How to compare research impact of toll- vs. open-access research

[Stevan Harnad]

The following data posted by Peter Suber in
http://www.earlham.edu/~peters/fos/fosblog.html
indicate that open-access articles (from BioMedCentral) average at least
89 times as many downloads as toll-access articles (from Elsevier). (The
89 is probably an undercount, because it does not include PubMedCentral
downloads.)

    PETER SUBER:
    "Elsevier has put some PowerPoint slides on the web summarizing
    its interim results for 2003. Slide #16 shows that there were 4.5
    million full-text articles in ScienceDirect on June 30, 2003, and
    slide #15 shows that there were 124 million article downloads in
    the 12 months preceding that date. This means that its articles
    were downloaded an average of 28 times each during the past year.
    http://www.investis.com/reedelsevierplc/data/interims2003b.ppt

    "For comparison I asked Jan Velterop of BioMed Central what the
    download figure was for BMC articles during the same time period. He
    reports that the average is about 2500 per year, which doesn't
    count downloads of the same articles from PubMed Central. This is
    89 times the Elsevier number. "

Combine these download data with the citebase data on the correlation
between downloads and citations
    http://citebase.eprints.org/analysis/correlation.php
and you will be able to estimate the dramatic way in which open access
enhances research citation impact, confirming what Steve Lawrence reported
in 2001 for computer science research:
    http://www.nature.com/nature/debates/e-access/Articles/lawrence.html
and what Kurtz et al. reported for astrophysical research:
    http://www.ecs.soton.ac.uk/~harnad/Hypermail/Amsci/2829.html

(In an ongoing collaboration with Charles Oppenheim we are currently
making controlled pairwise comparisons of citation impact between
open-access and toll-access articles that appear in the same journal and
year, comparing self-archived and non-self-archived articles, across time,
and across disciplines. We hope to extend these comparisons with the
help of ISI's citation database.)

Those individuals, institutions, research-funders, tax-payers and nations
who are interested in increasing the visibility, usage and impact of
their research output should take special note of these data! Apply the
estimates in reverse if you wish to estimate the amount of research impact
(and its rewards) being that is currently being *lost* daily, monthly,
and yearly by researchers, their institutions, and by research itself as
long as we delay providing immediate open access to all research output --
as we could already do today, by self-archiving it.
http://www.ecs.soton.ac.uk/~harnad/Temp/che.htm

Stevan Harnad