RE: On metrics

["Lisa Dittrich" <lisa@jbmr.org>]

I think defining terms here would help.  When I tracked usage at 
my old job, we differentiated "hits"--page views--from downloads, 
where someone actively downloaded a PDF of an article.  Not 
surprising, we found downloads a more significant measure of 
interest.  Whether or not the person who downloaded an article 
subsequently read it was, of course, impossible to know.  But at 
least they had enough interest to do the downloading.

Lisa Dittrich
Director of Publications
ASBMR Publications
www.jbmronline.org

-----Original Message-----
[mailto:owner-liblicense-l@lists.yale.edu] On Behalf Of Elizabeth R Lorbeer
Sent: Wednesday, October 10, 2007 7:01 PM
To: liblicense-l@lists.yale.edu
Subject: RE: On metrics

The number of downloads is a trivial metric since all it reveals 
is that an attempt was made to access an article at the journal's 
web site. More useful metrics include machine specific IP of the 
requestor, volume specific usage, and average length of time 
viewing an article.

Liz Lorbeer
University of Alabama at Birmingham
lorbeer@uab.edu

-----Original Message-----
[mailto:owner-liblicense-l@lists.yale.edu] On Behalf Of Joseph J.
Esposito
Sent: Tuesday, October 09, 2007 5:16 PM
To: liblicense-l@lists.yale.edu
Subject: On metrics

Members of this list may be interested in a recent announcement
by Berkeley Electronic Press, which can be found here:

http://www.bepress.com/download_counts.html

BEPress did a careful study of the downloads of its articles and
discovered that usage figures were highly inflated by Internet
robots that come to a site for any number of reasons, from
search-engine indexing to laying the groundwork for spam attacks.
BEPress has "come clean," but their example poses the very
important question of how reliable some of the usage reports from
other publishers are.  (Although I have consulted with BEPress, I
had no involvement with this project.)

I am no expert on the metrics for journals; someone who is may
want to take a look at this.  My hypothesis is that many
publishers are reporting inflated figures--and that acquisition
librarians may be making purchasing decisions in part based on
faulty data.  Of course, downloads are far from the only metric.

One tidbit that emerged from the BEPress data is that open access
publications were more susceptible to inflated download counts
than toll-access articles.  Thus this analysis whittles away a
bit at the alleged "open access advantage."

I would be interested to know whether this kind of situation has
been previously identified and whether the current measurements
for downloads take this matter into account.

Joe Esposito