote
> Yes, that does seem very weird. However, I take into consideration
that a
> number of malicious nincompoops have spared little effort in a
campaign to
> destroy his professional reputation. I wouldn't do what Lott did (I
think),
> but Marquis of Queensbury Rules may not apply when you are fighting
the
> kinds of people Lott is.
What kind of person you are depends on, among other things, how you
choose to act "when you are fighting the kinds of people Lott is".
-- Tony P.
Clayton E. Cramer
> David Friedman wrote:
> Even models of intelligence have to be both fat-tailed and fractal to
> approximate the real world. The real world differs from the models in
> having, at minimum, random variation on top of this.
Do you have a graph that demonstrates this? Perhaps I'm in over my
head on this, but my understanding is that the definition of a bell curve
is what happens when you take a sufficiently large population and
distribute some variable characteristic--and that you would be immediately
skeptical of any sampling method that produced a non-bell curve
distribution for any reasonably large sample.
Clayton E. Cramer
> dd...@daviddfriedman.com (David Friedman) wrote in message news:<86041cd5.03012...@posting.google.com>...
> > different way of analyzing the data gives a more reliable result." The
> > accusations of dishonesty or incompetence that I saw, mostly in the
> > early stage of the dispute when I was paying attention to it, were
> > themselves pretty clearly dishonest and being made by people with a
> > clear axe to grind. You can find a discussion of some of them on m web
> > page, but I haven't tried to keep up with the later controversies.
>
> Nonsense. Ehrlich, author of _Nine Crazy Ideas in Science_, has no
> obvious axe to grind.
Does he have an unobvious one? Perhaps he lives in an area where
he feels unsafe at night?
Tim Worstall
Nothing important....but your acknowledgement of the third hand proves 2 things :
1) You are an economist.
2) You are an honest one.
The first coming from the possesion, the second from the admittance :-)
Tim Worstall
David Lloyd-Jones
No, I don't, but the point is obvious, and generally conceded: there are
far more people with IQ's of 200 and of 50 than you can fit under a
curve with mode 100 and standard deviation 15 -- or any other plain bell
curve.
Of course you can fix this -- by chqanging your scoring plan so that the
first few right answers are worth more and the last few are worth less
-- but the phenomenon stays the same if you go to some other "measure"
of intelligence such as speed of nervous system or knowledge of music or
ability to parse Klingon verbs.
-dlj.
Christopher Auld
><p>Chris, <br>
></p>
><p>Here's your density-density correlation function: </p>
[ snip: more gibberish, in .html no less ]
Just to sum up, this:
>the curve is wildly
>irregular as you study intelligence in greater detail: all the
>underlying values are distributed in what look like ragged, probably
>fractal, curves, not smooth distributions.
is gibberish. David has been asked twice to explain what on earth
it means, but he refuses to do so. And this:
>The
>prototype of these guys was Sir Cyril Burt who just *knew* that
>intelligence was normally distributed, and it was so obvious that it
>wasn't worth the trouble of testing. So he just pencilled in the
>invented data on his charts and published it that way.
is an unfounded accusation against Burt, thoroughly refuted in
the literature 25 years ago. DLJ's deeply confused discussion
of the issues hardly overturns Stigler and Rubin's comments on
the allegations against Burt.
Don Libby
>
> David Lloyd-Jones <d...@rogers.com> wrote:
>
> ><p>Here's your density-density correlation function: </p>
>
> [ snip: more gibberish, in .html no less ]
>
> the literature 25 years ago. DLJ's deeply confused discussion
> of the issues hardly overturns Stigler and Rubin's comments on
> the allegations against Burt.
Fascinating discussion, sorry I missed it. Ah well, water under
the bridge, all recorded by Google for progeny to sort out
posthumously I suppose, pip, pip, eh, what?
Just a quick Q for perfesser Auld: if econometricians call
"logit" models "logistic regression" models, do they also call
"tobit" models "tobistic regression" models?
:-)
-dl
Christopher Auld
>Just a quick Q for perfesser Auld: if econometricians call
>"logit" models "logistic regression" models, do they also call
>"tobit" models "tobistic regression" models?
Nah, we call logit models logit models, it's everyone else
who's got the term wrong. Pop quiz: Which literary work
contains a character named -- not by coincidence -- Mr. Tobit?
Incidentally, the stuff DLJ quoted in a strange attempt
to justify the statistical gibberish he keeps uttering,
>> ><p>Here's your density-density correlation function: </p>
can be found by typing "density fractal" into Google. It's
part of a physics thesis submitted to the University of Oslo
in 1996, and of course has absolutely nothing to do with
the meaningless strings of jargon DLJ keeps entertaining
us with.
Robert Vienneau
(Christopher Auld) wrote:
> David Lloyd-Jones <d...@rogers.com> wrote:
> Just to sum up, this:
> >the curve is wildly
> >irregular as you study intelligence in greater detail: all the
> >underlying values are distributed in what look like ragged, probably
> >fractal, curves, not smooth distributions.
> is gibberish. David has been asked twice to explain what on earth
> it means, but he refuses to do so.
Let F(x) = The number of elements in the Cantor set less than or
equal to x. F has all the properties of a Cumulative Distribution
Function. That is, a random variable can have this distribution.
Poor Chris Auld.
Christopher Auld
>> >the curve is wildly
>> >irregular as you study intelligence in greater detail: all the
>> >underlying values are distributed in what look like ragged, probably
>> >fractal, curves, not smooth distributions.
>Let F(x) = The number of elements in the Cantor set less than or
>equal to x. F has all the properties of a Cumulative Distribution
>Function. That is, a random variable can have this distribution.
>
>Poor Chris Auld.
The creationist contingent attempts to salvage DLJ, just
on general principles! How entertaining. Now, the question
wasn't "is there anything to do with fractals that can also
describe random variables." The question was what it means
to say a random variable is distributed "in what look like
ragged, probably fractal curves, not smooth distributions."
The person who wrote that statement can't explain what it
means (in particular, he didn't say "I meant intelligence
follows the Cantor distribution"), but perhaps Robert can:
Robert, is F() a continuous function? Does it have "ragged,
probably fractal curves?" What does it mean to say a variable
is distributed according to "ragged, probably fractal curves?"
Which papers demonstrate the distribution of intelligence
if of the form F(x), and what are the relevant test statistics?
David Friedman
David Lloyd-Jones <d...@rogers.com> wrote:
> No, I don't, but the point is obvious, and generally conceded: there are
> far more people with IQ's of 200 and of 50 than you can fit under a
> curve with mode 100 and standard deviation 15 -- or any other plain bell
> curve.
>
> Of course you can fix this -- by chqanging your scoring plan so that the
> first few right answers are worth more and the last few are worth less
> -- but the phenomenon stays the same if you go to some other "measure"
> of intelligence such as speed of nervous system or knowledge of music or
> ability to parse Klingon verbs.
All of this seems to assume that there is some natural metric embodied
in tests. But consider a simple exam with a hundred questions on it.
Whether or not the ability it is measuring is distributed on a bell
curve, the scores can't get above a hundred, so the distribution of
scores, as the number of people taking the test gets large, can't fit a
bell curve.
More generally, why would you assume that getting a score of 1 instead
of zero on an exam shows the same amount of "extra intelligence" as
getting a score of 99 instead of 98?
Since there isn't a natural metric, one obvious approach to measuring a
common characteristic with a variety of tests is to define the variable
in terms of the distribution. I.Q. of 100 means the average, 115 means
one standard deviation above the average, and so on. You seem to think
there is something bizarre or dishonest about that.
And I am still waiting for your justification for the comment about
pencilling in data--or your retraction. My guess is that you are
confusing two different issues--mistaken criticisms of an article by
Burt about social mobility with your guess about what you think he did
wrong in measuring I.Q. But that's only a guess. Perhaps you could at
least tell us what the title was of the article in which he supposedly
published the invented data added to give a curve the right shape.
David Lloyd-Jones
>The creationist contingent attempts to salvage DLJ, just
>on general principles! How entertaining. Now, the question
>wasn't "is there anything to do with fractals that can also
>describe random variables." The question was what it means
>to say a random variable is distributed "in what look like
>ragged, probably fractal curves, not smooth distributions."
>
Not quite. The "random variable" is Chris's pathetic interpolation, as
he struggles to prove that *his* statistical methods are relevant to
intelligence in the real world. What I said -- and clearly meant, though
Chris seems to be unable to get it -- was that intelligence, eith in one
individual or as the intelligences of different people, does not have
the characteristics of a smooth function.
robert Vienneau has correctly taken the idea that it might have some of
the qualities of Cantor's various bizarre findings. I have pointed out,
for instance that it is not orderable; I don't know whether Cantor has a
creation with this property. It also changes in "colour" depending on
the scale at which it is examined, a common quality of Mandelbrot
beasts, and probably of some of Cantor's.
>The person who wrote that statement can't explain what it
>means (in particular, he didn't say "I meant intelligence
>follows the Cantor distribution"), but perhaps Robert can:
>Robert, is F() a continuous function? Does it have "ragged,
>probably fractal curves?" What does it mean to say a variable
>is distributed according to "ragged, probably fractal curves?"
>Which papers demonstrate the distribution of intelligence
>if of the form F(x), and what are the relevant test statistics?
>
>
Keep you eye on the ball, folks. The "continuous function," like the
"random variable" above, is Chris's invention, not ine and not Robert's.
And the task of proving that statistical methods of any kind whatsoever
are applicable to human intelligence -- not not test results, which are
a different topic and not part of this discussion -- is Chris's problem,
not mine, not Robert Vienneau's.
All the experience of the past hundred years in this benighted, and in
Burt's case beknighted, field indicates that Chris has a rather
difficult task ahead of him.
Myself I can imagine that some statistical methods might be usefully
applied to the results of "IQ tests" and the various "suitability" tests
of the sort pioneered by ETS. On the other hand it seems pretty clear
that the metaphor of a single quality, "intelligence" or whatever, with
a dispersion around some central value is not the way to go. It's been
tried in great depth at great expense, and it is not at all clear that
it has done *anything* useful, while it is simultaneously clear that it
has done great harm.
There must be alternative metaphors for human abilities. Here's a
thought :my friend Albert Szent-Georgi once said to me "An IQ difference
of thirty points means that no amount of explanation can make clear to
one person a problem which the other can solve by inspection -- and you
meet people over a range of three of those bridges in ordinary life
every day."
Mightn't it be useful to do skill-specific research identifying what
these gaps and bridges look like for specific realms of endeavour?
Certainly couldn't be less useful than all the effort that's been thrown
at the "dispersion around a central value" notion of "intelligence."
-dlj.
-dlj.
David Lloyd-Jones
>All of this seems to assume that there is some natural metric embodied
>in tests. But consider a simple exam with a hundred questions on it.
>Whether or not the ability it is measuring is distributed on a bell
>curve, the scores can't get above a hundred, so the distribution of
>scores, as the number of people taking the test gets large, can't fit a
>bell curve.
>
>More generally, why would you assume that getting a score of 1 instead
>of zero on an exam shows the same amount of "extra intelligence" as
>getting a score of 99 instead of 98?
>
I have quite explicitly not assumed that.
>Since there isn't a natural metric, one obvious approach to measuring a
>common characteristic with a variety of tests is to define the variable
>in terms of the distribution. I.Q. of 100 means the average, 115 means
>one standard deviation above the average, and so on. You seem to think
>there is something bizarre or dishonest about that.
>
No. What I find bizarre and dishonest is the notion that since we have
here a test, therefore the thing we say we are measuring a.) exists, and
b.) has qualities which are measurable and, importantly, rankable, by
the test.
>And I am still waiting for your justification for the comment about
>pencilling in data--or your retraction.
>
I've just ordered a number of books on Burt and the controversy,
including the book you recommended. I'll get back to you.
In the meantime maybe somebody could point out to poor Chris Auld that
"contradicted" does not mean the same thing as "refuted." For that
matter, the fact that he gibbers is only evidence that something is
gibbernogenic, not gibberish, and it is only such evidence if there is
firm evidence that he did not gibber before being exposed to the
material in question.
-dlj.
Stephen J Fromm
> And this time it's a Musical Award. Congratulations for qualifying two
> years in a row.
>
> In the category, "These Foolish Things (Remind Me of You)", the nominee
is:
>
> "Stephen J Fromm" <stephe...@verizon.net for:
>
> > > > You seem to like to appeal to authority ....
>
> > > > your appeal to authority ....
>
> > ...Seems like you like citing
> > authority to hide your own ignorance.
Go crawl back into whatever right-wing miasma you crawled out of, susupply.
Fact is, you know nothing about statistics---all you can do is cut and paste
like a twelve-year old. You don't understand the difference between citing
facts and figures and adding to a debate and mindlessly repeating the words
of others.
Christopher Auld
>>describe random variables." The question was what it means
>>to say a random variable is distributed "in what look like
>>ragged, probably fractal curves, not smooth distributions."
>Not quite. The "random variable" is Chris's pathetic interpolation, as
>he struggles to prove that *his* statistical methods are relevant to
>intelligence in the real world.
While I would like to claim that basic properties of distibutions
are "*my*" inventions, they are unfortunately not. What is more,
I am not trying to "prove" anything about the applicability or
lack thereof of various statistical methods to intelligence. I
pointed out that the charge against Curil Burt DLJ made is
unfounded, and anyone who wishes can read the letter of Stigler
and Rubin in _Science_ for further details.
I also repeatedly asked David what he meant by the statistical
claim that a density has "ragged, probably fractal curves," and
of course it turns out that that is gibberish. In lieu of an
explanation of what that curious phrase means, we get lots of
insults, gnashing of teeth, and name dropping peppering a
grossly incompetent discussion of statistical concepts.
Embellished, funnily enough, with claims that the discussion
has nothing to do with statistics --- apparently, David doesn't
realize that "bell curve" is a layman's term for the density
function of a random variable following the Gaussian
distribution.
Summing up, DLJ is statistically illiterate and, therefore, he
acts irresponsibly when he makes or repeats allegations of fraud
or incompetence against quantitative researchers.
David Lloyd-Jones
they might want to look at
http://pavlov.psyc.queensu.ca/faculty/butler/butler2.html , a report on
the "The Twin Papers" scandal.
Roughly speaking it's a case of Burt taking an old bunch of "findings"
and claiming that he'd found the same thing again with a larger
asemblage of data. The later data was entirely fictitious, i.e. the
original group of numbers recycled. This answers David Friedman's
enquiry about "simply pencilling in data," by the way. It wasn't the
case I had had in mind, but it is a case of inventing data and pasting
it in to fit a preconceived conclusion. There is no question of this
being a valid "normalization" of data: making one datum into three or
five data points is forgery, not normalization.
As I say, this should be taken with a grain of salt, since it comes from
Queen's. Still it looks to me better than some of the stuff that comes
out of that joint.
-dlj.
Christopher Auld
>If anybody here thinks a PhD from Queen's University is worth anything,
Ah yes: The sci.econ kook's mantra: 'You are wrong not
because of a lack of education, but because you have
an education.'
>enquiry about "simply pencilling in data," by the way. It wasn't the
>case I had had in mind, but it is a case of inventing data and pasting
Which is the point: Burt stands accused of several sins. The
particular accusation DLJ has been making for the last few days
has long been known to be unfounded. DLJ has now discovered there
are others, but that doesn't rescue his incoherent ideas about
certain statistical issues in psychometrics, nor the accusation
that rescaling to the normal curve constitutes fraud.
David Lloyd-Jones
>David Lloyd-Jones <d...@rogers.com> wrote:
>
>>enquiry about "simply pencilling in data," by the way. It wasn't the
>>case I had had in mind, but it is a case of inventing data and pasting
>>
> Which is the point: Burt stands accused of several sins.
*And on a mere excursion into Google I have him nailed dead to rights on one of the major ones: the guy faked his data.*
> The
>particular accusation DLJ has been making for the last few days
>has long been known to be unfounded.
>
Uh, not quite, Chris. You have reinterpreted my original "the guy faked
his data" into your "he normalized some of his data" and then you have
presented a number of people saying that this is OK to do.
You have defended him from a charge I didn't make -- but the more I look
into it the more it appears that your conjured-up elsewhere-defence is
also wrong. Even if he had never been straight out dishonest, which ws
clearly not the case, it is likely that what the professional
statisticians defend as usual professional procedure is in fact
unjustified, epistemologically unsound, ridiculous in its conclusions
and harmful in its policy results, and generally nutso-the-squirrel
stuff that could only come from an isolated academy -- perhaps with a
wee tidge of support from the massed human relations departments of the
land. In other words he was not merely a fraud and a forger: he was a
fraud and a forger whose bad work had bad results.(The "bad" refers to
the fact that he got caught, never a virtue in a forger.)
One has only to speculate about his knighthood to lead one into the
subject of whether he was not merely a bad fraud and forger but also a
wicked bad fraud and forger.
> DLJ has now discovered there
>are others, but that doesn't rescue his incoherent ideas about
>certain statistical issues in psychometrics, nor the accusation
>that rescaling to the normal curve constitutes fraud.
>
No, poor little Chris, I haven't "now" discovered that there were
others. This happens to be the one case in which proof comes easiest to
hand.
The fact that I only accused him of one -- of which he now turns out to
have been utterly guilty -- does not for a moment suggest that he wasn't
guilty of a bunch of other things as well, and I knew it all along. As I
said earlier, I've ordered the books: there'll be more, and you shall
cringe whining back to your burrow before this matter is out.
* * *
Just a check point, poor Chris: Do you not now concede, on the basis of
the "The Twins" case, that Sir Cyril Burt was a forger and a fraud?
Don't worry, this is not your last chance to surrender: there'll be more
later, as I get command of more of both the literature and the facts.
-dlj.
>
>
>
ro...@telus.net
wrote:
>If anybody here thinks a PhD from Queen's University is worth anything,
>they might want to look at
>http://pavlov.psyc.queensu.ca/faculty/butler/butler2.html , a report on
>the "The Twin Papers" scandal.
>
>Roughly speaking it's a case of Burt taking an old bunch of "findings"
>and claiming that he'd found the same thing again with a larger
>asemblage of data. The later data was entirely fictitious, i.e. the
>original group of numbers recycled. This answers David Friedman's
>enquiry about "simply pencilling in data," by the way. It wasn't the
>case I had had in mind, but it is a case of inventing data and pasting
>it in to fit a preconceived conclusion. There is no question of this
>being a valid "normalization" of data: making one datum into three or
>five data points is forgery, not normalization.
IIRC, after the original claims about Burt were aired so loudly in the
popular media, someone actually dug up the data, and wrote an
exculpatory article (completely ignored, of course) that completely
exonerated Burt.
-- Roy L
David Lloyd-Jones
> Uh, not quite, Chris. You have reinterpreted my original "the guy
> faked his data" into your "he normalized some of his data" and then
> you have presented a number of people saying that this is OK to do.
>
> You have defended him from a charge I didn't make -- but the more I
> look into it the more it appears that your conjured-up
> elsewhere-defence is also wrong. Even if he had never been straight
> out dishonest, which ws clearly not the case, it is likely that what
> the professional statisticians defend as usual professional procedure
> is in fact unjustified, epistemologically unsound, ridiculous in its
> conclusions and harmful in its policy results, and generally
> nutso-the-squirrel stuff that could only come from an isolated academy
> -- perhaps with a wee tidge of support from the massed human relations
> departments of the land. In other words he was not merely a fraud and
> a forger: he was a fraud and a forger whose bad work had bad
> results.(The "bad" refers to the fact that he got caught, never a
> virtue in a forger.)
It's always fun to write stuff like this, but sometimes a moment of
reflection, and perhaps clarification, is called for.
I originally made one charge, that Sir Cyril Burt was a fraud. This is
now settled, though not on the original "New Scientist" charge that I
had in mind. When I get the details he may or may not, place your bets,
turn out to be guilty of that, too. In other words the question "Was Sir
Cyril Burt a fraud in the sense that he fabricated data and lied about
the things he was publishing?" is settled. Yes. He was a plain and
simple fraud, and he got caught.
There is a broader question, which comes up, inter alia, in the defences
of Burt by the racist Philippe Rushton and the respectable but, imho
mistaken, Arthur Jensen: is Burt's work very basically crooked because
it triees to justify pre-decided conclusions through the tricky use of
epistemological fog and numerological trickery which are impressive but
of no relevance to the questions at hand?
(I tend here to dismiss the defence of Burt made by David Friedman, on
the basis of a book he's read, that his statistical methods were
legitimate within the craft of statistics. This is true but irrelevant,
like saying that Michael Skalkel used the overhand grip. )
This debate has made me think about this stuff in a way that I haven't
for thirty years -- which gets back to when I was a Governor of
Educational Testing Service -- a fine organization, whose staff's
critiques of IQ testing and such tend to be much sharper than mine.
Robert Vienneau's excellent and good humored note in my defence -- in
which he invoked Cantor's many and wonderful creations as sound, imho,
examples of the ways in which I was thinking human intelligence might
more usefuly be said to act -- got this debate going in a sound way, I
think.
I have tried to keep it going with my note about Szent-Georgi's "three
bridges of everyday life" view of intelligences.
There's lots of fun to be had here, and a good deal of work. but none of
it will be done by sour-minded little mediocrities like poor little
Chris Auld.
* * *
Verbum sap: ETS, which was the best of the Generation Of 1945, and got
some of it right, a great deal of it wrong, has for about the last four
or five years been turning the huge ship about in its way. Currently
they are a figure of fun in the media because so many of their new
initiatives are commercial -- which sorta suggests that some writers
miss how much fucking money flowed through the old model. My brief
Governorship (actually I was carrying my boss's proxy) was the first
time I was ever picked up from Princeton Junction by limousine, and then
flown back to Washington on an almost private plane.
-dlj.
David Lloyd-Jones
>David Lloyd-Jones <d...@rogers.com> wrote:
>
>I am not trying to "prove" anything about the applicability or
>lack thereof of various statistical methods to intelligence. I
>pointed out that the charge against Curil Burt DLJ made is
>unfounded, and anyone who wishes can read the letter of Stigler
>and Rubin in _Science_ for further details.
>
Christopher confuses assertion with proof. The fact is Burt was a fraud.
The fact that some, only some, of his statistical methods are defended
by other statisticians does not change this fact.
>I also repeatedly asked David what he meant by the statistical
>claim that a density has "ragged, probably fractal curves," and
>of course it turns out that that is gibberish.
>
No. It turned out that poor Chris was gibbering. he asked a stupid
question grossly at variance with the thread of this discussion, and I
sent him a reply which covered the essentials on fractals, namely that
they have really strange and interesting density functions.
>In lieu of an
>explanation of what that curious phrase means, we get lots of
>insults, gnashing of teeth, and name dropping peppering a
>grossly incompetent discussion of statistical concepts.
>
Counting the few lines above I see "Llloyd-Jones" and "DLJ" everywhere.
Who the hell are you, poor little Chris, to accuse me of name dropping?
>Embellished, funnily enough, with claims that the discussion
>has nothing to do with statistics --- apparently, David doesn't
>realize that "bell curve" is a layman's term for the density
>function of a random variable following the Gaussian
>distribution.
>
<giggle. stretch. yawn.>
>Summing up, DLJ is statistically illiterate and, therefore, he
>acts irresponsibly when he makes or repeats allegations of fraud
>or incompetence against quantitative researchers.
>
Not quite.
DLJ, while having rather more 9's behind his IQ's percentile that poor
little Chris, has a strong layman's acquintance with statistics, and a
whole hell of a lot more common sense than most people who try to misuse
them, e.g. Chris, a fisherman waving a bladder on a mountain-top, to mix
up about three classical allusions.
It's not that I don't take statistics seriously: it's that I understand
the whole thing seriously enough to use those skills where they're
useful, and to steer clear of nut-house doctrinals when they want to use
them promiscuously.
-dlj.
>
>
>
tonyp
"David Lloyd-Jones" <d...@rogers.com> wrote
>There is no question of this
> being a valid "normalization" of data: making one datum into three
or
> five data points is forgery, not normalization.
This reminds me of a brilliant article that appeared many years ago in
the "Journal of Irreproducible Results". It was titled "The Data
Enrichment Method". The method was specifically described for the
case where one wants to test whether altitude effects the probability
of a fair coin coming up heads. You might think this would require
thousands of coin tosses at sea level and thousands more at the summit
of Everest, since it is surely a very subtle effect. But using the
Data Enrichment Method one can reliably measure the altitude
dependence with just 10 tosses of a fair coin on each step of an
ordinary staircase.
The Data Enrichment Method requires no forgery at all. It only
requires that you accept the existence of that which you are trying to
measure. In the coin example, you merely have to accept that, for
example, a coin toss which gave heads on the third step would _also_
have given heads on any higher step, and tails on the eighth step,
say, would also have been tails on any lower step -- because of the
altitude dependence of fair coins!
I don't know if the authors had Cyril Burt in mind when they wrote the
article.
-- Tony P.
David Lloyd-Jones
the same time wrote:
>IIRC, after the original claims about Burt were aired so loudly in the
>popular media, someone actually dug up the data, and wrote an
>exculpatory article (completely ignored, of course) that completely
>exonerated Burt.
>
>-- Roy L
>
>
Roy,
Nobody gives a shit about your recollections by this time. (Hell, it's
my rough recollections that got us into this trouble in the first place.
The difference is that my rough recollections turn out to be correct.)
This is olde tyme type warfare, so we're all going back to the original
documents.
Your "exculpatory documents" were not ignored -- as can be established
by the facts not only that they have caught your dim attention but that
other people in this newsgroup have referred to them promiscuously.
No document, nor any set of documents, "completely exonerated Burt." The
fact that he forged data is indisputed. He is a fraud, sometimes a
careless one, but even at best his carelessness seems to have been
driven by his relentless ideological presuppositions.
In this discussion so far we have established that as far as the "The
Twins" case is concerned he was a common cheat, forger, and liar. As
more stuff comes in to me I'm fairly sure I'll find more like it. As I
get the books, perhaps three weeks from now, I'll lay it out in more
detail, probably with some commentary on that which is right and that
which is wrong among his defenders.
* * *
There is a larger question, one which might crudely be called
ideological, about whether mere averages and numbers ca be applied
sensibly to human qualities, and if how how fairly? This is not the main
question about Burt (or Lott). In their cases the question is settled:
for Burt, yes, he lied because he was so sure. For Lott, maybe, maybe
not, but he's out there in OJ territory.
The Burt/Lott/OJ question aside, what can we sensibly and usefully say
about human intelligence?
For about four generations now, roughly speaking since WWI, it has been
"generally" agreed that measuring some human ability, one with a
central, if vague, value, and a cardinable set of outliers from there,
ws the most interesting thing to looik at.
Vast amounts of money and academic effort have been spent on this for a
century, give or take, on this project. In its most refined form, Henry
Chauncey's as instantiated in ETS and the post 1950 admissions policies
of all the major US colleges, it has done very little, if any, good. It
has done imeasurable harm over the same period.
{I find, on Googling, that Chauncey died just a few days ago at the age
of 97. I never had the chance to meet him, but always felt as though I
knew him through the stories told me by our mutual friend Ben Wood, also
a major of that generation.
{Poor little Chris Auld accuses me of name-dropping, but I notice that
he uses the phrases "David Lloyd-Jones" and "DLJ" half a dozen times in
the post in which he does so. Who's the name dropper here, fella?
[Jee-zuss, do I have to put a smiley in here for the fools?]}
In Chauncey's shadow I say that his vision was as good as it gets -- and
it lost.
I'm going to bed now. Anybody interested can do the usual: follow the
Google-brick-road: Chauncey. "Educational Testing Service", take it from
there as your taste leads you...
-dlj.
David Friedman
> On Sat, 25 Jan 2003 22:39:32 GMT, David Lloyd-Jones <d...@rogers.com>
> wrote:
>
> >If anybody here thinks a PhD from Queen's University is worth anything,
> >they might want to look at
> >http://pavlov.psyc.queensu.ca/faculty/butler/butler2.html , a report on
> >the "The Twin Papers" scandal.
> >
> >Roughly speaking it's a case of Burt taking an old bunch of "findings"
> >and claiming that he'd found the same thing again with a larger
> >asemblage of data. The later data was entirely fictitious, i.e. the
> >original group of numbers recycled. This answers David Friedman's
> >enquiry about "simply pencilling in data," by the way. It wasn't the
> >case I had had in mind, but it is a case of inventing data and pasting
> >it in to fit a preconceived conclusion. There is no question of this
> >being a valid "normalization" of data: making one datum into three or
> >five data points is forgery, not normalization.
I again suggest that, before making such confident statements, you
read the book _Cyril Burt: Fraud or Framed_, which goes into
considerable detail on the various charges, concluding (roughly
speaking--its the work of mutliple authors who don't entirely agree)
that none of the charges of fraud has been clearly demonstrated,
although some (not, as it happens, the particular one you are
discussing) have enough evidence to cast serious doubts on his
honesty.
You might also want to think, just for a moment, about the
plausibility of this particular line of argument. The claim (the one
your source describes, not his new twist on it) is that Burt published
one article on identical twins separated at birth, a later article
with a larger sample size, and the correlation coefficients in the two
articles were the same--surely an unlikely thing to happen by chance.
Burt was, according to everyone on both sides, an extraordinarily able
man and a pioneer in the application of statistics to psychology.
Suppose he was also a fraud. Further suppose he decided to invent a
bunch of new separated twins in order to strengthen his argument (you
can assume that he invented the earlier ones too if you want). Just
how likely is it that, in writing his bogus piece, it wouldn't occur
to him that adding new data ought to change his (hypothetically
invented) correlation coefficients?
Some of the relevant facts:
While it is true that a surprising number of the coefficients in the
table of correlation coefficients (not all of them, but way too many
to be reasonably explained by chance) in Burt's final paper on
identical twins are the same as the coefficients in the corresponding
table in one of the earlier papers, some of those coefficients
disagree with the corresponding coefficients reported in the text of
the paper. It seems likely that when putting together the final paper
Burt, who was in his eighties at the time, miscopied some of the
figures, putting the correlation coefficients from the earlier
calculations in the table instead of those from the later calculation
with the larger sample. That would explain the discrepency between
table and text. It turns out that there is at least one other case in
his late papers of such miscopying.
The real puzzle is not the identical correlation coefficients, it's
the source of the additional twins. I will let you read the book to
see what the evidence is that presents a problem and what explanations
are offered.
> IIRC, after the original claims about Burt were aired so loudly in the
> popular media, someone actually dug up the data, and wrote an
> exculpatory article (completely ignored, of course) that completely
> exonerated Burt.
I think "completely exonerated" is an exaggeration. Two people wrote
books persuasively arguing that Burt had not been proved to have
committed fraud--persuasively enough so that the British Psychology
Association withdrew its condemnation of him. But the impression I get
from a later book, which is the one I read, is that there is some
evidence of fraud with regard to some of his work, just not
sufficiently strong evidence to clearly convict.
David Lloyd-Jones
>You might also want to think, just for a moment, about the
>plausibility of this particular line of argument. The claim (the one
>your source describes, not his new twist on it) is that Burt published
>one article on identical twins separated at birth, a later article
>with a larger sample size, and the correlation coefficients in the two
>articles were the same--surely an unlikely thing to happen by chance.
>
>
David,
I think you're missing the point.
The "larger sample size" did not exist. Burt confabulated it out of thin
air.
-dlj.
>
>
David Friedman
David Lloyd-Jones <d...@rogers.com> wrote:
And you know that how?
Part of the original evidence for that claim was the similarity of the
tables of correlation coefficients, however. The other part was the
supposed nonexistence of Burt's two female assistants--until it turned
out that one of them definitely existed and the other was at least
tentatively identified. Then there was the supposedly striking absence
of references to acquiring such data in Burt's very detailed
diary--until someone other than the biographer read the diary and
discovered that it wasn't a detailed diary at all but something more
like an appointment book.
Finally, there is various evidence--how conclusive isn't clear to
me--that Burt had not collected any additional data after the war. But
even if we assume that is true, there seems to be some evidence that
Burt's papers got messed up during the war and in his moves immediately
thereafter. So one possible explanation is that the data were collected
before the war but only retrieved and analyzed over a period of time
thereafter, as one document or another came to light. If that is what
happened his description of the situation was at least misleading, but
that doesn't mean the data is fake.
The most obvious problem for your thesis is that the results Burt
reported from his data are consistent with the results of similar later
studies by independent researchers elsewhere. That's a curious
coincidence if he invented the data.
Of course, some of these questions could be answered if Burt's papers
hadn't been destroyed--on the advice of one of the other people in the
field immediately after his death. It was only after that that the
accusations started to be made.
As I said, it's an interesting story. But you are trying to make it much
simpler than it was by simply accepting at face value the assertions of
Burt's critics, at least some of which, so far as I can tell, have been
demonstrated to be false.
David Lloyd-Jones
David,
I'm going to leave your extremely far-stretched maunderings below
exactly as you posted them.
I think it would be charitable on my part to suggest that this case adds
up to some paid employees saying that perhaps they had smelled a dog.
{For classicists in this group, there is one actual case of a dog
destroying homework, and it happened just around the corner from where I
spent several years of my childhood, at the corner of McCaulley Road
(named a hundred or so years later) and the road that ran across the top
of our property, in our tme called "The Avenue."
Newton's dog, Diamond, managed somehow, right there at the corner
window I walked past so often on my way to the park where I would always
practice my political skills saying "What kind of dog is that?" and then
listening as they told me all about their world, to destroy Newton's
papers which had taken twenty years' work.
You can parse a long sentence, can't you?
"Oh, Diamond, you know not what you do!" is the usual account of the
disaster.
Meself, I'm a little sceptical. Liebniz was doing infinitismal calculus
at the same time as (my neighbour) Newton, and everybody and his brother
was doing logs, including natural-base logs, at roughly the same time.
Once you get the first clue, it's really pretty easy: this is what I do
to get asleep at night when my long cool one is off studying.
Slightly further down McCaulley (I'm going American: I can't spell his
name any more) e ugliness -- indeed whose apparently deliberate assault
on the sensibilities of the human race -- is only matched by the Paris
fascist temple, "the Whited Sepulchure", I forget its name, is it Notre
Dame du Grace up there?
* * *
To sum up: in Newton's case, at least he had a dog.
-dlj.
susupply
"Stephen J Fromm" <stephe...@verizon.net>
who will no doubt be disappointed to hear that sequels, no matter how inane,
do not qualify for nominations,
wrote in message news:8kDY9.285$qb1...@nwrddc01.gnilink.net...
> Go crawl back into whatever right-wing miasma you crawled out of,
susupply.
>
> Fact is, you know nothing about statistics---all you can do is cut and
paste
> like a twelve-year old. You don't understand the difference between
citing
> facts and figures and adding to a debate and mindlessly repeating the
words
> of others.
Though perhaps, "mindlessly repeating the words of others", added to his
other self-unawarisms will one day bring him an Out-To-Lunchbox-MacKinnon
Lifetime Achievement Award:
>> Lambert/Ehrlich are ....
> > > Ehrlich, author of _Nine Crazy Ideas in Science_....
> > > See my posts on Ehrlich and Lambert's posts...
> > >> Lambert wrote....Ehrlich made similar points....
> > independent folks like Ehrlich (a
> > physicist) found ....
> > >> Read Ehrlich's book.
>> read what Ehrlich is saying...Ehrlich alleges....
Christopher Auld
David Lloyd-Jones wrote:
...lots of insults in the midst of some truly surreal rants.
Wading through this unpleasantness, we find that David wants
to rewrite history rather than defend his allegation:
> > The
> >particular accusation DLJ has been making for the last few days
> >has long been known to be unfounded.
>
> Uh, not quite, Chris. You have reinterpreted my original "the guy faked
> his data" into your "he normalized some of his data" and then you have
> presented a number of people saying that this is OK to do.
What David originally said was:
> >>The
> >>prototype of these guys was Sir Cyril Burt who just *knew* that
> >>intelligence was normally distributed, and it was so obvious that it
> >>wasn't worth the trouble of testing. So he just pencilled in the
> >>invented data on his charts and published it that way.
As I've noted, there are other allegations against Burt, but
the one David repeated is unfounded.
On the matter of what it means to say that something is
distributed with "ragged, probably fractal curves," David
opines
>No. It turned out that poor Chris was gibbering. he asked a stupid
>question grossly at variance with the thread of this discussion, and I
>sent him a reply which covered the essentials on fractals, namely that
>they have really strange and interesting density functions.
Even asking David what he meant was "stupid," but in a post I
must have missed he nonetheless stooped to lecture on the
"essentials of fractals, namely that the have really strange and
interesting density functions."
Of course, anyone who knows what a "density function" is is
now having yet another chuckle at David's abuse of terminology.
I think it's clear I am never going to get an answer to my
question, because David was just making stuff up. Which is
an odd lapse of knowledge for a genius with an extensive
background in the subject he's discussing,
>DLJ, while having rather more 9's behind his IQ's percentile that poor
>little Chris, has a strong layman's acquintance with statistics, and a
>whole hell of a lot more common sense than most people who try to misuse
>them, e.g. Chris, a fisherman waving a bladder on a mountain-top, to mix
>up about three classical allusions.
(I assume David means he's in the 99.9999th percentile of the
distribution of IQ rather than, say, the 6.99999th.) It appears
I also stand accused of misusing statistics, which would be
a pretty serious charge if it didn't merely place me in the
same league as lesser lights such as Rubin and Stigler and the
rest of the wacky academic "nutso the squirrel" quantitative
research community.
At least we now know why David keeps accusing everyone of
being stupid, idiotic, etc: When you're THAT smart, I guess
almost everyone seems a bit thick. I can only hope that
our resident super-genius deigns to share more fascinating
insights into statistical reasoning. We all have so much
to learn!
David Lloyd-Jones
>David Lloyd-Jones wrote:
>
>....lots of insults in the midst of some truly surreal rants.
>Wading through this unpleasantness, we find that David wants
>to rewrite history rather than defend his allegation:
>
>
>
>>> The
>>>particular accusation DLJ has been making for the last few days
>>>has long been known to be unfounded.
>>>
>>>
>>Uh, not quite, Chris. You have reinterpreted my original "the guy faked
>>his data" into your "he normalized some of his data" and then you have
>>presented a number of people saying that this is OK to do.
>>
>>
>
>What David originally said was:
>
>
>
>>>>The
>>>>prototype of these guys was Sir Cyril Burt who just *knew* that
>>>>intelligence was normally distributed, and it was so obvious that it
>>>>wasn't worth the trouble of testing. So he just pencilled in the
>>>>invented data on his charts and published it that way.
>>>>
>>>>
Chris,
What the hell have you been smoking? This quotation precisely confirms
what I have been saying (about what I had said) and exactly contradicts
all your surreal claims.
The "The Twins" case is exactly an example of this: he knew what he
wanted to have found, so he pencilled it in. And that's only what I've
found in the first day of searching. On the basis of the jourmalistic
record I am confident in predicting that the rest of the record will
turn up a lot more of the same kind of fakery.
He was a fraud, and you, young man, are a fool -- and a fool on dogmatic
autopilot.
I quite agree with you that this discussion has been unpleasant so far,
and it looks like continuing to be so as I continue to find more and
more fakey by your hero and his friends. Since you are so good at
rewinding your files, I suggest you look back and see exactly who
started the unpleasantness. Search under "Auld" and start scanning from
about 1994, you sour, arrogant, ignorant little twirp.
-dlj.
>
>
David Lloyd-Jones
> I quite agree with you that this discussion has been unpleasant so
> far, and it looks like continuing to be so as I continue to find more
> and more fakey by your hero and his friends.
David,
I think that even a dolt like poor little Chris deserves to know the
deep knowledge Anai has brought home from school:
"Why did the chewing gum cross the road?"
To stick to the chick.
This vital news brought to adults in economics everywhere by the
five-year-old survey network.
-dlj.
Christopher Auld
>What the hell have you been smoking? This quotation precisely confirms
>what I have been saying (about what I had said) and exactly contradicts
>all your surreal claims.
There are many allegations against Burt. One of them --
the one David repeated -- related to scaling to the normal
curve. This allegation is false. Contrary to the content
of David's rants, I have made no comment whatsoever on any
other charges against Burt, which include the suspect
twin's data -- a charge that has absolutely nothing to do
with the normality or lack thereof of intelligence. That is,
the "pencilling in of data" and the normality issue are
two seperate charges that David has conflated.
The rest of David's post(s) consists of reams of schoolground
insults. I suppose that one must by a super genius such
as David to understand why anyone would find such tantrums
persuasive.
--
David Lloyd-Jones
Christopher,
Just to review: I have so far demonstrated, to your satisfaction, that
Burt was a forger and confabulist.
What you say above "scaling to the normal curve" is entirely your own
invention. My claim is that he made up data and stuck it on a normal
curve, like a Gilbert and Sullivan clown.
So far I've dug up the "The Twins" case, which proves that what I said
was correct. You were wrong.
You didn't "misinterpret" or "overweight some of the paramaters." You
simply fucked up, had it wrong, lived, however briefly, in a universe
different from the human race. If you repeat your earlier claim, you'll
be a liar; if you don't withdraw it, you'll be a low-class little dodger.
You were wrong in an ugly, undignified, insulting, and distressing way.
I don't think you have a future in academia -- where contentiousness is
valued, but style and good manners are also important. I suggest you
look to your future, little man, and apply for a job at your local
income tax office.
-dlj.
>
>
ro...@telus.net
<dd...@daviddfriedman.com> wrote:
>Then there was the supposedly striking absence
>of references to acquiring such data in Burt's very detailed
>diary--until someone other than the biographer read the diary and
>discovered that it wasn't a detailed diary at all but something more
>like an appointment book.
In British usage, "diary" means an appointment book, not a personal
journal, as in American usage.
-- Roy L
Christopher Auld
...well, lots more ranting and insults. Geniuses can be
touchy, and David must be forgiven his tantrums given
that he has to put up with the nutso the squirrel arguments
of idiots such as
Stigler, S. and D. Rubin, "Dorfman's data analysis,"
Science, New Series, Vol. 205, No. 4412. (Sep. 21,
1979), pp. 1204-1206.
What would these isolated academics know about statistics?
It's not like they're super geniuses with a strong layman's
acquaintance with the subject or something. All us dolts
and twerps in academia should be in awe of DLJ and not
bother him with any more stupid questions.
David Lloyd-Jones
>Stigler, S. and D. Rubin, "Dorfman's data analysis,"
> Science, New Series, Vol. 205, No. 4412. (Sep. 21,
> 1979), pp. 1204-1206.
>
>What would these isolated academics know about statistics?
>
Hysterical little Chris seems to think the discussion is about
statistics, somehow.
It's not. It's about basic honesty, something Sir Cyril Burt lacked --
and was caught lacking.
I rather doubt that Stigler and Rubin ever had to give you this bit of
advice, little Chris, so I'll do it on my own authority: if you need a
new data set for a new set of twins, go and do the work; don't multiply
some old stuff lying around your office by a Cook's Factor and print
that. That's a no-no.
More advice: If you ever go on radio and television raving about the
findings of some survey you have done, be sure that your dog does not
eat the survey papers in their entirety.
No doubt you can contact Stigler and Rubin on the Net: they will confirm
the soundness of my advice.
-dlj.
Stephen J Fromm
>
> "Stephen J Fromm" <stephe...@verizon.net>
>
> who will no doubt be disappointed to hear that sequels, no matter how
inane,
> do not qualify for nominations,
>
> wrote in message news:8kDY9.285$qb1...@nwrddc01.gnilink.net...
>
> > Go crawl back into whatever right-wing miasma you crawled out of,
> susupply.
> >
> > Fact is, you know nothing about statistics---all you can do is cut and
paste
> > like a twelve-year old. You don't understand the difference between
> citing
> > facts and figures and adding to a debate and mindlessly repeating the
> words
> > of others.
>
> Though perhaps, "mindlessly repeating the words of others", added to his
> other self-unawarisms will one day bring him an Out-To-Lunchbox-MacKinnon
> Lifetime Achievement Award:
Wrong.
When I cite Ehrlich and Lambert, I'm saying "this party said X, and I
understand what they say and either agree with what they said or think they
made an interesting point." Without such shorthand, communication
would be impossible.
When you cite Auld, you're saying, "I don't have the faintest clue what his
lingo means, but I'm citing him because he's an authority." You wrote,
"Well, we've just seen Lambert made mincemeat of by sci.econ's favorite
econometrician." Note the language: "sci.econ's favorite econometrician".
Clearly a blind appeal to authority.
I doubt you know anything about statistics, and you haven't provided any
evidence to dispel my opinion.
[snip]
Christopher Auld
[ insults ]
>>Stigler, S. and D. Rubin, "Dorfman's data analysis,"
>> Science, New Series, Vol. 205, No. 4412. (Sep. 21,
>> 1979), pp. 1204-1206.
>>
>>What would these isolated academics know about statistics?
>Hysterical little Chris seems to think the discussion is about
>statistics, somehow.
Yeah, crazy me. What would a discussion about density
functions, test statistics and the statistical results
of a psychometrician have to do with statistics? Why,
that's an hysterical assertion!
>It's not. It's about basic honesty, something Sir Cyril Burt lacked --
>and was caught lacking.
For the fourth time: I have not made any comment on Cyril
Burt beyond noting that the allegations against Burt made
by Dorfman were unfounded. I have not made any other comment
about Burt, his methods, or his integrity. David seems
to have hallucinated a discussion in which Burt is "my hero"
and I am defending him broadly. Much of the controversy
exists precisely because Burt failed to adequately document
his data collection, for example, which doesn't bode well.
But, for more than the fourth time: The particular allegation
David repeated was that Burt "pencilled in the data" to
make it look normal. Dorfman made this charge after noting
that Burt's data look "too normal." This charge was
picked up by the media and, obviously, is still being
repeated today. But as Stigler and Rubin point out, Burt
plainly says in the paper in question that he rescaled the
data. This is not a dishonest or nonstandard procedure.
Stigler and Rubin go on a subsequent short piece in Science
a few months later to point out that even if one ignores
the rescaling, the test statistic Dorfman used to check if
the data are "too normal" tends to massively verify that
hypothesis even when the data are not normal. Using
Dorfman's chi-squared test would lead one to the conclusion
that almost all empirical science is fraudulent.
But this discussion has nothing to do with statistics.
[ snip: more ranting ]
Hey David, are you going to lecture more on the fractal
density of intelligence? Or was that the density function
of fractals? What color are they, for instance?
susupply
"Stephen J Fromm" <stephe...@verizon.net>
for whom, apparently, the logical fallacy is deeply personal Performance
Art,
wrote in message news:KIbZ9.3014$P64....@nwrddc02.gnilink.net...
> When I cite Ehrlich and Lambert, I'm saying "this party said X, and I
> understand what they say and either agree with what they said or think
they
> made an interesting point." Without such shorthand, communication
> would be impossible.
This is shorthand that is allowable only to the annointed on the left? BTW,
do you know that "impossible" doesn't mean "inconvenient"?
> When you cite Auld, you're saying, "I don't have the faintest clue what
his
> lingo means, but I'm citing him because he's an authority." You wrote,
> "Well, we've just seen Lambert made mincemeat of by sci.econ's favorite
> econometrician." Note the language: "sci.econ's favorite
econometrician".
> Clearly a blind appeal to authority.
This is, of course, the worst kind of Appeal to Authority; the "authority"
being oneself (and a mindreader to boot).
I also note that as a matter of hierarchy, Lambert (notorious even on
sci.econ for his obtuseness) and Ehrlich (a physicist, not a social
scientist) are outranked by "sci.econ's favorite econometrician". Which
description of Chris Auld being "shorthand" for a fellow who has been
demonstrating his competence here for several years--often with devastating
humor (though it's probably better to keep humor out of a discussion with
Stephen Fromm).
Further, it was pretty clearly Stephen Fromm who didn't have, "the faintest
clue", what Auld had written (and it wasn't exactly--bow to Ehrlich et
al--rocket science; trend lines from regression analysis), as this exchange
made plain:
[Fromm:]
<< Lambert/Ehrlich are specifically referring to fitting two
different curves, one on each side of the point of interest. Nothing in
your clip above refutes that.>>
[Auld:]
<< Of course it does. Lott used a perfectly standard regression-based
technique to test for what's called "structural change" in the
econometrics literature. This essentially involves fitting "two
curves" on either side of (in this case, known) break point, and
testing statistically to see if there is a significant difference in
the slopes before and after this point. >>
But, if you have some hitherto unspoken problem with Chris Auld's claim,
let's hear it.
And, I can't help notice something else about your posts, they don't refer
to anything that you've read in Lott's book. Have you actually read it
yourself?