Here (from the Annals of Applied Statistics).
“Thus, arguably, all of Section 3 is wrong until proven otherwise.” As with retractions in general, it makes me wonder about the rest of this guy’s work.
Dr. Anil Potti would be pooping in his pants spinning in his retirement.
A Forensic Analysis please? How did this happen? In hindsight.
Admit it, you did this just to show those other guys how it ought to be done right?
I think that errors will creep into any body of work built on the academic model (where we simply don’t have the resources for robust, independent replication prior to publication). The two key things, in my not so humble opinion are to:
1) Do the best job you possibly can and never stop double checking
2) Have the moral courage to own up to errors
Given the constraints, I see this as the best compromise path going forward.
I think “where we simply don’t have the resources for robust, independent replication prior to publication” is not a necessary state of things.
Just like devoting time to be a reviewer is considered a duty of being a good academic a convention ought to evolve where researchers devote a fraction of their time / efforts to replication.
You say you often want to have more people apologize for errors, cite your own examples of places where you had errors and then resolved them through discourse, and then post other people’s admission of errors and say ‘As with retractions in general, it makes me wonder about the rest of this guy’s work’.
I may be getting errors and retractions mixed up and if so I apologize, but I don’t fully understand the distinctions here.
You might want to follow that link ;).
I was being ironic. But intonation is notoriously difficult to capture in typed speech. I actually have a lot of respect for this guy’s work. Too much, probably.
haha, my bad :)
Is Joesph Nevins also soiling himself? Very much not in retirement?
Can’t the analysis be redone without the error? If so, why is it not included with the correction?
I’m guessing it took a few minutes to carefully craft the correction, and would take days or weeks to redo the analysis and write it up.
Yup. In retrospect we should’ve replicated the exact analysis in the paper and discussed how the conclusions changed. The short answer is that the results changed so much that much of the analysis seemed wrong, in the sense of no longer making sense from a political perspective.
We put in quite a bit of work trying to understand things better by looking at data from several years and improving our analysis in various ways. Eventually, though, Yang (the new student working on the problem) graduated, and we ended up with no writeup of anything. (I don’t blame Yang for that; if a writeup had been important enough to me, I could’ve worked on it.) At some point it seemed to make sense to post the correction, rather than waiting until the new analysis got done.
[…] is the data coding error. These authors actually got of lightly in that regard. I stumbled across this post, that captures an unequivocal retraction due to an Excel coding error (for bonus points, look […]