Worst. Defence. Ever.

Howard Nemerov has a post defending Lott and responding to Chris Mooney's Mother Jones article. Unfortunately, he gets his facts wrong, leaves out inconvenient facts and indulges in fallacious arguments. I'll go through his post and correct these, but first some general comments.

  1. Even though his article is a response to Chris Mooney's article Nemerov does not link to Mooney's piece. If he did, his reader's might have been able to discover how badly Nemerov misrepresents the article.
  2. Nemerov tries to pretend that the dispute is just about politics. He doesn't mention any critic by name, instead Lott is being attacked by a single entity he calls "the left". Nemerov does not mention any of the criticism of Lott from pro-gun and right-wing folks.
  3. Nemerov repeatedly hurls insults at "the left": "hypocrisy, smut, and innuendo" ... "Let's give the left it's [sic] 15 minutes of whine" ... "what we have here is liberals who are not intelligent and educated enough to understand statistical analysis" ... "the left is too stupid to keep its lies consistent". Such insults only subtract from Nemerov's argument---they do nothing to persuade me that he is correct and suggest that he does not have any better arguments. And "smut"??? Where does Nemerov think the "smut" is in Mooney's piece?

Nemerov starts off on the wrong foot by managing to completely miss what Mooney's piece was about:

An article from Mother Jones magazine's website attempts to discredit the work of John Lott, author of "More Guns, Less Crime." The brouhaha seems to center around one alleged lost survey, in which he found that "98 percent of the time that people use guns defensively, they merely have to brandish a weapon to break off an attack." This survey was documented as lost in a computer crash.

Mooney briefly discussed the mysterious survey, but his article was about the coding errors Lott made in his "More Guns, Less Crime" data and Lott's behaviour when his errors were discovered. Nemerov does not mention the coding errors in any way, shape or form. Nor is true that the survey "was documented as lost". There is no documentation showing that it was lost. In fact there is no documentation showing that it existed in the first place. And notice that Nemerov found the article on the Mother Jones website. He has no excuse for not linking to it.

So let's suspend critical thinking for a moment and other authoritative surveys about defensive gun use reached similar conclusions.

Other surveys did not reach similar conclusions. Lott said his survey found 2% firing. The lowest number any other survey found was ten times higher:

Survey Percent firing Source
Kleck 24 Kleck 1995
NSPOF 27 Duncan 2000
NCVS 1987-1990 28 Duncan 2000
NCVS 1987-1992 38 Rand 1994
NCVS 1992-2001 21 NCVS online analysis system
Field 34 Kleck 1995
Cambridge Reports 67 Kleck 1995
DMIa 40 Kleck 1995
Ohio 40 Kleck 1995

After summarizing Bellesiles' misconduct Nemerov writes:

John Lott is accused of not coming up with a single piece of data, which was not the heart of his book.

Lott is, like Bellesiles, being accused of fabricating research results. It's not just about not coming up with the data, (which is serious enough by itself) but whether he did the research at all.

Secondly, until the very end liberals staunchly defended Bellesiles, and some continue to bemoan that his fate is merely the doing of the vast right wing/gun lobby conspiracy.

Gee, just like Nemerov staunchly defending Lott to the end while complaining about how "the left" is attacking him.

Nemerov than accuses "the left" of using a double standard because:

Al Franken receives a Harvard "fellowship" and uses Harvard students to write a book published under his name, and is rewarded by a place on the New York Times best seller list. John Lott uses a pseudonym in an Internet chat room---a common practice---and is guilty of fraud.

Dear readers, here's where I have to appeal for help. If you can figure out what the parallel between Franken getting a fellowship and Lott using a pseudonym is supposed to be, leave a message in the comments. As far as I can tell, Nemerov just took a rocketship to Planet Non Sequiter. Oh, and the fraud part of Mary Rosh was not the mere use of a pseudonym, but the dishonest use of it with the five-star reviews of More Guns, Less Crime and claiming that Lott was "the best professor I ever had" and so on.

Another supposed sin is John Lott's tendency to use complex explanations of his work. "They tend to be mind-bogglingly complicated, involving things like ordinary least squares and Poisson distributions." I only took one statistics class in college, but it seems to me that if he did not include this technical discussion, they would be beating on him for not being scientific enough to be credible.

Well, no. Chris Mooney actually provided a link to Goertzel's Sceptical Inquirer article so Nemerov could have found out what it was arguing, but it looks like Nemerov either didn't read it or didn't understand it. Goertzel is arguing that the dueling statistical arguments are unpersuasive.

It sounds like what we have here is liberals who are not intelligent and educated enough to understand statistical analysis, and rather than acknowledge their own educational shortfalls for handling the topic, they prefer to insult someone who is their intellectual superior. But this is the socialist way: if one climbs too high, bring him back down into the mediocre morass with the rest of us. After all, the desire to excel is selfish arrogance, and traumatizes and victimizes those without the drive to succeed. Education is not necessary, because we are entitled to happiness and income derived from suing McDonald's for spilling hot coffee on ourselves and eating ourselves into obesity.

There isn't any actual argument in this paragraph to respond to, I just thought it was funny the way he worked the McDonald's coffee thing into his rant.

These attacks are followed by more standard liberal hypocrisy: ''The Stanford Law Review critique, authored by Yale's Ayres and Stanford's Donohue, analyzed more recent crime statistics, extending Lott's original 1977-1992 crime dataset to include data through the late 1990s. As it turned out, after 1992, partly due to the end of the 1980s' crack cocaine-related crime wave, crime rates dropped dramatically in states with large urban centers, many of which had not passed right to carry laws."

Thanks for highlighting the fact that the left is too stupid to keep its lies consistent. If the violence issue is controlled by drugs and not guns, then why all the whining about the need for more gun control?

It's unwise to call others "stupid" when there is a chance that you might have missed the point as Nemerov has here. Just because drugs affect crime, it does not follow that other things do not.

I agree with the article on one issue: we do not need to discuss least squares, regressions, and distributions. The heart of ''More Guns, Less Crime'' resides in FBI crime data, collected from all counties in the U.S. over many years. It is available to any who want to check Lott's conclusions, so regular folks like us can see for ourselves that crime rates drop when law-abiding citizens exercise their Second Amendment rights.

The issue where he says he agrees with the article is the issue that prompted his rant where he accused liberals of not being intelligent enough to understand statistics. Now he is agreeing with the article on this? Consistency is not Nemerov's strong suit. Anyway, in case the thought was at the back of your mind, this shows that Nemerov is not another one of Lott's sock puppets. Lott would never say that we don't need regressions.

For example, you can clearly see that for the 2001-2002 time period, Michigan, with a new CCW law, saw a 5% drop in total crime per 100,000 residents, and a 2.6% drop in violent crime, while murder rates stayed the same. Neighboring Ohio, without a CCW law, saw less than a 1.7% drop in total crime, and less than two tenths of a percent drop in violent crime, while murder was up 15%!

I don't think that Nemerov is Lott, but he sure can cherry-pick like him. Michigan has two neighbouring states without a CCW law. The other one is Wisconsin. How did Wisconsin do compared with Michigan 2001-2002? Wisconsin saw a 2.7% drop in violent crime, a slightly greater decrease than Michigan, while murder went down 22%! Why didn't Nemerov tell us about this?

These comparisons involving just two states and two years are useless---you can always find a comparison that says whatever you want.

Nemerov then writes "would you care more or less about a bunch of intellectual ivory tower types splitting statistical hairs?" I wish he would make up his mind. Do those darn liberals believe that "Education is not necessary", or are they "intellectual ivory tower types"? Hmm?


More like this

One thought regarding the apparent reversal of Lott's findings due to the decline of crack; that proves only that the addition of more data with a clear driving variable (no more crack) with the same model will yield different results.

At its heart, this is a tautology. Things changed, so we know things changed. It does not negate the original results unless we can demonstrate that the same variable confounded the original results. This not being done, the Stanford study is largely without merit.

I, as a gun rights backer, am personally troubled by some of Lott's actions. However, the use of a pseudonym is not one of them. Any number of influential authors have done so, and their esteem does not--should not--suffer for this.

By Robert Perry (not verified) on 15 Dec 2003 #permalink

Robert, there's a clear difference between using a pseudonym for general publication and using one to misrepresent yourself and to talk up yourself. The first is perfectly acceptable, but the second is blatantly dishonest.

Robert, no-one has good data on the incidence of crack use. What A&D found was that if looked at pre-1992 data carry laws were associated with a decrease in crime, but looking at post-1992 data you got an increase. That strongly suggests that there is some other, unmeasured, variable producing Lott's results. "Crack" is the one that seems to make the most sense. Whatever it is, Lott's results have been refuted.
And as jon pointed out, it's not Lott's use of pseudonyms that is wrong, but his misuse of them. That's why I have been referring to them as "sock puppets", since that term also describes the way he used them.