Luis Angeles nobly confessed to a serious coding error in his paper on kids and happiness. At least to me, that’s a (moderately) negative signal of competence and a (strongly) positive signal of honesty. If you were a journal editor, how would you treat Angeles in the future? Why?
May 3 2010
Russ Roberts interviews Nassim Taleb in a fast-paced discussion. They kick around a number of interesting ideas. One of them is that leverage is linked to overconfidence. When you believe you can predict the future, you borrow a lot. When you doubt the future, you do not. Taleb is also a fan of small firms. He se...
May 3 2010
My admirable colleage Garett Jones points out that according to my source for German government employment, Greek government employment is only 14% of the labor force, not a third as the New York Times originally reported.The NYT apparently deleted the offending paragraph rather than correcting the statistics. An...
May 3 2010
Luis Angeles nobly confessed to a serious coding error in his paper on kids and happiness. At least to me, that's a (moderately) negative signal of competence and a (strongly) positive signal of honesty. If you were a journal editor, how would you treat Angeles in the future? Why?
READER COMMENTS
rapscallion
May 3 2010 at 2:43pm
I’m not sure how strong a signal of honesty it is. If you know there’s a big mistake, not telling people about it is risky because future researchers might discover it, and then things will only be worse. He probably took that into account.
Nevertheless, unless any other papers of his have been significantly challenged in the literature, I wouldn’t take this as too indicative of his general ability. He’s published a fair amount and this is just 1 paper.
Mo
May 3 2010 at 7:12pm
What about Emile Oster and the missing girls? She has written a newer paper saying her first paper was total wrong.
Or, what about the Levitt paper on abortion where there was a coding error? Or the one where he instruments with police elections and it isn’t valid? Or all the places were people don’t cluster their standard errors?
As far as I can tell these folks are still flying high and publishing even though they have been shown to be wrong.
RL
May 4 2010 at 9:13am
A quasi-related story: One of the findings one looks for on certain medical imaging studies to diagnose an arteriovenous malformation (AVM) is an appearance of the vessels that looks like a coiling snake. This had been described in my training as a “serpiginous” appearance. I recall many years ago finding a letter to the editor in one of the premier radiology journals written by a famous academic neuroradiolgist, seeming to take to task the authors of a recent publication where they referred to this appearance as “serpiginous.” The letter writer began by explaining that the proper word for something that looks like a coiled snake is “serpentine” and that “serpiginous” actually is a dermatologic term referring to a lesion with a raised edge. As I read this I’m thinking, “Well, that IS pretty interesting, but what a pedant this guy is…” Then the letter writer went on to say, “I must take responsibility for this error, as the original misuse of the ‘serpiginous’ descriptor was in my original paper describing the findings of AVM in….” At that point, my respect for the letter writer went up immensely. He’d made an error 20 years earlier in a seminal paper, and literally hundreds of authors had followed him blindly to the point that the wrong word was the accepted term in all the textbooks. And the guy writes in to correct the error…
Alan Crowe
May 5 2010 at 10:28am
This reminds me of Jaynes dissolving Hempel’s Paradox by insisting that the rival hypotheses be made explicit. Jaynes cites Good, I. J. (1967) ‘The white shoe is a red herring’ Brit. J. Phil. Sci. 17, 322.
What are scientists like? Crudely there are two types, A and B.
With type A their work is 90% correct. They discover their errors in 5% of the cases and issue corrections. The other 5% cause confusion until forgotten.
With type B their work is 80% correct. They blunder across and conceal errors in 10% of their work. The remaining 10% of errors cause confusion until forgotten.
Obviously in this world Angeles admission identifies him as type A and is a positive signal of competence. Had you considered the possibility that it could be a positive signal of competence?
Comments are closed.