Lott’s correction policy

In the comments to Lott’s post where he failed to notice that the study on guns and road rage reported a multi-variate regression in table 1, someone asked him about my post pointing out his mistake. Lott’s response:

I haven’t bothered looking at Lambert’s page, but he typically
doesn’t have a clue what he is talking about. Assuming that he
actually made the claim that Hemenway et al reported regressions in
the tables, that is definitely not the case. The Tables just show
the means, or conditional means. They are not regression estimates.
Clue: you can’t have a regression where you have a dummy variable
for both male and female (you can only pick one or the other)
because otherwise you will have perfect collinearity with the
intercept term. The package will force you to drop one of the three
variables.

Clue: the table contains bivariate and multivariate (in the columns labeled “multivariate”) results. Lott is just looking at the columns that present the bivariate results.

Anyway, it looks like the penny finally dropped, because he has now deleted his claims about multivariate regressions, the comment that asked about my post, and his response above. As usual, Lott tries to rewrite history to cover up his mistakes.

Comments

#1 SayUncle
February 7, 2006

I had a few emails with Lott on this particular issue. Every time I start to like the guy, he does something like this. He just removed the comment without addressing it or giving a reason why. Just gone. Not good. It’s ok to admit you screw up.
#2 Tim Lambert
February 7, 2006

Lott fabricated a defensive gun use survey rather than admit to screwing up.
#3 z
February 7, 2006

“I haven’t bothered looking at Lambert’s page, but he typically doesn’t have a clue what he is talking about.”

Department of unconsciously ironically self-referential sentences wherein people reference themselves, unconsciously ironically?
#4 z
February 7, 2006

“It’s ok to admit you screw up.”

I thought I had made a mistake once; but I was mistaken.
#5 z
February 7, 2006

What Lott doesn’t say is that although he doesn’t bother seeing what is written about him here, Mary Rosh keeps him appraised in detail.
#6 Fed Up
February 7, 2006

Is there a Lottwatch site somewhere that has kept track of all these self-aggrandizing ego-manical shenanagans of his?
#7 Chris Jarrett
February 7, 2006

When someone claims that Tim is making false claims about what Lott wrote on his blog (and it will happen), Google cache is a great way to show Lott’s original page and his altered page. Search on his web page for Hemenway and check out the cached version. Tim, if you didn’t already, you should save a copy of the earlier version.

Lott’s inability to admit a small error is yet another example of why he didn’t get tenure at UPenn and was unable to obtain a permanent academic position.
#8 ben
February 7, 2006

“Is there a Lottwatch site somewhere that has kept track of all these self-aggrandizing ego-manical shenanagans of his?”

Yeah, it’s right here.
#9 Barry
February 7, 2006

I’d like to add to Jarrett’s point – Lott’s got a Ph.D. in econ from U Chic, the top school. If he can’t even get a position at Dumbf*ck U, that’s very, very bad. Remember that, even after a whole string of minor state schools which are basically jumped-up colleges, there’s still a string of minor, highly politicized right-wing religious schools that might like a ‘scholar’. And he should be able to snag some under-the-table foundation money, so he’d be a freebie for that college. But he’s so screwed up, that nobody seems to want him, in any obscure crevice of academia.
#10 Barry
February 7, 2006

Lott: “I haven’t bothered looking at Lambert’s page, but he typically doesn’t have a clue what he is talking about.”

z: “Department of unconsciously ironically self-referential sentences wherein people reference themselves, unconsciously ironically?”

Most ‘laws’ in social science are more like weak to medium strength probabilistic relationships, but there is one iron, absolute, bet-the-house principle I’ve discovered: The Iron Law of Right-Wing Freudian Projection. These guys will almost never accuse the left of doing something that the right is not already doing, or plans on doing as soon as possible.

This ties in with a saying that I came up with: If you live in a glass house, perhaps you shouldn’t throw stones; but if you have mud on you, sling as much as possible on your opponent. Then you’re both muddy, and it’s harder for people to figure out who’s dirtier. This is one of Rove’s most basic tactics.
#11 Chris Jarrett
February 7, 2006

Um, Barry? I believe that Lott’s PhD is from UCLA, not UChicago. Lott was a fellow at Chicago for a year or two. It’s my understanding that, even though he sent CVs out to pretty much every University in North America, Lott’s only academic job offer was from a small university in Australia.
#12 Meyrick Kirby
February 7, 2006

Okay, this is weird!

I’ve tried looking at the cache on Google, as Chris Jarrett mentioned.

The search results show parts of the offending comments:

here

but when I look at the cached version I get the blog posting before any comments were made:

here
#13 Meyrick Kirby
February 7, 2006

Does this mean Lott has contacted Google and demanded they backdate their cache?
#14 Chris Jarrett
February 7, 2006

John Lott is so pathetic that, in his blog, he’s asking people to vote for the ‘top gun’ blogger. It’s even more lame than his unethical practice of plugging his SSRN papers. But what I want to know is why SayUncle isn’t included in the poll of top gun blogs. If you want to mess with his mind, go vote for someone else at: http://countertop-chronicles.blogspot.com/2006/02/gunnies.html
#15 ben
February 7, 2006

yes, please do vote for someone else. Preferably Mr. Completely, who has the best chance of beating Lott and if you don’t know the other guys, or maybe Kim du Toit, Alphecca or The Smallest Minority, who also have a shot, so to speak.
#16 dsquared
February 7, 2006

the guy who included “%black” and “%white” in his model for race bias in Florida 2000 voting undercounts has a bloody nerve talking about multicollinearity.
#17 Chris Jarrett
February 7, 2006

Ironically, Clayton Cramer didn’t get the memo from John Lott and Cramer’s blog has retained the original comment from Lott. I wonder if Cramer will delete the error to help cover for Lott. Interestingly, in that post, Cramer states “Dr. Lott has always made his data sets available to other researchers.” Perhaps someone should ask John Lott for his dataset from the 1997 survey.
#18 SayUncle
February 7, 2006

‘what I want to know is why SayUncle isn’t included in the poll of top gun blogs’

Heh. me too.
#19 Meyrick Kirby
February 7, 2006

Now that’s better. Lott’s current altered post:

here

While over a Clayton’s blog we have the original offending material:

here

I’ve posted the mistake on Lott blog. What’s the chance he refuses to publish it?!

P.S. I’ve saved a copy of Clayton’s page, just in case things are altered.

P.P.S. Can some please explain the caching issue I brought up before. I’m a bit of a techie, and I find this interesting. Anyone have any ideas?
#20 Tim Lambert
February 8, 2006

Clayton Cramer has a history of covering up for Lott, so it will be intreesting to see what he does.

I think the chance of Lott refusing to allow your comment is close to 98%.

Google has indexed the later version of the page but the earlier version is in the cache. I’ve noticed this happening on other occasions.
#21 Scott Church
February 8, 2006

Regarding the Google cache idea, another possibility would be to look up Lott’s site at http://www.archive.org. If they have record of it (I haven’t had the chance to check yet) you could get various versions of his site from different timeframes up to several years back. The more recent the site updates, the more likely they’ll have something. It’s worth a check.
#22 Scott Church
February 8, 2006

Regarding Lott’s comments,
“Clue: you can’t have a regression where you have a dummy variable for both male and female (you can only pick one or the other) because otherwise you will have perfect collinearity with the intercept term. The package will force you to drop one of the three variables.”
It’s ironic that Lott would make a remark like this seeing as how one of his most infamous regression studies–one that to this day is widely cited in Right-Wing forums around the world–depended on exactly this type of multicollinearity for its results. During the investigation of the year 2000 Bush/Gore presidential runoff in Florida by the U.S. Commission on Civil Rights (USCCR), the dissenting opinion from Commissioners Thernstrom and Redenbaugh harshly criticised one of the USCCR’s most significant findings: that the widespread voter disenfranchisement resulting from the negligence of the Florida Secretary of State’s office fell disproportionately on minority voters.
The USCCR based much of this conclusion on a regression study done by Alan Lichtman of American University. Thernstrom and Redenbaugh hired Lott to do an opposing study calculated to debunk Lichtman’s. Naturally, Lott’s study concluded that race had no correlation whatsoever to voter disenfranchisement–a conclusion that has been repeated uncritically ever since by nearly every conservative forum in America that has ever addressed that election.
Well, it turned out that several of the models Lott ran in that study had severe multicollinearity issues of their own. One or two in particular had racial variables for both black and white/hispanic. Unfortunately, Lott overlooked one little detail: blacks, whites, and hispanics collectively comprise nearly all of Florida’s voting population. Thus, these two “independent” variables mirrored each other almost perfectly (a paper I wrote last year on Florida 2000, PDF version here and HTML version here, discusses Lott’s study in some detail). So of course he found no racial bias. When Lichtman reran Lott’s models with one of the two racial variables removed his results vanished. Thus, when properly characterized, Lott’s own models refute his claims about that election!
And here he is complaining about “perfect collinearity with the intercept term”–as though any fool would know to avoid it!
Hmmm……..
#23 Chris Jarrett
February 8, 2006

The regression shown in Hemenway et al. (2006) does not contain a dummy variable for both male and female gender varibles. There is a multivariate odds ratio only for Male (1.2) and that means that men are more likely to report having engaged in a road rage incident. Like the other comments about the study, Lott is simply wrong and he’s demonstrating his rather poor comprehension levels about the study and the methodology used. Lott reminds me of the worst reviewer I’ve had on a journal article. The comments show that the person not only didn’t read the article very well, they show that he doesn’t understand the methods he’s criticizing.
#24 Chris Jarrett
February 8, 2006

As a follow-up on Cramer, in his post, he writes:
“If a historian published a paper that made digested claims about something–and wasn’t prepared to make either the database or a complete list of sources available from which that database came–it would quickly lead to suspicion of fraud.”

I suppose he thinks that the ‘suspicion of fraud’ only applies to historians and does not apply to economists…

The irony of that statement is too funny.
#25 Barry
February 8, 2006

Chris Jarrett: “Um, Barry? I believe that Lott’s PhD is from UCLA, not UChicago. Lott was a fellow at Chicago for a year or two. It’s my understanding that, even though he sent CVs out to pretty much every University in North America, Lott’s only academic job offer was from a small university in Australia.”

My bad.
#26 z
February 8, 2006

“Lott’s inability to admit a small error is yet another example of why he didn’t get tenure at UPenn and was unable to obtain a permanent academic position.”

Well…. the more cynical of us attribute it to an inability to cover up the small errors in a convincing fashion…..
#27 Lettuce
February 8, 2006

Barry:

The Iron Law of Right-Wing Freudian Projection. These guys will almost never accuse the left of doing something that the right is not already doing, or plans on doing as soon as possible.

This is known as the “waking up, looking in the mirror, thinking you see a liberal syndrome.”
#28 Harald Korneliussen
February 9, 2006

“Lott’s only academic job offer was from a small university in Australia.”

That wouldn’t be the University of New South Wales, would it?
#29 Tim Lambert
February 9, 2006

No, I think it was Australian National University, which isn’t a small university, but one of the “big eight” (ANU, UNSW, Sydney, Monash, Melbourne, Adelaide, Western Australia, Queensland).
#30 Chris Jarrett
February 9, 2006

Oops. My error there for writing off of vague memory and not checking.
#31 Chris Jarrett
February 9, 2006

Lott’s now added a ‘correction note’ to his post.

“Correction: The original note mentioned that only one regression had been run by these authors. In fact, it turns out that four regressions had been run. The points listed above are now correct.”

That’s a typically dishonest statement from Lott. He’s trying to make it look like he’s being honest yet he fails to note his original errors that lead to his deletions.
#32 z
February 9, 2006

Slightly off topic but worth posting:
Correction in yesterday’s New York Times:
“Because of an editing error, a recipe last Wednesday for meatballs with an article about foods to serve during the Super Bowl misstated the amount of chipotle chilies in adobo to be used. It is one or two canned chilies, not one or two cans.” http://www.nytimes.com/2006/02/08/dining/08dcxn.html?_r=1&oref=slogin
#33 Meyrick Kirby
February 9, 2006

Well, Lott’s admitted to making a mistake, although the exact content remains deleted. I thought the standard method of correcting blog contents is to draw a line through the incorrect content so it is still visible.

Additional my comment appears to have been blocked by Lott. What is his policy on comments? Am I a spammer?
#34 Chris Jarrett
February 9, 2006

Lott really didn’t admit to a mistake, or at least didn’t admit to the mistakes he made. He then tries to cover it up by claiming that he said something other than what he actually wrote.

If you look at the comment box, you’ll see that he has “turned on” the moderation switch for that post. He has to approve any comment message that appears in that post.

I guess he’s a Tim Blair acolyte.
#35 Ian Gould
February 9, 2006

>Like the other comments about the study, Lott is simply wrong and he’s demonstrating his rather poor comprehension levels about the study and the methodology used.

My own knowledge of statistics ends at first year University level hence I can’t follow the technical detail of many of these issues.

However, when I first became aware of the various controversies regarding Lott, I read descriptions of him in various soruces such as The Economist which desribed him as a highly competent and respected statistician whose technical expertise wasn’t being questioned – just his ethics.

After a couple of years of tracking Lott, I’m beginning to suspect he’s simply incompetent.
#36 Meyrick Kirby
February 10, 2006

Chris Jarrett:

If you look at the comment box, you’ll see that he has “turned on” the moderation switch for that post. He has to approve any comment message that appears in that post.

Yes, but what is Lott’s policy? Block anyone who criticizes?
#37 Chris Jarrett
February 12, 2006

DemoChoice Web Poll: Best Master Gunnie

Round 9
——————————————-
Mr. Completely 646 (61.5%)
John Lott 405 (38.5%)
None of these 248

Results
Mr. Completely Elected
John Lott Defeated

Mr. Completely has enough votes to guarantee victory and is declared a winner.