Craig et al.'s review of the OA citation advantage

[Stevan Harnad]

On Sat, 26 May 2007 bernd-christoph.kaemper at ub.uni-stuttgart.de wrote:

> Could we do a thought experiment?
>  From a representative group of authors, choose a sample of authors
> randomly and induce them to make their next article open access.
> Do you believe they will see as much gain in citations compared to
> their previous average citation levels as predicted from the various
> current "OA advantage" studies where several confounding factors
> are operating? Probably not - but what would remain of that
> advantage? -- I find that difficult to predict or model.

 From a random sample, I would expect an increase of around 50% or more
 in total citations, 90% of the increased citations going to the top 10%,
 as always.

> As I learned from your posting, you seem to predict that it will
> anyway depend on the previous citedness of the members of that
> group (if we take that as a proxy for the unknown actual intrinsic
> citation value of those articles), in the sense that more-cited authors
> will see a larger percentage increase effect.

I don't think it's just a Matthew Effect; I think the highest quality
papers get the most citations (90%), and the highest quality papers are
apparently about 10% (in science, according to Seglen). 

> To turn your argument around, most authors happily going open
> access in expectation of increased citation might be disappointed
> because the 50% increase will only apply to a small minority of
> them. 

That's true; but you could say the same for most authors going into
research at all. There is no guarantee that they will produce the highest
quality research, but I assume that researchers do what they do in the
hope that they will, if not this time, then the next time, produce the
highest quality research.

> That was the reason why I said that (as an individual author)
> I would rather not believe in any "promised" values for the possible
> gain. 

Where there is life, and effort, there is hope. I think every
researchers should do research, and publish, and self-archive, with
the ambition of doing the best quality work, and having it rewarded with
valuable findings, which will be used and cited.

My "promise", by the way, was never that each individual author would
get 50% more citations. (That would actually have been absurd, since
over 50% of papers get no citations at all -- apart from self-citation
-- and 50% of 0 is still 0.)

My promise, in calculating the impact gain/loss that you doubted,
was to countries, research funders and institutions. On the assumption
that the research output of each roughly cover the quality spectrum,
they can expect their total citations to increase by 50% or more with
OA, but that increase will be mostly at their high-quality end. (And the
total increase is actually about 85% of 50%, as the baseline spontaneous
self-archiving rate is about 15%.)

Stevan Harnad

> That doesn't mean though that there are not enough other
> reasons to go for open access (I mentioned many of them in my
> posting).

There are other reasons, but researchers' main motivation for conducting
and publishing research is in order to make a contribution to knowledge
that will be found useful by, and used by, other researchers. There are
pedagogic goals too, but I think they are secondary, and I certainly
don't think they are strong enough to induce a researchers to make his
publications OA, if the primary reason was not enough to induce them.

(Actually, I don't think any of the reasons are enough to induce enough
researchers to provide OA, and that's why Green OA mandates are needed
-- and being provided -- from researchers' institutions and funders.)

> With respect to the toll accessibility index, I completely agree.
> The occasional good article in an otherwise "obscure" journal
> probably has a lot to gain from open access, as many people
> would not bother to try to get hold of a copy should they find it
> among a lot of others in a bibliographic database search, if it
> doesn't look from the beginning like a "perfect match" of what
> they are looking for.

You agree with the toll-accessibility argument prematurely: There are as
yet no data on it, whereas there are plenty of data on the correlation
between OA and impact.

> An interesting question to look at would also be the effect of
> open access on non-formal citation modes like web linking,
> especially social bookmarking. Clearly NPG is interested in
> Connotea also as a means to enhance the visibility of articles
> in their own toll access articles. Has anyone already tried such
> investigations?

Although I cannot say how much it is due to other kinds of links or from
citation links themselves, the University of Southampton, the first
institution with a (departmental) Green OA self-archiving mandate, and
also the one with the longest-standing mandate also has a surprisingly
high webmetric, university-metric and G-factor rank:

    http://openaccess.eprints.org/index.php?/archives/79-guid.html

Stevan Harnad