Re: Does the arXiv lead to higher citations and reduced publisher downloads?

["Peter Banks" <pbanks@diabetes.org>]

Although I respect Kristin's work and suspect that there is a 
small OA citation advantage, I am not convinced by these data.

For one thing, I doubt that most of the results reach statistical 
significance. For the fields other than mathematics, there are so 
few OA papers with 1 or more citations that the margin of error 
is likely be greater than any apparent difference between OA and 
non-OA papers. Even the trends aren't clear; looking at the data 
for papers with one citation, there seems no difference for 
engineering, an OA advantage for philosophy, and a non-OA 
advantage for political science. I think we're trying to find 
significant differences in what is probably noise.

I also don't understand how these data exclude Phil's hypothesis. 
Since Kristin seems to define quality in terms of citations, then 
the logic seems self-referential: how would one detect a 
difference in citation due to instrinsic quality when one has 
defined quality as number of citations?

Peter Banks
Publisher
American Diabetes Association
Email: pbanks@diabetes.org

>>> kristin_antelman@ncsu.edu 03/20/06 5:54 PM >>>
Phil Davis wrote:

> Based on our analysis, we found that a quality differential is
> a more plausible explanation -- the reason why arXiv-deposited
> articles receive more citations is simply because they are
> better articles, not because of some advantage conveyed through
> increased access.  If Open Access can explain the citation
> advantage (and we did confirm one), it is only responsible for
> giving an advantage to already highly-cited articles.

Data I collected for philosophy, political science, engineering 
and mathematics do not support this hypothesis that OA causes 
more citations for better articles only (given that one uses 
overall citations as a rough measure of quality).

These data were collected for my article, "Do Open Access 
Articles Have a Greater Research Impact?" (C&RL Sept 2004, 
http://eprints.rclis.org/archive/00002309/), but at that time I 
had not looked at the distribution of OA and non-OA articles by 
citations. Graphs of those results are posted at 
http://www.lib.ncsu.edu/staff/kantelman/OA_by_citations.xls.

These data show OA citation advantage across all articles with 
more than zero citations.  It could be argued that OA helps to 
get the first citation.  It's also striking, I think, how similar 
the graphs are even though the rates of OA vary greatly between 
these disciplines (between 17% and 69%).

________________________________________
Kristin Antelman
Associate Director for the Digital Library
NCSU Libraries
Box 7111
Raleigh, NC 27696-7111
(919) 515-7188 Fax (919) 515-3628