Tuesday, October 02, 2012

Misconduct Accounts for Two out Three Retractions

The big news of the day is the PNAS paper -- Misconduct accounts for the majority of retracted scientific publications -- by Ferric Fang, Arturo Casadevall and Grant Steen who looked into the reasons behind retracted papers in the PubMed database -- all 2047 of them. There is quite a bit of buzz -- see this post at Retraction Watch for some commentary and links to press coverage.

G. Mudur's news story in today's The Telegraph covers the Indian angle and presents some additional details which are only implicit in the paper. A couple of quick comments:

  1. PubMed has less than 200,000 papers by Indian researchers out of over 25 million entries. It's safe to say India accounts for less than 1 % of the papers in PubMed.

    But, India accounts for 3.4 % of fraudulent papers, 10 % of plagiarized papers, and 9 % of duplicate papers. [Look at the graphic in Mudur's story.]

  2. My own study last year had flagged just one Indian paper for fraud (specifically, falsification); but Fang et al appear to have flagged more than 25 papers for fraud.

    My guess is that this difference is probably due to papers that could have been flagged either way. For example, a bunch of PubMed papers (in addition to several tens more that are not in PubMed) from Prof. P. Chiranjeevi's group at Sri Venkateswara University at Tirupati could easily have been flagged for fraud because (a) it was a part of a massive, deliberate scam, and (b) the group also changed names of some of the chemical entities in their plagiarized papers.

    [I flagged them for plagiarism, primarily because that's what those papers really are; moreover, the retraction notices are not entirely clear about the nature of the offence].