OA advantage = EA + AA + QB + OA + UA

[Stevan Harnad]

On Sat, 4 Sep 2004, jcg wrote:

> Dear Stevan,
>
> In your presentation of the Kurtz study that is available at:
> http://listserver.sigmaxi.org/sc/wa.exe?A2=ind04&L=american-scientist-open-access-forum&O=A&F=l&P=44671
>
> one finds both round and square brackets. I take it that the square brackets
> are your interventions or clarifications while the round brackets are just
> brackets used by the author.

Dear Jean-Claude,

That's correct.

> Here is the passage that interests me:
>
> For me the bottom line result of my little experiment is that even for
> astronomy, with its near ideal [i.e., small, universally affordable]
> toll-based system, fully 20% of all potential article reads are thwarted
> by the access controls (and lack of electronic versions). (The fact that
> many of the inaccessable papers are in the ArXiv probably does not change
> this much, as the additional effort involved [from leaving ADS's unified
> resource to go to another system] is a great deterrent.)

It's a little confusing as written, but I think what Michael means
here is that the special situation of astro -- with just about 100% of
astrophysicists worldwide having just about 100% institution-licenced
(toll) access to the full, small, closed circle of astro journals --
there is still an impact advantage for those articles that are also
in Arxiv, because authors prefer to search Arxiv, seamlessly, with
all the preprints and postprints together there, rather than to search
across various licensed sites (and in some cases there is no online
version, just paper!).

I expect that the astro publishers' licensed sites will remedy this to
a certain extent soon (all online, seamless navigation), but they cannot
remedy the fact that they don't have the PREprints!

And so the fine-structure of the OA citation-impact advantage now appears to
consist of the following additive components:

(EA): Early advantage, a permanent citation increment that comes from
making the paper accessible already at the preprint stage. (This is not just an
earlier reaching of the eventual peak, but a true increment.)

(AA): Arxiv advantage: Even in astro, where the postprints are all 100% OA through
universal licenses, and apart from the preprint EA effect, there is an additional
citation advantage for those articles that are not just OA through the licensed
sources, but also through Arxiv. (This effect will also occur, I expect, for the
entire network of distribute OAI institutional archives, searched via OAIster or
citeseer, once the content there also approaches 100%. So the first "A" can
also be read as (Self-)Archiving Advantage.)

(QB): Quality bias: There is a tendency in astro (and probably in other fields as
well) for the self-archiving authors to be the authors of the higher quality
articles. So this is a component of the OA citation advantage of which the cause
is not OA but self-selection.

(OA): In addition to these 3 factors there is the true OA citation
advantage (absent in astro, because it is already 100% OA), which is the
higher citations that an OA article has over a non-OA article published in
the same journal and year, over and above its EA, AA and QB advantage. My
prediction is that this 4th component of the overall citation advantage
for Open Access articles will be the biggest component in fields where the
%OA is low, and will shrink, relatively, as the proportion OA approached
100%. (In astro, there is 100% OA, so this third component is zero.) This
means that until we reach 100% OA, the biggest benefits of OA are in
fact relative, *competitive* advantages; hence this is another strong
incentive for reaching 100% OA as soon one as we can -- and it means
that authors who provide OA before their field has reached 100% OA will
enjoy an extra advantage until it does.

But even at 100% OA (at which level, according to Michael Kurtz, in astro, the
total number of citations is not scaled up higher than with non-OA), there
will not only still be the EA (preprint) advantage but also:

(UA): Usage advantage. With 100% OA, articles are *used* three times as much
as with non-OA (even though they are not cited more). The QA vanishes, of course
(as all articles are OA) and so will the AA (as access will be integrated and
seamless). But there will still be a "selectivity" effect, in which authors will
be able to select in a far better-informed way what they do and do not choose to
cite, with the toll-affordability bias that there was in non-OA days now
completely eliminated. And variance in the size of the UA will be predictive,
because UA is an early-performance predictor of later citation impact:
http://bmj.bmjjournals.com/cgi/content/full/329/7465/546
http://citebase.eprints.org/analysis/correlation.php
http://www.ecs.soton.ac.uk/~harnad/Temp/timcorr.doc

> PS Some people tell me privately that you now object to shades of green in the
> Romeo color coding scheme. Is this really true?

No, not at all. But it depends which Romeo colour coding scheme you mean!
The one in romeo.eprints.org has only two colours, green and gray. Gray means
no green light for either preprint or preprint yet. Pale-green means preprint
only. Full-green means postprint. I of course completely agree with that code,
because I designed it!
http://romeo.eprints.org/stats.php

What I object too is the absurd spectrum of unnecessary and confusing
colours that are still being used by SHERPA Romeo: white, yellow, blue,
green (and even red and gray!) to mark distinctions that are completely
irrelevant to OA and only of interest to acquisitions librarians and
IP officers.
http://www.sherpa.ac.uk/romeoinfo.html
http://www.sherpa.ac.uk/romeo.php?all=yes

Best wishes,

Stevan Harnad