There is a bit a scandal on in Germany – a leading business academic has been accused of misconduct and a leading journal ‘Research Policy’ has retracted some articles.
Ulrich Lichtenthaler, who is affiliated with the University of Mannheim, has come under suspicion of inflating his publication record using unethical methods.
Additionally, a number of his published papers apparently contain severe mathematical errors and methodological inconsistencies.
Okay – so the first allegation is self-plagiarism. That is naughty and can be hard for referees to detect. I have less sympathy for the other allegations.
… there seems to be an omitted variable bias problem that would invalidate the conclusions of the Research Policy 2010 paper. In both cases, this raises severe doubts as to the validity and robustness of the conclusions drawn in the two Research Policy papers (and indeed in the other parallel papers). If the referees and editors involved in handling the two Research Policy papers had been aware of this (i.e., if their attention had been drawn to the other closely related papers and they had spotted this inconsistency), they would undoubtedly have rejected each of the Research Policy papers on methodological grounds.
Omitted variable bias is always a problem and referees should think about it. To be fair they didn’t know about the other papers, yet referees are supposed to be experts in the field and perhaps should have known about them.
Then there is this:
… in some cases where the coefficients and standard errors are about the same size, variables are reported as highly significant, This problem is more evident for independent variables than control variables.
No excuses – referees should have spotted that. Unless the author actually fabricated the standard errors in the paper, the referees should have picked up on this.
Now I don’t want to detract from the seriousness of the allegations against the academic, but it does highlight the fallible nature of published research and the peer review process.
(HT: Mike)
Quite so, Sinc.
E only reason this even matters is because of government policy. No amount of publishing nonsense effects actual productive enterprise, as it will depend on repeated and verified science.
I would have thought the Lancet’s autism and Iraq death toll* enormities settled for all time the proposition that peer review is, substantially, a meaningless racket.
* The journal argued that 897 billion Iraqis were killed by Bush.
Its the problem in climate modelling. If the GCM modellers included the significant variables for solar magnetics (eg pSCL) and ocean cycles their calculated numbers for sensitivity would very likely come out at about the instrument values (0.6-0.7 C/doubling of pCO2).
Unfortunately if they did this their next years’ long term climate modelling budget would be statistically similar to zero. A lot of money can ride on those accidentally omitted variables.
Academics, harried by the over weaning demands of students and bureaucrats, sometimes cannot give papers they are reviewing the attention they need and errors can slip through.
Peer review is like democracy – it is very flawed except that all other forms of review are worse.
As an editor I am struggling to get papers thoroughly reviewed. My solution is to invite a number of reviewers, usually people well known to me, in the hope that at least 1-2 will be thorough. If not, I undertake the review myself. I don’t usually check the math but I check for ‘secondary’ clues as to the validity of an argument.
It is never possible for a reviewer to assertain that a paper is free of serious errors. One can only think of experimental papers, where so many things can go wrong…
An alert academic who was refereeing a paper submitted by Lichtenthaler had a hunch that he had seen the same paper by the same author before but in a different journal. A cross reference to that work was missing, though.
Publishing similar or closely related paper twice without cross referencing is considered unethical in academia, partly because it unnecessarily consumes the unpaid work of other researchers who act as referees for the journal.
Called the gravy train where I come from in the remote areas of the outback where social studies predominate. Studies favoured ‘anthropology’, archaeology was ‘banned’ in the late 70s, try find a book on this science during this era to 2010. Narrative, narrative and self-selected respondent surveys is all you will find.
Change the methods (sampling, questionnaire, observation, participant/non-partic, focus group, narrative, interpretative analysis etc etc) to suit the [bias in] methodology to FIT the theoretical perspective (feminism, post-positivism, interpretivsm etc etc) and one has a career for life, publishing crap.
heh presto, a neat fit with constructionism or subjectivism.
Aaron, that is why a red pen or the big X was developed. Science and the practice of, demands such.
But red pens are not so abundant in recent times, there aren’t too many in academia that practice or teach science these days.
Reviewing is a thankless task and it is not surprising that it is often done badly. Of course there are two types of errors in reviewing (Type A and Type B), one when it is too slack and lets rubbish get into print and the other when the reviewer decides to be a smart-arse and block my papers.