Search This Blog

Sunday, January 01, 2012

Friend or Foe

Retraction Watch estimates close to 400 papers were retracted in 2011. A couple comments came up with retraction numbers below 100 for the year:

Thanks so much for providing this valuable service. Of interest, a PubMed search for “Retracted Publication” or “Retraction of Publication” for 2011 yields only 35 papers. So if you and your colleagues weren’t on the job, we’d be missing out on ~1 in 12 of these papers!

Pat French

“Retraction notice” gives 63 hits. Still, the problem is exactly that it is not clear how to retrieve all the retractions, which may well be in the PubMed database. There should be a standard tag!

Jon Beckmann

We think that retractions are important pieces of scientific information. Unlike a Friday night bar conversation where you make the claim that the Buffalo Bills won the Superbowl back in 1995, the accuracy of claims made in science matters. When the folks at Nature faced the prevalence of chicanery taking place in big science labs and big science journals they concluded:
these days image detection software and the vigilance of media outlets such as Retraction Watch can catch irregularities—be they due to innocent error or misconduct—much sooner. The ability to track these changes provides benefits to biomedicine, as experiments in the scientific literature lay the foundation for future experiments.

The peer review process employed by science journals is the opposite of the scientific method. Peer reviewers don't repeat any experiments thus their judgments are not empirical. Peers have a bias that is not eliminated by the process. The authors and their organizations carry with them a cache that brings with it bias that is also not eliminated. The science isn't the only thing being evaluated. In a perfect world a journal would publish articles and offer only minimal interference into what the authors intend to say. Once published the true peers can evaluate the paper and agree or disagree with it. Scientists after all, are the peers.

The question that peer reviewed journals have to wrestle with is whether or not Retraction Watch is their friend or enemy. With over 1.5 million hits online Retraction Watch is clearly popular among scientists. Who else would read about retractions? It appears that the demand is there for more information. Transparency is our friend.

1 comment:

Yohanan Weininger said...

Re pubmed statistics – my advice as a retired librarian:

1) READ THE Pubmed Entrez HELP! Great manuals...

2) relevant specific search functions in Pubmed include [PT] publication type items “Retracted Publication”[pt] and “Retraction of Publication”[pt] See their definitions in[pt]
gets 3381 hits today and
gets 4077 hits..

3)>”Retraction of Publication as Topic”:
>Authors’ withdrawal or disavowal of their participation in performing research or writing the results of their study. Year introduced: 2008(1984)
On this subject there is a MeSH, a keyword, as indexed to pubmed records on the subject or books in the catalog etc… the Medical Subject Heading, MeSH database definition
Major MeSH search gets 321 hits from 1986 to Apr 2015..[MAJR]

4.) Note a free text search on title word “Retraction”[ti] retrieves more recent records as the MeSH keywords are not automatically indexed but entered by librarians after a week or several months after the record. (See the EDAT, CRDT and the MHDA date fields in the “Medline” (all fields) display format.)

Of course you can run the pubmed searches as automatic alerts to your email via your own MyNCBI (Pubmed’s free personal storage customization feature)

Extra for anyone uninitiated into the google web search help

BTW what is the CCS line on old school library reference desk librarian? (info support is just small part of edifice. It's how the knowledge users use information resources that matters.)