---------- Forwarded message ----------
Date: Tue, 14 Mar 2000 11:54:18 -0400
From: David Goodman <dgoodman@Princeton.EDU>
Subject: Re: Collection development policies for serials (Mary Ann Walker)

The only important factor is use, real or prospective.
Tyoically, a library buys the titles that receive or will receive the most use.
This can be measured by physical counts, and by citations.
Physical counts are not accurate exceopt in closed-shelf collections, but with
immense work can give useful data. An excellent example of how it should be
done is the recent work at the University of Wisconsin.
As for citations, I reprint below a discussion of this I posted license-l this
past August

Citation indexing is one of the key tools, and can give potentially two
types of data:

First, for or the journal as a whole, it can tell how much it is cited
(either total citations to all article in the journal, or citations per
article published, known as the impact factor), and by what journals it is
cited. This information is not exactly free, but is available as the CD or
Web product Journal Citation Reports. This however is information about
the content of the journal, not the content of the journal as it applies
to any one University's interests. There are , for example, many
theoretical ecology titles that are not that much used by the world in
general, but are certainly key journals to my library; there are many
hematology journals very highly cited by those in that subject, but only
slightly used or cited here.)

Second, Information about citations as it applies to any one institution
can be obtained from Science citation Index. Unfortunately it can not be
done with Web of Science, but only the versions on Dialog (& other such
services). Basically, you search for a title as citation source, and your
institution as citing location. This is quite inexpensive if one doesn't
need to analyze further. To see what articles are being cited or the
departments of those citing them is, because of the Dialog pricing
structure, very expensive (unless one does this, there's a ambiguity about
citing institution, as this is not a standardized field. ISI also can
supply all this for an institution as a customized service, for a fee.) I
have in fact done this for all Biology and Neuroscience journals that
Princeton subscribes to, and the Rutgers chemistry Librarian, Howard Dess,
published an article about this in "Science and Technology Libraries."
(Note that this is not the same as just seeing where a University's
faculty publish their papers, which is typically a much more restricted
group of titles.)

But this citation use is only part of the use: It only measures use that
results in references in formally published papers. It doesn't reflect
what is read as background but not cited. It doesn't reflect unpublished
theses. It doesn't reflect publication as technical reports, or as books.
It doesn't reflect student papers. It doesn't reflect class assignments.
It doesn't reflect what was found not useful, but which had to be examined
to make that determination. However, it is generally found to roughly
correlate with usage as measured by reshelving studies, except that at
least in biology review journals are typically even higher in the
reshelving use than the citation use (data for my library are in
preparation). One can certainly argue that in a scholarly institution
citation data offers the best single measurement. And in the absence of
data for a particular institution, the JCR data can be very useful indeed.
In both cases, they can of course be validly used only with awareness of
the limitations.

Prospective use can only be an estimate, based on similar titles.

All of the above assumes that all use and all citation is equal. In some
cases, that is not politically expedient, and it will be necessary to give
more weight to faculty use than undergraduate, or the opposite. In a
teaching-oriented institution a very good source for citations is term
papers. Ask faculty for copies of what they consider excellent ones.

I never trust faculty opinion. I tell the faculty "You completely and
totally determine what we buy. You do so by selecting what you work on and
what sources you use, and by what you tell your students to work on and
what sources they use. I will buy whatever you and your students actually
use and need." I don't actually say, but I also also mean: "I am in a
better position to judge this than you. It doesn't take subject knowledge;
it takes knowing, talking to, and observing the users."

Faculty, in my experience, usually tell librarians what titles they think
they and their student _ought_ to use, not they do use. Sometimes they
specify whatever they used when they were in graduate school many years
ago. If you follow their selections you will get a collection composed of
prestigeous but little used titles. If you can locate faculty whose advice
you can really trust, that's ideal; in my 20+ years of experience serving
a marvellously able group of talented researchers and teachers, I have had
the unusual good fortune to actually know one.

What journals to get in electronic versus print is a different question.
I would in most situations and for most titles get whatever is available
electronically as electronic-only. I will be giving a talk explaining this
at National Online.

--
Dr. David Goodman
Biology Librarian, and
Co-Chair, Electronic Journals Task Force
Princeton University Library
dgoodman@princeton.edu         http://www.princeton.edu/~biolib/
phone: 609-258-3235            fax: 609-258-2627