The Open Access Citation Advantage

[Stevan Harnad]

A CSIR-NISCAIR (India) study
<http://www.currentscience.ac.in/Volumes/107/05/0733.pdf> by Prathap (2014)
compared OA and non-OA by comparing journal impact factors for 13 Indian
journals betore and after they converted to Gold OA (2004-2014) and found
no difference.

Comparing journal impact factors before and after conversion to OA are not
the best way to test the OA citation advantage, which should be tested at
the individual article level, within journals, for OA and non-OA articles
published at the same time. Many things change across time.

Prathap cites the work of Davis
<http://scholarlykitchen.sspnet.org/2014/08/05/is-open-access-a-cause-or-an-effect/>
who
in turn cites the RIN study
<http://www.nature.com/press_releases/ncomms-report2014.pdf> and the Nature
Blog article
<http://blogs.nature.com/ofschemesandmemes/2014/07/30/investigating-the-open-access-citation-advantage>
about
the OA citation (and download) advantage.

The RIN within-journal study found a small citation advantage for paid
hybrid Gold OA *Nature Communications* articles compared to non-OA Nature
Communications articles 2010-2014.

Phil Davis cited an error in the RIN study -- OA and non-OA may have been
mixed up — and re-iterated his preferred hypothesis that the OA citation
advantage is a self-selection artifact (but also suggested that because of
the RIN error, the actual OA citation advantage, though a self-selection
artifact, might be much bigger!).

I have not since heard an official confirmation or correction of any error
by RIN or Nature.

As to the self-selection hypothesis, I tried to post a commentary on the
Nature Blog where the RIN study was discussed by Ellen Collins of RIN
<http://blogs.nature.com/ofschemesandmemes/2014/07/30/investigating-the-open-access-citation-advantage>
(but
I’m not sure whether it will appear: the software never acknowledged
getting it, and I posted twice):

The RIN study did not compare non-OA vs. OA in general but non-OA versus
hybrid Gold OA in particular (hence a self-selected decision as to whether
or not to pay for hybrid Gold OA).

I pointed out that our 2010 within-journal OA/non-OA  comparison (Gargouri
et al <http://dx.plos.org/10.1371/journal.pone.0013636>) had shown — for
individual articles, with a far larger sample, across all journals and
disciplines, and for unpaid Green OA self-archiving rather than for hybrid
Gold OA payment — that there is an OA citation advantage of the same size
regardless of whether the (Green) OA is self-selected or mandatory.

Most of the OA advantage studies either compare on too short a time period
(too early for the OA advantage to be detectable) or they average across
too long a time period in which the *timing of the OA itself *is unknown.
The timing of OA is important, however, because some publishers embargo OA
for a year, and late access may never make up the lost access
<http://openaccess.eprints.org/uploads/citegrowarxiv.jpg>.

Future repository-based studies, where the date on which the paper is made
OA can be determined, relative to the paper’s acceptance data and
publication date), will allow OA timing to be controlled better.

Timing (as David Hume <https://en.wikipedia.org/wiki/Causality> would
agree) is crucial to inferring causality (even if it cannot “prove” it).

Meanwhile, I’d say that it still remains most likely that there is indeed a
significant OA citation advantage, and that the advantage is causal, not
just an artifactual side-effect of author self-selection. Ditto for the
even bigger OA download advantage (and the two effects are unlikely to be
unconnected!).

Stevan Harnad
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.asis.org/pipermail/sigmetrics/attachments/20140914/44ec695e/attachment.html>