Re: Citation indexing and Re: Ejournal use data, was: Elsevier and=cancellations

["Gerard Kohler" <kohler@fee.uva.nl>]

=09charset=3D"iso-8859-1"
Content-Transfer-Encoding: 8bit
X-edited-by: aokerson@pantheon.yale.edu
Date: Mon, 23 Aug 1999 12:09:36 EDT
Reply-To: liblicense-l@lists.yale.edu
Sender: owner-liblicense-l@lists.yale.edu
X-Listprocessor-Version: 8.2 -- ListProc(tm) by CREN

> On Aug 20, 11:30pm, Hunger99@aol.com wrote:
> > Didn't Garfield pioneer this area by looking at citation indexing?
> > Doesn't Science Citation Indexing still put out numbers on what journal=
s
> > and what articles have the most impact in a field?  Are these not bette=
r,
> > more direct, measures than how often a journal is accessed?  It would
> > appear that accessing may be related to the impact of a journal or
> > article, but what we would like to know for understanding progress and
> > development in a field is impact, or some such measure. Aren't citation
> > measures available to librarians for free from SCI?  And SSCi? dh
> >-- End of excerpt from Hunger99@aol.com
>
Subbiah Arunachalam responded

> Citation measures based on Journal Citation Reports data are produced and
> analysed regularly by ISI, Philadelphia. The data, now available on CD an=
d
> on the Web, are also used by many others - such as CHI, the Leiden
> group, ISSI in Hungary. But one has to pay for the data. It is not free.

When using stuff like JCR, ssci and sci for "cleaning up" serials
collections bear in mind that these bibliometric indicators can serve very
well to provide an indication, or summary statistic, of the overall
quantity and quality of scientific output but

[As in its JCR (journal citation reports), the ISI writes:]

"While the JCR is an essential guide to scientific and technical journals,
it should not be relied on as the sole source of information when
comparing and evaluating publications. (.) Users should be aware of
general citation patterns when applying the information in the JCR. For
example, citation frequency, impact, immediacy, and other indicators may
vary widely between different research specialties. In some fields,
five-year impact factors may be more appropriate than the two-year impact
data presented in the JCR. Also, review articles (and review journals)
tend to be cited more frequently than other types of research
communications. On the other hand, letters and other shorter
communications may be cited less."

In my practice, using ISI products in combination with measuring daily use
of serials, some drawbacks of a sole ISI-method are clear:

Not all journals are covered by ISI, hence, not for all journals, citation
data are available. ISI claims to cover the most important journals, but
ISI is not very open concerning the criteria that it uses for
including/excluding journals. ISI covers some 10% of more than 100.000
journals in total.

The highest scoring journals in science and medicine have impact factors
above 20, much higher than the journals in social sciences or humanities.
And in my practice (covering economics as well as econometrics), only
using impactfactors will kill many marketing and business serials but only
using the measurement data will kill almost all econometrics serials.

Furthermore impact factors might be biased in favour of theoretical work
The idea here is that theory is more generally applicable, hence,
theoretical work (at least if it is good) will be cited more frequently
than applied work, which, after all, is limited to a specific context.

The impact factor is a ratio of two numbers, number of citations divided
by the number of citable documents in the journal. The question is what
counts as a citable document and what not. Some journals have different
types of articles, such as regular articles, notes, comments, letters,
etc. In these cases it might happen that the numerator includes all
citations while the denominator includes only the 'regular' articles; if
many citations are to notes, letters, etc., then this may result in a
substantial upward bias (of say 50%) in the impact factor.

The impact factor published by ISI measures short term impact: a two year
citation window is adopted. This might be problematic since in some
disciplines, (eg in economics) the number of citations to an article
usually reaches a maximum after 4 years.

"Quality" has many different dimensions, hence, it is somewhat ambiguous a
concept. The number of citations tell us how often an article or journal
has been used. Review articles and review journals present ideas in an
accessible way and will be more widely cited. Some fundamental ideas are
initially understood only by few people and the papers containing them may
not be frequently cited in the first couple of years after they appeared.

The impact factor is calculated on the basis that all citations are
equally important. However, if one wants to investigate the impact of a
journal on a certain field, one would want to weigh journals in that field
more heavily than others. More generally, one might wish to vary the
weight with the importance of the citing journal. Gerard Kohler

**********************************************************
Prof. N.G. Piersonbibliotheek
Gerard  K=F6hler;  kohler@fee.uva.nl
tel +31205254275 fax 5836
Agenda http://agenda.fee.uva.nl/default