« Lott v. Levitt - Count I Dismissed | Main | More on Judicial Independence »

12 January 2007

Comments

Robert VerBruggen

For the record -- as you obviously have no clue about this specific case, the context in which my comment was made -- Lott claimed his results were statistically significant.

Robert VerBruggen

No, you're wrong, at least insofar as a researcher claims his findings are statistically significant.

You're right that if I walk down the street and count the proportion of people with blue eyes, I might get a different result than someone else does without making any mistakes. But if I claim my results are accurate within a certain bounds, and others cannot replicate this, I am inept or fraudulent.

zaoem

Robert Verbruggen said "If something is not reliable, that means the experimenter was either inept or fraudulent.'

This is simply wrong. Reliability refers to the quality of a measurement method that the same data would have been collected each time in repeated observations of the same phenomenon (this is the standard definition from Babbie's intro textbook). There are many social phenomena that are difficult to measure accurately and for which we would obtain quite different measurements each time we try, even if a researcher is not fraudulent and uses the best available measurement method.

Robert VerBruggen

To be completely honest, I haven't read Freakonomics in its entirety. I read the paper the abortion-cut-crime theory is based on, and I've read a lot of Lott's work on guns. So I can't tell you how I reacted to the passage cold -- I never reacted at all until Lott blogged about the passage and why it was wrong.

I don't have extensive training in statistical/hard science (majored in journalism and political science), but I do know that experiments and other studies are supposed to have (A) reliability and (B) validity. Reliability (you could call it "replicability" to square with the language of the suit) means that when the experiment is replicated precisely, different experimenters get the same results. Validity, on the other hand, is whether the experiment measures what the experimenter is trying to.

If something is not reliable, that means the experimenter was either inept or fraudulent. Challenging validity is much less of a charge.

There is no doubt that Lott's work has reliability (when other people "replicate" it, they get the same result), and scientifically speaking, Levitt's passage said otherwise. What's debatable is whether Lott's equations/regressions/etc. actually measure the effect of concealed carry on crime.

In short, I believe Levitt's wording is off, and he should have changed it when Lott informed him of it. I do understand why many (including me) hesitate to go so far as to call it libel, but Levitt persisted in publishing something knowing it was incorrect.

William Ford

Robert,

While I said "I don’t think it matters" whether one applies the standard of the reasonable general reader or the reasonable academic reader, this was probably a bit off. I think the case is easier under the reasonable academic standard because it is easier to document the word's multiple meanings to academics. I don't really know how the typical general reader would read this passage, but the general dictionary definitions leave room for an innocent construction.

In your post, you said "replication" has the specific meaning Lott says it has. Did you read the passage before Lott sued? If so, did you originally read page 134 the way Lott reads it? Regardless of when you read it, what persuaded you it has this narrow academic or scientific meaning? (Lott's court filings or maybe another source?)

Thanks,

Bill

Robert VerBruggen

Thanks so much for the link. I do agree that the "popular view" is that Lott defamed Levitt to the academic audience. I think that view is accurate, but the judge didn't come out and say that.

For the record, while the judge struck down this allegation, the other one -- regarding Levitt's e-mail to another economist, which claimed Lott had bought an issue of an academic journal to publish views he agreed with -- will proceed.

What bothers me about Levitt here is not that the passage ran. As a journalist, I'm more than aware of how subtle wording issues can really, really tick off sources. There would be a real chilling effect on discourse if nuances like this always led to lawsuits.

What bothers me is that, after Lott pointed out the error, Levitt left it in for future editions. If he'd changed the sentence to "some other academics, however, have come to different conclusions" -- a far more accurate assessment of the body of evidence -- the whole thing could have been avoided.

To the best of my knowledge there's no real legal difference between publishing something once vs. publishing it again after being informed it's incorrect (except that actual malice is easier to prove, ie publishing with knowledge it's wrong), so this is more of a right/wrong argument.

William Ford

I don’t agree that a formal loss in court can never be a victory or a vindication (which I recognize are not always the same thing). It depends on the plaintiff's goal(s). If the plaintiff’s goal is to grind his or her opponent into the ground with legal expenses, then a formal loss might still be a victory. As a more relevant example, if the plaintiff’s goal is to vindicate his reputation, then a formal loss might be a partial vindication, if the court implies that the plaintiff was defamed to one relevant audience.

My original question was whether people will read the outcome as a partial vindication. We'll have to wait and see. Of course, I don't agree that Lott was defamed because I don't agree with Lott's reading of page 134 or his claim about the meaning of replication. Thus, I don't think Lott needed the sort of vindication that some people may tease out of the court's opinion.

At Truth on the Market, Keith Sharfman has some thoughts on the reputational interests at stake in this case.

http://www.truthonthemarket.com/2007/01/12/lott-v-levitt-low-stakes/

Joshua Macy

I don't see how a summary judgment against a party can ever be viewed as a partial vindication for that party. If he had prevailed on the motion, Lott would still have had to prove it in court--but the ruling is that even if it went to trial and he proved every fact and inference that he alleged, as a matter of law he couldn't prevail. I don't know about the popular view, but that's about as far from vindication as I can imagine.

The comments to this entry are closed.

Conferences

December 2014

Sun Mon Tue Wed Thu Fri Sat
  1 2 3 4 5 6
7 8 9 10 11 12 13
14 15 16 17 18 19 20
21 22 23 24 25 26 27
28 29 30 31      

Site Meter


Creative Commons License


  • Creative Commons License
Blog powered by Typepad