Tuesday
Mar172009
by 
Bishop Hill
Bishop Hill Whitewash in Albany
Mar 17, 2009 Climate An interesting post for climate watchers over at Freeborn John. It's the murky tale of how dodgy data finds its way into the surface temperature records and the duplicity of the University of Albany, New York in trying to whitewash the alleged fraud.
Word "alleged" added, 18/3 following a comment.
Reader Comments (17)
While I would agree that there seems to be evidence of an error (and an important error), there are several ways to explain errors. I have not seen the direct information (e.g. emails between Wang and Zeng) which would allow me to conclude that any individual knowingly fabricated a claim.
In fairness to UoA, there must be limits to what it can reveal to the public. The allegation is fraud, and if that cannot be proved beyond reasonable doubt (or the equivalent standard), then the allegation must be dismissed. Scientists have the presumption of innocence, too.
yours
per
The only way I can see of explaining it as an error would be if there was an entire second set of documentation giving entirely different station histories, that has since been lost, and which co-workers apparently had no knowledge of since they have subsequently contradicted it.
But it is true that all are innocent until proven guilty, so we'll wait and see.
Wang is now at UoA, while Zeng is in China. This may be one of these collaborations where communications have been difficult; collaborators from non-english speaking countries frequently misunderstand, or misuse, written and verbal english. Bad communications between collaborators would be enough to explain an error, without any fraud whatsoever.
It seems to me that it is extremely easy for Wang to create reasonable doubt that he acted fraudulently. By contrast, finding evidence to show that he did anything fraudulent would almost certainly require email and/ or mail from 1989 or before.
In fact, Wang has a contemporaneous note (via an email of 2007) that Zeng has assured him that the original statement in the paper is correct (according to Informath's notes).
this is great theatre, but it is a really serious claim that someone has committed fraud. Making that claim speculatively will not engender respect.
per
The claim was made that the stations had not moved, and yet that was either not known or the records showed that they had. Of what worth are such claims if you simply have to set up a suitable chain of plausible deniability to be able to say any the hell thing you want? The moon is made of green cheese, and I have a contemporaneous note from a NASA astronaut to say so.
If you put your name to a paper, you vouch for its claims. Otherwise the fraud is in claiming somebody else's results as your own.
"If you put your name to a paper, you vouch for its claims."
quite, but not realistic. The statistician (whose name may be on a paper) does not understand the bench work, and the bench workers almost certainly don't understand the statistics; utterly common. Collaborators coming together, perhaps from different labs; who is going to go back and check the lab notebooks ? Clinicians in hospitals, where a medic does the case notes by themselves; who can check ? In practice, almost all papers depend on taking someone else's word for something they did.
I agree that it's usual for collaborators to each know their own area, and not know much about each other's. I would say, though, that if the bench work constitutes a significant part of the result, the bench worker gets their name on the paper, and takes both responsibility and credit. If a lead author is going to take all the credit for the work of others, then they take all the blame for their mistakes and frauds too. It's their job to check. If they think the odds are good enough and choose not to, they're taking the risk on themselves.
If you put your name to a paper, you vouch for its claims. If you choose to make a claim regarding evidence you don't have and have never seen, it's fabrication. Even if somebody else told you it was so, and even if it was reasonable to believe them. You're the one claiming the credit for the claim, so you're the one responsible for it.
To say that every scientist does it in practice is simply to say that every scientists commits fraud. I don't believe that to be so. In any case, had Wang been open with the data and simply admitted the error, apologised, and withdrawn the paper when it was pointed out, I wouldn't have a problem with it. It's not the crime, its the cover-up that gets you.
as regards a cover-up, according to Keenan's report, Wang corresponded with him in 2007, saying that he had checked with a co-author, and the co-author had confirmed that the statement in the paper was correct. So according to due process, Wang has checked it out.
I think there is an important issue about getting the facts of the matter correct. If it can be shown that the original comment is wrong, then errata can be published, etc. This is a way forward, but requires cooperation from the people involved. Once you have undertaken the nuclear option of accusing people of fraud, it strikes me that even the most accommodating individual is going to be unlikely to be helpful, and many scientists would be positively obstructive from there on in.
It is probably worth recording that most of this debate is not in the formal peer-reviewed literature; and that is where reality is established. It would be useful to see clarification of the rights and wrongs of this in the peer-reviewed literature.
per
You're joking, surely! You're trying to wind me up! "reality is established"? I'm sorry, I'm sure you don't mean it that way, but that just cracks me up!
There is a simple answer then. All Keenan has to do is produce an email from somebody in which he was earnestly assured that it was indeed fraud, say later that he chose to believe it and you can't check everything, and get the resulting accusation past peer review. Oh, but hang on a sec. The accusation has indeed been published in the peer-reviewed journal Energy and Environment. QED. Due process has been followed, and the Reality of Fraud is established!
I'm glad that's all settled then.
Seriously, and all joking aside, somebody somewhere made a claim that they must have known wasn't true or supported by any evidence. So would you at least agree that fraud occurred, even if you disagree about who can be held responsible for it? And if so, how would you recommend such fraud be dealt with, to ensure accountability, if nobody can be held responsible for the contents of a published paper?
You seem to be very clear that someone has committed research misconduct (at least !), even though I have pointed out a few of the many possible alternative explanations. You have a view of the "responsibility" associated with publishing a paper which is at odds with US and UK regulators. This is all jolly good stuff, but making serious charges on the basis of layer upon layer of speculation is not that impressive.
For your information:
http://ori.dhhs.gov/
http://www.ukrio.org.uk/home/
I think you go too far in your support for the peer review literature. Peer review is a relatively recent thing Einstein and Watson & Crick were not peer reviewed) and by most accounts is thoroughly unreliable anyway. The truth will be reached by replication and debate and transparency.
That's actually even worse. To rely on peer review is a case of Appeal to Authority, and antithetical to science. To restrict this to a special club of peer reviewed journals run by a particular organisation who act as gatekeeper is even more clearly an Appeal to Authority. At least when you cite peer review, you're talking about a general process, not a specific organisation.
Actually, it would be pretty funny if this really was the standard, because the IPCC appears not to be listed either. (I couldn't find them, anyway.) The idea that the IPCC would be 'forced' to take notice is quite funny as well - you will remember all the manipulations that went on around "Caspar and the Jesus Paper" necessary to get the required 'rebuttals' into the report. Of *course* they're going to lionise Jones again.
The thing I think you didn't make clear is whether you are only talking about peer review being necessary to enter the *IPCC's* reality, and thus have an effect on the policy-influencing mainstream, or whether you really meant it.
Being snotty about the peer-reviewed literature is all very well, but not entirely realistic. Peer-review is not perfect, but it is the best system we have (that I am aware of) for such a complex problem. Replication and debate serve different purposes to the peer-review/ literature process.
toodle-pip !
per
The problem the journals aim to address is to help scientists hear about new and interesting developments in their field. Peer review forms a first-line filter to ensure that what is published is interesting, novel, looks like competent science, provides enough information for others to follow it up, and doesn't have any glaring gaps or obvious mistakes in its arguments. The aim is to provide papers that the journal's paying customers are happy to read, and filter out those that make them want to drop it in disgust. It is an editorial, and in a quite fundamental way, a *commercial* function, much akin to the way a newspaper vets its stories.
It is NOT a check on the accuracy of validity of the science, except in the most shallow sense. While they vary widely in their degrees of diligence, peer reviewers will not generally replicate the results, chase up or necessarily even *look at* the data, re-do lengthy calculations, or check every step of the arguments. All of that comes later. All it does is see if it passes the 'smell test', and while some papers do require a lot of specialised knowledge, in many cases any competent scientist can manage that. It does not check for fraud.
The accuracy and validity of the science is checked by means of the *scientific method*. To the extent that scientific matters are ever "decided", it is by a theory reliably withstanding all attempts to falsify it unscathed, by scientists honest in their intention to do so and open and willing to accept such falsification, with data that could reasonably be expected to falsify it if it was in fact false, over such a period and with such an effort made that we can conclude that it would take an extraordinary new body of scientific evidence to overturn it.
Passing peer review just means the journal thinks it's worth somebody's while to start to check it. If anyone starts telling you that it has to have passed peer review to be credible, or worse, that if it *has* passed peer review it *is* credible, that should be an instant red flag to warn you of someone who doesn't understand the scientific method, or the real purpose of journals. Or who does, but is trying to deceive you.
It would take a determined attack by sceptics with the *full cooperation* and assistance of the climate science community over several years, until the sceptics gave up defeated, to verify the science. That's the way it is supposed to work. And if somebody phones up asking for details and data, *volunteering* to check your work for free, you should at least pretend to be delighted, because it means your work is on the fast track to becoming scientifically accepted.
In this case the author didn't, and it took an FOIA request on a colleague to get even minimal cooperation. Given the results, I think we can guess why. We will, of course, have to wait for the the results of a proper enquiry. But even if Wang is cleared, this is not the way science is supposed to be.
I don't really see peer-review as editorial or commercial functons; they are certainly part of an editorial or commercial process, but their function is to provide that first-line scientific filter.
"it has to have passed peer review to be credible, or worse, that if it *has* passed peer review it *is* credible"
to a first approximation, and with exceptions, I would support these statements. It is, as you say, a first-line filter. As to what "the real purpose of journals" is, I suspect that must remain unknown :-)
In practice, very few people are happy to have their work audited, and there probably weren't even applicable standards in the 1990s. It is unreasonable to ask for auditable material to be retained from way back then for a scientific institution.
A lot of this stems from your perspective. I believe that there is at least one nobel laureate who argues that you should be able to publish more-or-less untrammeled, since your publications will reflect only on yourself. A lot of his work has been replicated, so he may not see replication as a big issue.
per
Wang's defense during the university's investigation was leaked (as part of Climategate). The defense effectively claims that Wang got all the data from Zeng. I have posted some comments about this at
http://www.informath.org/apprise/a5620/b19.htm
My belief is that none of this makes the university's actions, discussed in the original posting and further on my site, any less discreditable.