Deltoid

Back in January, in my very first posting I have the following comment on Clayton Cramer’s posting of a letter from Lott:

[There used to be some comments here about the results of Lott’s new survey. Clayton had posted that Lott’s new survey found a “bit more than 4%” firing. He informs me that he removed this because it was a mistake and could not post the correct figure because Lott does not want the results made public. Consequently I have deleted my comments on this figure.]

When I asked him why he had removed the “bit more than 4%” firing figure, Cramer was somewhat evasive. After a couple of emails I was eventually able to pin him down:

Forgive me, but I still don’t understand. You wrote “I appear to have scrambled a few details”. Do you mean:

1. The “bit more than 4%” was incorrect, but you can’t tell me what the actual number is because Lott asked you to keep it secret.

Yes.

2. The “bit more than 4%” was correct, and one of the details you scrambled was that you were supposed to keep it secret.

No. I scrambled several things in how I explained it.

When Lott finally released the results from his survey he claimed that there were 5% firing, which seems to me to be a “bit more than 4%”. So, I recently emailed Cramer, suggesting that it appeared that Lott had lied to him. He posted a reply:

No, I answered the second half of your question. The number was a bit more than 4%, and I had given out information that I was not supposed to have given out. I figured that within a few weeks, you would see the actual published number, and see that I had only answered the second half of the question. It was rather a difficult position, because I had unintentionally revealed more than I should have, and yet I didn’t want to confirm this data unnecessarily.

I will let the reader judge whether Cramer’s statements to me were deliberately false, or merely answering the second half of my question.

Glenn Reynolds has posted a defence of Lott by Benjamin Zycher. Lindgren has a reply on the same page that should definitely be read by anyone interested in Lott and the survey—there are some very interesting details in there about how Mustard’s and Gross’s stories changed in ways that made their accounts more favourable to Lott.

While Zycher repeatedly accuses Lindgren of incompetence, he only bothers to offer one piece of evidence in support of his claims:

Any undergraduate student receiving a B or better in introductory Econometrics would be able to pick the Ayres/ Donohue work apart. This is for a number of reasons, the most fundamental of which is—and this is an error more appropriate for freshman Statistics 1—that their own interpretation of their estimated coefficients is simply wrong. They discuss two variables purporting to measure the effect of concealed-carry laws, but then fail to understand that it is the joint effect of the two variables, rather than merely one of them, that represents the estimated effect in the model.

Ayres and Donohue have extensive discussions on the interpretation of the two variables in their paper. Those discussions appear on pages 1220–1222, pages 1264–1268 and pages 1277–1280. For example (page 1277, my emphasis):

To calculate the five-year impact of the shall-issue law under the hybrid specification it is necessary to add together the impacts of the intercept and trend terms for individual years and then sum the yearly impacts.

Or page 1264 (my emphasis):

according to the hybrid model, in the year after passage the main effect of the shall-issue law is a 6.7% increase in violent crime, which is dampened by the 2% drop associated with the negative trend variable, for a net effect of 4.7% higher crime. After three and a half years, the conflicting effects cancel out at which point crime begins to fall.

I don’t understand how Zycher managed to miss all of this.

Daniel Davies criticises Zycher for the obscurity of Zycher’s allegation about a fundamental error in Ayres and Donohue. And he dubs me “the Ultimate Lott Trainspotter”. Where did I put my anorak?

Mark Kleiman calls for the American Enterprise Institute to convene a panel to investigate Lott.

Randy Barnett highlights Reynolds’ scepticism about Zycher’s claims.

Randy Barnett adds his voice to those calling for the American Enterprise Institute to conduct an investigation into Lott’s conduct. He also writes:

Since Jim Lindgren’s unsuccessful effort to verify some of Lott’s claims from the criticisms of Tim Lambert, I have not defended Lott publicly in any way. Nor would I now rely on his empirical conclusions absent some outside examination of the sort that was eventually given the work of Michael Bellesiles.

Ralph Luker writes

Lott’s sponsors, primarily the American Enterprise Institute, but also the Federalist Society, may now want to distance themselves from him.

Tyler Cowen reacts to the calls from Mark Kleiman, Glenn Reynolds and Randy Barnett for a panel to investigate Lott’s conduct:

My first reaction is to suggest that we already have such a panel every time John, or anyone else, submits a manuscript to a refereed journal on the topic.

Cowen seems to believe that the purpose of the panel would be to investigate whether Lott was correct in his “More guns, Less Crime” research. There is already a panel examining that question. It is the National Academy of Sciences panel on firearms research. Lott mounted a preemptive attack on the panel, accusing the members in general and Steve Levitt in particular of anti-gun bias. No, what is needed is an investigation into the question of whether Lott’s conduct is ethical. In particular:

• he almost certainly fabricated a mysterious survey and certainly behaved unethically in making claims for which he had no supporting data
• he presented results purporting to show that “more guns” led to “less crime” when those results were the product of coding errors and then tried to conceal the fact that his results were the product of coding errors by changing his model

Cowen then says that “market-oriented economists” believe that even if Lott’s work is fraudulent:

• John, at the very least, did show: “More guns, not nearly as much more crime as had been thought.” This was a real contribution and it holds up.
• John did not so clearly show “More guns, less crime.” “More guns, less crime” conceivably could still be true, but it remains to be demonstrated.

However, if Lott’s work is fraudulent then he has shown nothing, since we cannot trust any of his findings. Nor should we be confused by his shorthand “More Guns”. He studied the effects of concealed carry laws, not the general availability of guns in society. Only a very small percentage of the population obtains permits when they are available so obtaining small effects for carry laws is not surprising and does not imply that guns in general do not have large effects either beneficial or harmful.

I also recommend Mark Kleiman’s excellent post on why the character of the researcher matters, especially when doing econometrics.

You have to scroll down towards the end of this monster posting. Zycher has proven to be profoundly careless with his facts in his post, which makes this comment of his rather self-referential:

The modern art of blogging—of which you are one of the truly prominent practitioners—has many virtues, among them the stimulation of discussion and the ability to correct errors and set records straight quickly. But among those virtues one searches in vain for carefulness; the familiar tradeoffs are heavily weighted toward edginess and speed.

I can only hope that he corrects his errors and sets the record straight quickly.

Clayton Cramer tells us that he doesn’t believe the criticism of Lott

because his opponents are clearly fierce anti-gun advocates, and not above a few tricks of their own.You read their work outside of the statistical area, and it’s clear that the are either not very careful thinkers, or are playing fast and loose.

Earlier, he told me:

I don’t know if you actually found Dr. Lott engaged in a lie or not, but I do know that your irrational desire to see victims disarmed and murdered by criminals puts you on my list of people not to bother with anymore.

Cramer doesn’t dispute any of the criticism of Lott made by John Donohue, or by Jim Lindgren, or by myself. Instead, he picks on two minor points in Donohue’s paper “The Final Bullet in the Body of the More Guns, Less Crime Hypothesis“. Cramer is trying to argue, not just that Donohue is wrong on these two small points, but that he is so wrong that it calls Donohue’s entire scholarship on Lott into question. Cramer faces a heavy burden here—he has to show not that there is a difference of opinion, or that Donohue made a mistake, but that no fair-minded person could have made such an error.

On the first point Cramer writes:

Donohue points to the recent example of actor Sean Penn, who has a California concealed weapon permit (apparently issued in violation of California law), and how two of his guns were stolen from his car. While Donohue acknowledges that Penn “succeeded in getting one of the relatively few gun permits in the non-RTC state of California,” raising this issue as part of a discussion of RTC laws makes no sense at all.

Cramer appears to have missed the reason why Donohue raised this issue, even though it was in the same paragraph:

By some estimates 1.5 million guns are stolen every year, which means that anything that increases gun ownership (and carrying a la Sean Penn) is likely to put more guns into the hands of criminals.

On the second point Cramer presents this passage from Donohue’s paper:

During the first 5 and one-half years of the Texas RTC law, the Violence Policy Center was able to identify that 41 permit holders were arrested for murder or attempted murder (the number would be too low if the researchers didn’t capture every permit holder in their count or if some permit holders committed murder and didn’t get arrested, and would be too high if some were falsely accused). The Violence Policy Center, License to Kill IV (June 2002). The current murder rate in the U.S. across all groups is roughly 5 per 100,000, so if one takes 150,000 as the average number of permits over the first five year period, one would expect roughly 7.5 murders per year from gun permit holders (if they killed at the same rate as the average American today), which totals 41 murders over the full period.

and proceeds to criticise it because the 41 permit holders arrested include attempted murders, those who don’t end up being convicted, those who did not use a gun and those who used a gun on their own property (where a carry permit was not relevant).

Now Cramer’s criticism would make sense if Donohue was arguing that the 41 murders and attempts were somehow enabled by the carry laws, but that is not the point of the passage. It is actually part of footnote 18, and footnote 18 is offered in support of this sentence:

But is it clear how concealed handgun laws could increase crime even if only 1 or 2 percent of the public goes to the trouble of getting a concealed gun permit and that group seems not to have unusually high rates of criminal involvement?¹⁸

That’s right, Donohue is arguing the opposite of what Cramer implied. It is not relevant whether the alleged criminals used guns or not if you are looking at how often permit holders commit crimes. Just as in his complaint about the reference to Sean Penn, Cramer seems to have just read a sentence or two and not looked at the surrounding context.

Cramer concludes:

If the claim is that John Lott has violated professional standards in how he has presented his information, Donohue is in no position to cast any stones. Using VPC’s information, while not discussing its serious shortcomings, is clearly misleading.

The important claim about Lott is not that he violated professional standards in how he presented his information (though he did that as well), but that he violated professional standards in how he conducted his research. It is professional misconduct to present a result (the famous 98% brandishing statistic) for which there is no supporting data. It is professional misconduct to present a result (the new, improved 95% brandishing statistic) based on a sample size of just seven. It is professional misconduct to modify your model just so you can contrive a significant result.

Donohue used the information from the VPC to argue that permit holders do not have a high level of criminal involvement. Donohue also noted the limitations of the VPC data when you used for that task. The shortcomings that Cramer complains about are only shortcomings if the data was used for another task, something that Donohue did not do. Cramer’s post was clearly misleading, using an out-of-context quote to imply that Donohue did something that he did not do. Using Cramer’s reasoning he is now in no position to throw stones at, say, Bellesiles.

Postscript: Both Kevin Baker and Greg, while accepting Cramer’s claim to have found errors in Donohue’s work, aren’t buying Cramer’s claim that this defeats the criticism of Lott.

Tom Spencer offers an explanation for Clayton Cramer’s attack on Donohue:

I think it’s safe to say that Cramer long ago crossed the line between reason and emotion regarding his view of John Lott’s situation. Like any true believer, he simply can’t believe the truth right in front of him. Like Bellesiles, Cramer’s hero is a fraud and he can’t accept it.

Mary Rosh was Lott’s most determined online defender, posting many messages, almost all of them defending Lott and displaying close familiarity with Lott’s work. Benny Smith is Bellesiles’ most determined online defender, posting many messages, almost all of them defending Bellesiles and displaying close familiarity with Bellesiles’ work.

Andrew Ackerman has an excellent article in the Emory Wheel, laying out the evidence that Benny Smith is really Michael Bellesiles. Firstly, there does not appear to be any Benny Smith in the Detroit area. That suggests that the name is a pseudonym, but of course does not mean that it is really Bellesiles. Secondly, Smith knows things, such as the cover art for the new edition of Arming America, that only Bellesiles and a few others know. Ackerman concludes:

It is becoming increasingly apparent that one of two scenarios exists: Either Bellesiles and Smith are in cahoots and Bellesiles is using Smith as a mouthpiece in order to garner self-respect in the wake of his downfall, or Benny Smith is Bellesiles.

Benny Smith has actually denied that the second scenario is true. In my recent run in with Benny Smith on HNN, I asked him:

Perhaps you can now satisify my curiosity? Your posts sometimes seem to display a great deal of knowlege about Bellesiles’ doings. Does he send you comments and information to post?

Benny Smith replied:

And regarding your own curiosity, I have not received any materials directly from Professor Bellesiles.

Either Smith is lying here, or he really is Bellesiles.

Postscript: Ted Barlow thinks that Smith is Bellesiles and notes some important differences between Lott and Bellesiles: same people still support Lott and Lott’s employer tolerates his activities.

Lott has an article which purports to show that Rush Limbaugh was right when he claimed that Donovan McNabb was overrated because the media wanted to see a black quarterback do well.

Lott looked at whether media coverage was more favourable to black quarterbacks than to white quarterbacks and found that stories about black quarterbacks were slightly more likely to be positive (67% to 61%). He then did a multivariate analysis controlling for factors like the whether the quarterback’s team won and finds that after doing this, black quarterbacks are 27 percentage points more likely to get favourable coverage.

My reaction to his analysis underlines the damage that Lott has done to himself with his conduct. Even if all the data is correct and his regressions have been correctly calculated his analysis is not in the slightest bit persuasive. The reason is that his behaviour in the coding errors case suggests that he just keeps trying different models and just cherry picks the one that gives the result he wants. Would Lott want to get a result that supports Limbaugh? Well, check out this Mary Rosh posting:

You have got to download this paper. Lott has done an amazing piece here. Fits in perfectly with Rush Limbaugh’s program today.

On his show Limbaugh apparently claimed that giving women the vote was a bad idea and the same day Mary Rosh was on freerepublic.com promoting a Lott paper supporting him. It seems that John Lott is a Rush Limbaugh fan.

Seb from “Sadly, No!” has examined Lott’s data and analysis and sure enough has found evidence suggesting that Lott is up to his cherry-picking tricks again. First of all, Lott only gets his initial 67% to 61% advantage to black quarterbacks because he only considers a subset of the stories. When you look at all the data, the result was reversed, with black quarterbacks getting 53% favourable stories and white ones 57% favourable. The difference seems too small to be meaningful, but it is the opposite of what Lott reported.

Second, Seb noticed that Lott included some variables in his regression for the extent of media coverage that made no sense. Why would having more stories make those stories be more or less likely to be positive? Seb reran the regression without those variables and found that Lott’s result went away—the quarterback’s race no longer had a significant effect. Now this dosn’t prove that Lott cherry-picked his model since we don’t know if he tried doing it Seb’s way, but given Lott’s past conduct it seems quite possible.

In any event, the data Lott collected indicates that there is no clear bias in the media either in favour of or against black quarterbacks.

Benny Smith has replied to Andrew Ackerman’s article that suggested that Benny Smith was really Michael Bellesiles. (My earlier comment is here.) Smith has a good argument against one of the pieces of evidence pointing to him being Bellesiles—he has turned up a website that has the new cover art for the new edition of Arming America. The fact that this wasn’t available from the publisher or Amazon had pointed to Benny Smith being close to Bellesiles.

Now, there is something odd about Smith’s responses to the questions about how he learned about the cover art. Despite repeated questioning, he didn’t post the URL until Ackerman’s article appeared—it seems like he was playing games with us.

Of course, none of this proves that Smith isn’t Bellesiles. Based on all the other things that he has posted I still believe that it is likely that he is Bellesiles but the case is not as strong now.

Update: I wondered whether the image had been available when Benny Smith posted his message, so I asked Benn Ray of Atomic Books how long the cover art had been on the web site and he replied:

The cover image displayed on my site is from the catalog of a distributor of Soft Skull Press’ books. It has been up on our site for several months.

The creation date listed for the image listed here is 12 Jul 2003, so I think we can conclude that the art was there when Smith posted.

Update 2: Ralph Luker comments:

Second, nothing Benny says proves that he is not Michael Bellesiles’s “sock-puppet.” He’ll have to provide more hard evidence to make the case for that. Third, Benny is a poor reader. From the Emory Wheel’s headline through Andrew Ackerman’s penultimate paragraph, the article poses the question of whether Benny is Michael Bellesiles’s “sock-puppet” or his double. It makes no absolute claim that Benny is Michael. It’s still an open question. No retraction or apology is necessary. Finally, it’s an open question, too, whether it is the more embarrassing to be someone’s “sock-puppet” or to be so lacking in critical distance as to be mistaken for one.

Update 3: Jimmy Wales has looked at the IP numbers of some of Benny Smith’s emails. These suggest that it is unlikely that Smith is really Bellesiles. It appears that Smith was posting from Michigan when Bellesiles was still at Emory.

Update 4: Benny Smith has posted another comment where he admits that he deliberately withheld the information about where he found the cover art in the hope the the Emory Wheel would be embarrassed. This makes his demand for an apology (scroll to bottom and read Benny Smith’s comment) rather disingenous.

Chris Mooney has published an article on Lott in Mother Jones. The whole article is well worth reading, but the way that Lott kept changing his story about the coding errors is particularly interesting:

In the face of this evidence, how can Lott continue to claim the coding errors don’t matter? In an interview conducted on August 18, Lott told me that he had posted “corrected” tables on his website for all to see. But when I downloaded Lott’s “corrected” version of the contested table, it showed the same numerical values as that of Donohue and Ayres—that is, the coding errors were gone—but bizarrely claimed the properly coded data still indicated statistically significant drops in murder, rape, and robbery. That’s because Lott had introduced a new twist: Rather than simply fixing the incorrectly coded data, he omitted a key calculation regarding statistical significance used in the Plassmann-Whitley paper. (For statistics geeks, it’s called “clustering at the state level.”) Faced with no other way to save his thesis, you could say that Lott changed the rules—rules his own team had laid down—in the middle of the game.

Confronted with this, Lott’s subsequent actions raise even more questions. On the website, Lott claimed the “corrected” table used “clustering,” when it did not. In a heated interview on August 19, Lott said this labeling claim must be an error. But the very next day, he e-mailed a file containing precisely the same table, claiming that all the tables on his website were “clearly and properly labeled.”

On September 2, Lott changed his story yet again, emailing me that “the file should now be returned to what had been up there before.” But when I downloaded the new file, the key table had been altered to remove the questionable clustering assertion, but had inexplicably reverted to the incorrectly coded Plassmann-Whitley findings that Donohue and Ayres had long since debunked, and Lott himself had admitted to me were incorrectly coded. And despite all these changes, as of October 13, Lott’s website still labels the table as last being corrected “April 18, 2003.”

Glenn Reynolds thinks that Chris Mooney’s piece is “not a bad summary” but opines that it is somewhat slanted. Reynolds thinks that the article should have noted that Saul Cornell’s work is funded by the Joyce Foundation. This is a curious complaint, since Reynolds earlier objected to opponents of Lott making arguments based on the fact that Lott was funded by the Olin corporation foundation. [Correction: I meant to say foundation there. My point is that it is wrong to imply that someone’s opinion has been bought, whether that person is Saul Cornell or John Lott.]

Randy Barnett writes

The time is ripe for AEI to conduct a thorough investigation of all these issues—with a panel of accomplished outsiders—given that so many have relied on Lott’s conclusions without the expertise to substantiate them for themselves.

He also has some interesting comments on Ayres and Donohue’s results and peer review. This statement, however, is not correct:

Lott always claimed that right to carry laws were accompanied by a small increase in property crimes, which he speculated might be the results of deterrence caused by concealed weapons.

Actually, in the second edition of More Guns, Less Crime Lott analyzed data to 1996 and switched to different model that showed property crime declining by almost as much as violent crime.

Clayton Cramer also has comments. First he claims:

At least one significant difference is that anyone of normal intelligence could read the critiques of Bellesiles’s work and immediately see that Bellesiles had altered quotes and misrepresented sources. The allegations against Lott require some in-depth knowledge of statistics to understand.

This is not true. Some of the allegations require statistics to understand, but there are plenty that don’t. You want altered quotes? Lott even misquoted himself. You want misrepresented sources? Lott quoted extensively from this article without ever mentioning that it contradicted his story about what happened in the Appalachian School of Law shootings.

Cramer then claims:

That Lott was able to reproduce similar results with a 2002 survey—even though the 1997 and 2002 results conflict with lots of other studies—suggests that the 1997 survey could have taken place.

This is badly wrong. Lott has repeatedly claimed that that the 2002 survey produces similar results to the alleged 1997 survey, but he is not telling the truth. With just seven defensive gun users, the 2002 survey sample is way too small for any meaningful estimate of the brandishing number. Even if we ignore this problem, it still doesn’t reproduce the 1997 number—if you examine Lott’s data, the results are not what he says they are.

Cramer next claims:

As the Mother Jones article points out, however, even Ayres & Donohue’s work “shows statistically insignificant decreases in murder, rape, and robbery” as a result of non-discretionary permit issuance laws.

This is also wrong. Ayres and Donohue found that if you just corrected his coding errors Lott’s analysis “shows statistically insignificant decreases in murder, rape, and robbery”. Lott absolutely refuses to admit this and this fact does not seem to bother Cramer at all. Ayres and Donohue’s work found (small) increases in crime following the introduction of carry laws.

Tom Spencer is not impressed by Cramer’s posting, describing him as being in a “damn-near pathological state of denial”.

Chris Mooney has a post linking to blogspace coverage of his article.

Mother Jones has posted Chris Mooney’s interview with Lott. Definitely worth reading. First read Mooney’s commentry and follow the links to the transcripts.

Kevin Drum links to Chris Mooney’s article. In his comments, Jane Galt writes:

I agree that Lott has ruined his credibility, but I believe JadeGold has it backwards; Bellesiles work has been cited in several court decisions related to gun control and the second amendment, while I’m aware of no laws that have been made pursuant to Lott’s work, which, after all, was only published five years ago—too soon, generally speaking, to have a major legislative effect.

Concealed carry laws have been passed in five states since Lott’s work was published and his work has been widely cited in the arguments over these laws. The most recent case was Missouri, where the Second Amendment Coalition of Missouri (SACMO) mailed a copy of The Bias Against Guns to each State Senator before the vote to override the governer’s veto of the law. Here is SACMO’s account of what happened next:

Forgive me if I seem to be crowing about this next part, but heck, I am. When I walked into Senator Jim Matthewson’s office on Tuesday afternoon to visit him, I inadvertently interrupted him while he was reading the John Lott book, The Bias Against Guns that we had sent him! Holly Cow! But there is more. SACMO is headquartered in the big swing vote’s district, Senator Mike Gibbons. We have been helping him understand the issue in several ways over quit some time. Now get this!

On the floor of the Missouri Senate, while giving a fiery defense of right to carry, HE REFERRED TO A PASSAGE FROM THE BOOK! WOW! And better yet, several Senators nodded their heads in understanding. I about jumped off the balcony. Now I knew that we had done well, but, my God, I think we knocked that one out of the park. Thank you to all of you that sent money to SACMO to help make that possible.

The margin of victory in the override was one single vote, so Lott’s work seems to have been instrumental in getting legislation passed in at least one state.

Mark Kleiman writes:

What seems to me even more striking, though Mooney doesn’t mention it, is the difference in the way the two are treated in the mainstream press: while no news article about Bellesiles could fail to mention the controversy about Arming America, Lott—who made up an on-line persona who praised him to the skies and claimed on his behalf academic appointments the real John Lott never received, and who still claims to have done a survey with 2000 respondents which reached an utterly implausible finding and of which no evidentiary trace can be found—still gets treated as merely one side in a normal two-academics-disagreeing dispute. [*]

Ralph Luker comments:

Among the active commentators, I tend to find The Volokh Conspiracy’s Randy Barnett the most persuasive. His call for an even-handed inquiry into John Lott’s scholarship may be met by a National Academy of Sciences report from an expert panel which will examine Lott’s work and is due for release in the late fall. It may be the equivalent of the report by Emory University’s panel of experts in the case of Michael Bellesiles and we will see if the American Enterprise Institute is prepared to bite this bullet.

Randy Barnett is calling for an investigation into the question of whether Lott has perpetrated a fraud. That’s not what the NAS panel is investigating. They will assess whether his results are correct, but it is not their business to decide whether his results are the product of fraudulent research or not.

Brian Linse comments

I find the article to be devastatingly on target, and I believe that anyone still defending Lott after reading it must be suffering from fundamentalism induced denial.

Lott has responded to “Sadly, No!”’s comments on his analysis that purported to show that the media was biased in favour of black quarterbacks. On his blog he writes:

During the weeks when quarterbacks played about 67 percent of the news coverage for black quarterbacks was positive and about 61 percent of the coverage for whites was positive. Including all weeks (including when quarterbacks didn’t play and even weeks before some started to play) narrowed the difference to 64 percent for blacks to 60 percent for whites.

I downloaded Lott’s data and checked his arithmetic. His data shows 333 stories about black quarterbacks and 178 of these were positive. That’s 178/333=53% of the news coverage being positive. For whites, 574/1013=57% of the coverage was positive, i.e. coverage of white quarterbacks was more likely to be positive. How did Lott manage to screw up such a simple calculation? “Sadly, No!” has the explanation.

Lott has posted some criticism of Chris Mooney’s article.

Let’s see how many errors he has succesfully identified:

1) Paraphrasing claim from the Chronicle of Higher Education stating that the “coding errors had not been reviewed by a third party.” I was never asked by the Mother Jones reporter about this reference. In fact, after the Chronicle piece was published I immediately e-mailed David Glenn at the Chronicle to point out that two different points had been merged together in his piece.

It’s hard to see what Lott’s beef is here. The Chronicle made Lott’s point narrower than he intended, but Lott did in fact claim that the arguments about the coding errors had not been reviewed by a third party.

2) “The cause, according to then Stanford Law Review president Benjamin Horwich, was a minor editing dispute involving literally one word; Lott, however, complains of an editorial ‘ultimatum’ from the journal.”

The SLR issued a “Clarification” which states clearly that there were “certain revisions,” not that there was one change involving one single word. The “Clarification” also noted that “the impression that some have gotten from Ayres and Donohue’s Reply piece is incorrect, unfortunate, and unwarranted.”

There were disputes over certain revisions but Lott got his way on all of them except one. Horwich clearly states

There was one reinstatement of older language that I simply refused to make because it never made sense in the first place. I’ve already written far too much, so I won’t bore your readers with the details, but basically it was a one-word correction to a sentence that was otherwise incoherent. The impact on Lott et al.’s piece was nothing more than to remove a tangential two-sentence footnote.

Lott’s final sentence is misleading—the incorrect impression that the SLR is referring to is the one that Lott removed his name because he no longer stood behind the paper. Nowhere does Mooney make this claim.

3) “Fraud [Mooney’s article actually says “Faced” here] with no other way to save his thesis, you could say that Lott changed the rules—rules his own team had laid down—in the middle of the game.”

[I deleted a long paragraph of Lott’s here because it contained nothing relevant to Lott’s dropping of the clustering correction.]

Interestingly, in none of Ayres and Donohue’s own regressions, not a single one, do they use “clustering” to determine the standard errors. According to our analysis many results continue to show statistically significant drops in violent crime even when using “clustering,” particularly the year-by-year estimates reported in the figures for murder, rape and robbery. All the results for murder, rape, and robbery continue to show statistically significant drops when the methods used by Ayres and Donohue on their own regressions are employed.

None of this is an explanation for why Lott dropped the clustering correction. Yes, in Ayres and Donohue’s original paper they followed Lott’s earlier work and did not use clustering. In his reply, Lott decided to use clustering. He only dropped the clustering after seeing Ayres and Donohue’s reply, which did use clustering and showed that Lott’s results went away after the coding errrors were corrected. Nor is true that murder, rape and robbery show statistically significant drops when the methods used by Ayres and Donohue on their own regressions are employed. When Ayres and Donohue’s methods are employed, murder, rape and robbery show increases in most states. You only get the significant decreases by employing Lott’s methods and also removing the clustering correction.

4)”Yet Lott’s critique is once again misleading: His own newspaper op-eds aren’t peer reviewed, and Lott admits that Regnery Press, his latest book publisher, does not use peer review.”

Of course newspaper op-eds are not refereed; that is not customary. However, I have used the op-eds to popularize the work that I have published in journals. While More Guns, Less Crime was refereed and The Bias Against Guns was not, Parts of the empirical work in the three empirical chapters were based upon my past work published in refereed journals. For example, all of the material in Chapter 8 was from a refereed article.

This doesn’t even address Mooney’s point, let alone show it to be an error. If we should dismiss Ayres and Donohue’s work because it didn’t appear in a peer-reviwed journal, then we should also dismiss most of The Bias Against Guns since it was not peer reviewed. Yet Lott does exactly the oppostite of this. He repeatedly cites his own unpublished study that purports to show that carry laws dramatically reduce mass public shootings, while ignore Duwe’s peer-reviewed journal publication that found that carry laws had no effect on mass public shootings.

5) “On the website, Lott claimed the ‘corrected’ table used ‘clustering,’ when it did not.” This quote is wrong. I was arguing that the tables were labeled as not having clustering.

Mooney’s statement is absolutely correct. You can see a copy of one of the “corrected” tables here. In the table Lott states “clustering is assumed by state”.

To summarize, Lott has not managed to successfully identify one single error in Mooney’s article. And he has managed to prove this comment from the article to be correct:

“Lott will never say, ‘that’s a good point.’ Lott will offer you some rebuttal,” says Georgetown gun policy expert Jens Ludwig.

Lott has posted some comments by Florenz Plassmann on the controversy about “More Guns, Less Crime”. Plassmann writes:

In their response to our paper, Ayres and Donohue have pointed out that our extended data set contains errors. Correction of these errors leads to estimates that differ from those that we published in our paper in Tables 3–8. As far as I can tell, the public discussion of the three papers in the Stanford Law Review has focused exclusively on this fact, and has used it as a reason to dismiss our entire paper as irrelevant.

Plassmann does not mention that correcting the errors causes the results in tables 3–8 to mostly go away. The discussion of the papers has focussed on the coding errors because Lott’s behaviour in refusing to admit to the errors for four months and dropping the clustering correction has made it into a big issue. An issue that is not just a normal academic dispute about who is right, but one that calls into question Lott’s integrity. I have discussed the other half of the Lott/Plassmann/Whitley (LPW) paper here.

Ayres and Donohue put much emphasis on the results of their “hybrid” model. We explain why we disagree with their interpretation of this model and therefore with their interpretations of their own estimation results (pp. 1326-1329, see especially our Figure 2).

But Ayres and Donohue say that this model is misspecified. They go on to consider a disaggregated model, a model that LPW entirely ignores. They also addressed the “fitting a straight line to curve” problem by adding quadratic terms and found that this did not change the results (see footnote 109 of their paper and further extensive discussion in their reply). LPW just ignored this.

We argue that it is incorrect to analyze count data with Ayres and Donohue’s least-squares method, and that a count analysis with, for example, a Poisson model would be more appropriate (the first two paragraphs on p.1354).

LPW has almost twenty pages of analysis of the data using least-squares. If they believe that it is an incorrect technique, why devote a third of the paper to it? Nor, in any event, do they argue on page 1354 that the previous twenty pages were incorrect. They point out some possible problems with least-squares, use a Poisson model as a check and observe: “All results reported in Tables 7 and 8 are similar to those that we reported using weighted least squares.”

In their response to our paper, Ayres and Donohue have analyzed the extended data set using “clustering by state,” thereby replicating the analysis that we reported in our Table 3a. The public debate has used their results as an argument that the results that we published are flawed (although there has been no discussion of our results in the other tables), and that Ayres and Donohue have made a convincing case against “More Guns Less Crime.”

The results in table 3a are the ones LPW present as giving an overall picture of the effects of carry laws. They can’t salvage things by pointing to some other less important table.

However, Ayres and Donohue (1) do not use “clustering by state” in any of their own analyses, and (2) do not even mention that they use “clustering by state” when they repeat our analysis with the corrected data set.

Why does it even matter whether they explicitly mentioned it? No-one is disputing the fact that they did. It was perfectly clear, both from looking at the results they presented and also from the Stata files they posted that they had used clustering. They also stated that had replicated the analysis in LPW, which implies that they did use clustering, since that is what LPW used.

Ayres and Donohue (as well as everybody who agrees with their work) need to decide:

(i) either “clustering by state” is necessary, which means that Ayres and Donohue correctly use “clustering by state” when they repeat our analysis. In this case, all of Ayres and Donohue’s own estimates are wrong, because these estimates are obtained without “clustering by state.” This makes it impossible to claim that Ayres and Donohue have made a convincing case against “More Guns Less Crime.”

Nice try, but clustering does not affect the value of the estimates at all. If you don’t correct for clustering, your standard errors are too small. That means you might make the mistake of deciding that an effect is statistically significant when it really isn’t, but you won’t miss an effect that is really there. If you don’t use clustering and find that “More Guns” don’t lead to “Less Crime”, adding clustering will not change your result.

(ii) or “clustering by state” is not necessary.

I think it is appropriate to use clustering. But why doesn’t Plassmann tell us whether it should be included? He did use it in the LPW paper so presumably he felt it was appropriate then. Has he changed his mind? If he has, how come?

Chris Mooney has a couple of comments on Lott’s response to his article.

You can also read comments from Nick Confessore on Tapped and Atrios.

Update: And Randy Barnett repeats his call for an independent panel to assess the merits of this controversy.

[Note: This was an attachement to a Nov 17 posting to firearmsregprof on Nov 17. My comments are in italics like this. TL]

by David B. Mustard

I take issue with two points of Jim Lindgren’s representation of my statements about John Lott’s 1997 survey, as posted on instapundit.com.

(more…)

Randy Barnett updates his earlier post where he pointed to Lott’s response to the Mother Jones article to make it clear that he found Lott’s response to be thoroughly evasive and that he is not going to be relying on Lott.

Meanwhile, Lott apologist Clayton Cramer says that some of Lott’s responses “are pretty telling criticisms of the accuracy of some of his opponents.” Cramer is sounding more and more like Bellesiles apologist Benny Smith.

Jim Dallas finds Chris Mooney’s article compelling and tells us how, after writing in the University of Texas paper that “More Guns, Less Crime” had been refuted, he got an irate email from Lott.

On August 18, in his interview with Chris Mooney, when he was asked if there were coding errors, Lott replied:

There are a couple minor errors, the data is on, the data is available for anybody to look at, anybody can go and download the data, I’ve made it so that people can go and easily replicate the results, if you went to the website…

Two days later Lott admitted that there were “a few hundred data entries that contained mistakes”. Now when Lott talked to Mooney, he must have known how many errors there were, since he had corrected them when constructing the corrected tables that he posted to his website in May. He lied to Mooney about how many errors there were.

I’ve been getting some comment spam, so I have installed Fletcher Penney’s blacklist plugin, and modified it so it just blocks comment posting from the blacklisted IPs. If you post a comment and it doesn’t show up, sorry. Email it to me and I’ll post it for you.

After misrepresenting my views in this statement,

I don’t know if you actually found Dr. Lott engaged in a lie or not, but I do know that your irrational desire to see victims disarmed and murdered by criminals puts you on my list of people not to bother with anymore.

Clayton Cramer is at it again. In this post he claims:

Tim Lambert keeps attacking John Lott’s work because it is being used to prevent proper Australian-style gun control from reaching America. I can see why.

For the record, I have no desire to see victims disarmed and murdered by criminals. And the reason why I criticised Lott’s work was because I believed that he was wrong. More recently I have come to the conclusion that his work is fraudulent as well.

Cramer really has no excuse for his misrepresentations. He has been an active participant in the firearmsreg mailing list for years. On that list I have never once said that I wanted the US to adopt Australian-style gun laws. I have also stated that I do not have an opinion on what the laws in the US should be. I have also stated that I did not agree with the restrictions introduced in Australia in 1996, so I don’t see why he would think I would want to see similar laws introduced in the US when I didn’t want them introduced in Australia.

Cramer also makes a very silly argument against Australian gun laws—he posts a story about a drive-by shooting here and implies that the gun laws have failed. But all this shows is that gun restrictions do not completely eliminate gun crime. I don’t think anyone ever claimed that they would, so he is arguing against a strawman. What the people who advocated the laws argued was that they would reduce gun crime and the latest crime statistics from my state show a 26 per cent decrease in ’shoot with intent’ incidents and a 36 per cent decrease in assaults with a handgun. Of course, these decreases may well have nothing to do with the gun laws passed a few years ago, but the fact is that Cramer ignored the evidence that might have been relevant to his claim about the laws here and instead posted something irrelevant.

Glenn Reynolds links to a news story about crime rates in England and declares “More gun control, more crime”. However, the story actually states that there was less overall crime. The story also states that gun crime has increased to 0.15 gun homicides in England and Wales per 100,000 population in the previous year, compared with 3.6 per 100,000 in the US. Reynolds take: gun control is “Not a smashing success, so far”. Gun control may not be responsible for the difference, but it seems a bit much for pro-gunners to point to a gun homicide rate one twenty-fourth of that of the United States as evidence for a failure of gun control.

Tom Spencer has an “I told you so” posting on Clayton Cramer. He points out that he correctly predicted Cramer’s behaviour back in January.

I thought that after his experience with Bellesiles, Cramer would eventually come to the realization that Lott was behaving in the same way as Bellesiles, but I was wrong and Spencer was right.

Update: Tom Spencer has a few more comments, suggesting that Cramer’s hypocrisy is typical of right-wing bloggers.

Clayton Cramer has mounted a defence of Lott on the survey question. Just as Bellesiles tried to persuade people that the attacks on his work were politically motivated, Cramer tries to persuade us that the attacks on Lott are politically motivated. Now, Jim Lindgren has been prominent in raising question about Lott’s mysterious survey, but Cramer wrote Lindgren out of the survey controversy and replaced him with John Donohue. Cramer doesn’t mention Lindgren once, instead asserting that Donohue is using the lack of evidence for a survey “to cast doubt on Dr. Lott’s integrity.” However, contrary to the impression that Cramer leaves, Donohue has hardly mentioned the survey, while Lindgren has written an extensive report and further comments on the survey. Why does Cramer replace Lindgren with Donohue? Well, he can’t paint Lindgren as partisan, while he has made a (specious) case that Donohue is partisan. I am the other critic that Cramer mentions and according to him I am only attacking Lott because of my “desire to see victims disarmed and murdered by criminals”.

Continuing with his “it’s just a partisan attack” theme, Cramer writes:

Unfortunately, Dr. Lott stumbled into a highly politicized area, and when you are confronting people like Tim Lambert or John Donohue, you need to have very persuasive evidence. Why? Because both of them are partisans on this issue (Professor Donohue’s protestations are not persuasive), and they are going to demand very high standards of proof.

What has been asked of Lott is not that he meet some super high standard of proof, but the normal standard in any scientific endeavour. If you present a finding from a study you conducted, then you are expected to have the data to support your finding. In fact, it is scientific misconduct to report research that cannot be supported because the data no longer exists.

Cramer then proceeds to downplay and ignore the evidence that points to Lott not having done a survey while exaggerating the evidence for the survey’s existence.

While Cramer mentions the singular lack of evidence for the existence of the survey, he does not mention the fact that Lott attributed the 98% figure to other surveys for over two years. Lott only said that the number came from his own survey when Otis Dudley Duncan pointed out that the other surveys did not yield a figure even remotely like 98%. He does not mention the fact that Lott claimed that the survey was conducted over 3 months in 1997, but managed to present the 98% statistic (and attribute it to other polls) on Feb 6, 1997, well before his survey was completed. Or the fact that nine published surveys give results that differ markedly from 98%. Or the fact that he repeatedly changed his story about how the survey was conducted. Or the fact that he has been caught lying about other matters rather than admit to a mistake.

Cramer presents three pieces of evidence suggesting that Lott did a survey. Each piece of evidence is far weaker than he thinks.

First, there is David Mustard’s statement that he believes it likely that Lott told him about the survey in 1997 and is certain that Lott told him about it before October 1999. However, as Lindgren explains, Mustard’s account of when he heard about the survey has kept changing into versions more and more favourable for Lott:

When I discovered that Mustard had told Frank Zimring on the telephone in the summer of 2002 that he knew “nothing” about Lott’s 1997 survey, I called Mustard and we had a series of long talks.

Mustard confirmed the substance of his conversation with Zimring, but said that his general statement of knowing nothing about Lott’s 1997 survey followed a series of specific questions from Zimring about the survey, which he couldn?t answer. Mustard said that he meant that he knew nothing specific about the survey since he was not involved in it. In Mustard’s conversations with me, he also backed off his claim that he was fairly confident that he heard in 1997 about the survey, saying that he was certain that he learned about it before his October 1999 testimony, but he couldn’t remember whether he heard about it weeks, months, or years earlier. He said that his memory of talking with Lott about follow-ups in 1996 was firm and his memory of what he knew in 1999 was firm, but between late 1996 and late 1999 he did not know when he first learned of the 1997 survey. Nonetheless, Mustard then released a public statement covering much the same ground as he had covered with me, but adding claims about both 1998 and 1997. About 1997, Mustard wrote: “I believe it likely that John informed me of the completed survey in 1997.” I have not talked with Mustard since, so I never learned the basis for his recovered belief that Lott informed him in 1997 or his statement about 1998. I can only say that Mustard did not have either of those recollections when I spoke with him at length a few weeks before.

We do know that Mustard heard about the survey in 1999, but that actually makes things worse for Lott. Even though the survey was supposedly conducted in 1997, Lott made no direct or indirect reference to it until May 1999, when Duncan pointed out to him that his claim that the 98% came from national surveys was wrong. I suspect that he invented the survey then so that he would not have to admit that the statistic was wrong. Soon after that, he likely told Mustard about the survey. I got an email from him out of the blue in June 1999, telling me that he had conducted a survey. I have compiled a list of the direct references (where he says he conducted a survey) and the indirect references (where he states that the number of defensive gun uses is about two million, the number from his survey, rather than the 2.5 million that Kleck found). Notice that the first mention is May 13, 1999. Although it is mentioned frequently after that date, there is no mention before that date, even though he made many “98% brandishing” claims before then.

Second, there is David Gross’ statement that he believes that he was surveyed by Lott. However, as Lindgren explains, his account of the survey he was in doesn’t match Lott’s survey very well at all:

When I asked him if he remembered anything about who called, he said that he “was beginning to think” that the call came from students in Chicago, perhaps at Northwestern or the University of Chicago, but he was very uncertain about whether the call came from a Chicago area source. In his public statement issued after he talked with me more than once, however, Gross’s very uncertain memory became a bit more certain, suggesting that the call probably came from the University of Chicago. That and the timing (which he was also not certain about) were the only things that pointed to him having been called by Lott as opposed to another survey organization.

As I delved into the other studies being done in the 1996-97 period, I found that Gross’s description of the questions that he was asked fit a 1996 Harvard study by Hemenway & Azrael better than Lott’s account of his study questions. First, Gross said that the person who called him was interested in a defensive gun use that happened a few years before he was surveyed, but was not interested in a defensive use that occurred many years before that. This would not fit Lott’s survey, since Lott asked only about DGUs in the prior year. It would fit the Harvard study perfectly, which asked about DGUs in the prior 5 years, but excluded events before that. Further, Gross said that he gave a narrative account of the event, which the caller was interested in. Lott’s study had asked closed-end questions, which would make the narrative superfluous, while the Harvard study was one of the first to ask for a narrative account of DGUs. Last, Gross reported that there was a question about state gun laws, which Lott did not ask, but the Harvard study did.

To be fair, Hemenway doesn’t think it likely that Gross was surveyed in the 1996 Harvard study. He writes:

He doesn’t seem to have been. No one from his state, of his gender and general age responded Yes to the self-defense gun use question, and no one in the whole survey told a story highly similar to his story…

None the less, there doesn’t seem to be any reason to think that Gross was surveyed by Lott rather than by another survey. Another important point is that Gross was one of the prime movers behind the recently passed concealed-carry law in Minnesota. It is highly unlikely that someone with this great a motive for preserving Lott’s credibility would be included in a random sample of 2,000 people.

That doesn’t mean that Gross made the story up—there is another possibility. Gross may have been contacted because there was a news story about his defensive gun use. You see, if you are designing the questions for a defensive gun use survey, you might want to test the questions out on a few people to see if they elicit the information you are after. The trouble here is that if you call people at random you will have to call hundreds of people before you get enough people who have used guns to properly test your questions. So what you can do is call some people who you know have used a gun for defence and try your questions out on them. I found a news story published in 2002 that reported Gross’s activism and his defensive gun use. Obviously that story could not have caused someone to contact him in 1998 (or 1997 or 1996) and ask questions about defensive gun use, but if there was a story in 2002 there could have been an earlier story as well.

So it’s conceivable that Lott was thinking of doing a survey in 1997 and Gross was contacted to test some possible questions. (This would also explain why the questions that Gross reported don’t match Lott’s survey.) But even this most generous interpretation of the evidence doesn’t help Lott, since even if you show he was planning to conduct a survey, there is still no evidence that he actually conducted a survey, and plenty of evidence that he didn’t.

Cramer’s third piece of evidence is the one that he mistakenly regards as the strongest. He writes:

The 2002 survey that Dr. Lott did (and it is well established to have happened), gave roughly similar numbers to the 1997 survey results. It seems unlikely that Dr. Lott could construct a survey that would give him numbers to match a previous survey, unless that previous survey was real. He could certainly just make numbers up to match—but running a real survey certain to give the right numbers would require an amazing level of knowledge of how real people would answer the questions.

The trouble with this piece of evidence is that it just isn’t true. Lott has repeatedly claimed that the 2002 survey gave similar results to the 1997 survey. But it didn’t. The 2002 survey “shows” that people fire their guns five times as frequently as Lott claims his 1997 survey found. (I put the scare quotes around “shows” because the sample size of the 2002 survey is actually far too small for it to tell you anything useful about how often people fire their guns, but let’s ignore that for now.)

So Lott didn’t have to construct a survey that gave numbers that matched a previous survey. All he had to do was construct a survey that gave different numbers, make up some numbers that matched his previous survey, and then claim that they came from his new survey. Furthermore, this fact has been explained to Cramer in online discussion before but he keeps repeating Lott’s claim that the 2002 survey gives similar results.

Cramer’s final argument is that for something as serious as fraud, it is necessary to prove the case beyond a reasonable doubt. But this isn’t a criminal trial. If you think the evidence only shows that Lott probably fabricated the survey, then you should conclude that Lott probably fabricated the survey. You don’t have to acquit him or anything.

by Michael Maltz

Anyone who has looked closely at the data used by John Lott in coming up with his “findings” in More Guns, Less Crime (MGLC) would come to the same conclusion, MGLC = GIGO. First, the data are so full of holes as to be unusable for the analyses he conducted. Second, Lott badly miscalculated the crime rates. Third, he ignored a major discontinuity in the data. [And fourth, as Ayres and Donohue have shown, even if the first three points were not valid (which they are), he did it wrong.] It’s a bit technical, so bear with me.

(more…)

Michael Maltz has sent me some comments on the severe problems data Lott used for his “More Guns, Less Crime” thesis. Note that this is a different set of problems to the coding errors that Lott has been dodging recently.

Also of interest is Lott’s response to the severe problems—he ignored them and accused Maltz of fraud.

The Oct 12 Wichita Eagle has a story with a brief comment from John Donohue on Lott’s work:

Among his most vocal academic critics is John Donohue, a Stanford University law professor whose published critiques of Lott’s works cite errors in handling crime data.

“It’s abundantly clear that there is no support for his thesis,” Donohue said. “It borders on fraud for anyone to try to make the case that there is a drop in crime.”

He said Lott’s earlier work failed to account for the peak and subsequent decline in violent crime related to the advent of crack cocaine in the 1980s and said later updates had coding errors.

Lott responds on his blog:

See Plassmann and Whitley’s paper (p. 1361) for a discussion of the claim that “Lott’s earlier work failed to account” for cocaine.

If you look at at the page he cites, you find that Donohue is correct. Lott did not account for the peak and subsequent decline in violent crime related to the advent of crack cocaine. He included a variable for cocaine price, which does not measure cocaine consumption. Even Lott admits that it does not measure demand for cocaine.

Lott continues:

The language used by Donohue is very disappointing, but it has become extremely typical of the type of statements that he makes and this particular statement is probably relatively mild. I suppose that he feels that these statements will draw more attention to his claims. One has to wonder what other academics think about this.

Well, this academic feels that Donohue’s description of Lott’s behaviour is accurate. And it’s not hard to figure out what Michael Maltz, or Mark Kleiman, or Brad De Long think of Lott’s conduct.

Kevin Connors admits that he has been “quite remiss in following the efforts to debunk Prof. Lott’s work”, but unfamiliarity with the case isn’t going to stop him from having an opinion on the matter. Connors takes issue with Brian Linse’s description of Lott’s work as fraudulent:

simply because a theory is flawed, that constitutes no grounds for labeling it fraudulent.

If Connors had been following the case against Lott he would know that it isn’t just that his work is flawed (that was shown long ago) but that it is dishonest. After correcting his coding errors made his “More Guns, Less Crime” result go away, Lott changed the way he did the calculation to bring his results back. And when I asked him for an explanation, he changed it again and tried to pretend that it had never been changed.

Pro-gun writers frequently make false accusations that researchers they oppose do not share their data. For example, Lott claimed that Ayres and Donohue’s data was not available even though it was on Ayres’ webpage.

Another example is Carl Moody, who claimed that Kellermann was lying because his data was not available. When I pointed out to Moody that Kellermann’s data had been available for years and years he refused to apologize to Kellermann.

In More Guns, Less Crime Lott also claimed that Kellermann’s data was unavailable, even though it was available on the web. This was explained to him on Usenet, but he insisted that something was missing while refusing to say what was missing.

In the comments to this post, Andrew Salemme gives another example. In Armed: New Perspectives on Gun Control, Don Kates claims that Kellermann only released the data for the 316 matched pairs used in his analysis. This is an extremely careless claim, since all he had to do was look at the data to see that it included information about 420 murder victims.

Donald Luskin has a weblog mainly devoted to criticism of prominent economist Paul Krugman. I have found it to be a good example to help guide my criticism of John Lott’s work. A good example of how not to do it, that is. Luskin indulges in frequent name-calling: for instance he called Krugman a “nervous, stammering, shifty-eyed, twitching, ill-tailored, gray homunculus” and Brad DeLong a “pudgy dweeb”. He frequently uses ad hominem attacks on Krugman, for example, calling him “Shorty” and posting Krugman’s teaching evaluations. How does the fact that Krugman is short prove that he is wrong? Reading stuff like Luskin’s makes you more likely to think that Krugman is right—if there were actual good arguments against him, then maybe Luskin would have offered those instead of the ad hominems.

I’ve done my best to avoid Luskin-like behaviour here, both because it is wrong and also because it is counter-productive. If I slip up, I hope my readers will let me know.

There is an obvious word to describe someone with such as intense interest in someone else as I do here and Luskin does on his blog and that’s “stalker”. For example, here’s Kevin Drum:

I thought the fact patterns indicated that Lott was lying, but Tim, the acknowledged expert in Lott stalking, was more cautious.

and here’s Glenn Reynolds:

Donald Luskin is stalking Paul Krugman—claiming that Krugman is feeling the heat over errors in his columns (Luskin calls them “lies”), but still blowing smoke.

All good fun, you would think. But now Donald Luskin is threatening Atrios with a lawsuit unless he removes a post about Luskin entitled “Diary of a Stalker”. The premise of the letter from Luskin’s lawyer is that Atrios (and presumably everyone else who referred to Luskin this way) accused Luskin of committing “the crime of stalking”. This is obviously false, but presumably the real threat is to expose Atrios’ real identity.

I think that bloggers, whatever they think of Krugman, might not approve of Luskin’s behaviour. If they don’t, a simple way to express this disapproval would be to refer to Luskin as a stalker, as I do in the title of this post.

Update: Half of the folks on my blogroll have blogged on this. Naturally, everyone is critical of Luskin. I’ve followed Mark Kleiman’s suggestion and delinked Luskin. If you want to see what he says on his blog (not recommended), you’ll have to find it yourself.