ART:Takahashi,Impact Factor & Small Fields

[Gretchen Whitney]

CONCERNING TOPICAL BASED IMPACT FACTORS
The following letter is reproduced with the permission of the authors.
This is an interesting approach to solving the problem of computing impact
factors for small fields. In my correspondence with Dr. Takahashi I
cautioned him about relying on the assumption of objectivity in MEDLINE
since it involves human indexing which has advantages and disadvantages.
The latter includes inconsistency between indexers and changing
terminology. Nevertheless, I believe their method will produce  useful
results, even though the universe of articles retrieved may not include
all the articles that would be relevant to certain fields especially those
appearing  in cognate journals. Similar cohorts of articles could be found
by a combination key word and cited reference search of SCI or by creating
a co-citation cluster of articles on a topic. I recently performed an
analysis of apoptosis(cell death) by retrieving all papers with either of
those two terms in the titles of articles published over a fifteen year
period. Eugene Garfield
---------------------------------------

An Alternative to Journal-Based Impact Factors (Letter)

Ken Takahashi(1), Tar-Ching Aw(2), David Koh(3)

Published in: Occup Med (Oxf) 1999; 49 (1): 57-8.

1 Institute of Industrial Ecological Sciences, University of Occupational
and Environmental Health, Japan
2 Institute of Occupational Health, University of Birmingham, Birmingham,
United Kingdom
3 Department of Community, Occupational and Family Medicine, National
University of Singapore, Singapore

Correspondence: Ken Takahashi, M.D., Associate Professor, Institute of
Industrial Ecological Sciences, University of Occupational and
Environmental Health, Orio, Yahatanishiku, Kitakyushu City 807, Japan
FAX: +81-93-601-7324
e-mail: ktaka at med.uoeh-u.ac.jp

Editor - The Impact Factor (IF) of a peer-reviewed journal confers a
degree of credibility and importance to papers published in it.  The
higher the IF the better the recognition accorded to the papers within.
The journal IF is an index calculated by dividing the number of current
year citations to the source items published in that journal during the
previous (usually two) years.  Thus it is a measure of the frequency with
which the "average article" in a journal has been cited in a particular
year or period (1).  Under the current system, many occupational health
journals have IFs which rarely exceed two, whereas many well-established
multi-disciplinary journals for other subjects have much higher IFs Ð
sometimes even exceeding twenty. Individual papers in the same journal are
therefore valued to a similar extent regardless of differences in quality
as long as they appear in the same journal, and if that journal has a high
IF then the paper is rated highly by the system.  However, the
contribution of any individual paper should not be assumed from the IF of
the journal alone.  It is recognized that there is poor correlation
between citation counts of individual papers and journal IFs (2).  Journal
IFs also vary according to differences in the publication customs (3) and
decision process in selection of papers for publication across different
disciplines.

If citation counts are nevertheless accepted as a legitimate reference
point for judging scientific contributions, an improved index is needed to
allow the comparison of papers by subject area or discipline.  To this
end, we propose the use of topic-based IFs.  This can be defined as the
average number of citations received by all articles on a specific topic
over a defined time.  This index allows comparison of citations for any
specific article with the average number of citations for a group of
articles on the same theme.  This index deliberately avoids grouping
citation counts of papers dealing with different topics and thus is in
contrast with journal IFs and other ÔadjustedÕ IFs proposed (4).


If topic-based IFs are to serve as useful indicators, they will have to be
produced systematically.  At present, information from two existing
databases can be linked to achieve this goal, i.e., the citation database
of the Institute for Scientific Information and MEDLINE.  Papers in
MEDLINE are indexed with descriptors using medical subject headings.  Such
headings can be used to group articles on the basis of related topics.
For individual published papers, there are usually several major
descriptors indexed.  Hence, one paper can contribute to several topic
groups based on the major descriptors.  The topic-based IF can then be
calculated by dividing the number of citations received by an article
during a specific time period by the number of articles in the topic group
to which the article belongs (Table).


Citations from all sources should be included in the exercise.  Articles,
journals, and researchers from a narrow subject area in the biomedical
field will then be in a better position to compare the impact of their
papers against those of their group.  This is because the basis for
comparison will be publications on specific themes rather than the journal
IFs which favor papers in a limited number of journals that are relatively
well-established.  The current system perpetuates the practice of
preferential submission of research papers to such journals (sometimes
regardless of the topic).  Topic-based IFs alleviate problems with
differences in publication customs, readership or the number of
academics/researchers working on a specific topic area.  The journals can
identify the strengths and weaknesses of papers on specific topics, as a
number of topic-based IFs can be calculated for different types of
articles.  This will also provide researchers and administrators with a
better basis for comparing the impact of their research. (601 words)

References

1. Garfield E. Citation comments. Current Contents 1994 (June 20); 22
(25): 3-8.

2. Seglen PO.  Why the impact factor of journals should not be used for
evaluating research.  Brit Med J 1997; 314:498-502.

3. Garfield E. Citation data is subtle stuff - a primer on evaluating a
scientist performance.  Scientist 1987; 1(10):9.

4. Huth EJ. Mapping the land of medical journals: some new applications
for citation data from Science Citation Index.  In: Lock S, ed. The Future
of Medical Journals Ð in commemoration of 150 years of the British Medical
Journal. London, UK: BMA Publishing Group, 1991: 81-92.


 Table. Hypothetical data to calculate "topic-based" impact factors using
topics related to occupational health

---------------------------------------------------------------------------

                                       N of       N of citations    Topic-
                                       articles   rec'd by          based
                                       indexed    the group of      Impact
                                       by topic   articles          Factor
TOPIC                                  (1)        (2)               (2/1)
--------------------------------------------------------------------------
Occupational diseases                  1000       1500              1.5
Occupational diseases-epidemiology*     400        880              2.2
Occupational exposure-adverse effects   200        360              1.8
Pneumoconiosis                          150        200              1.3
Asbestos                                300        570              1.9
--------------------------------------------------------------------------
* Example of descriptor combined with subheading.





<><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><>
Gretchen Whitney, PhD                                     tel 423.974.7919
School of Information Sciences                            fax 423.974.4967
University of Tennessee, Knoxville TN 37996 USA           gwhitney at utk.edu
http://web.utk.edu/~gwhitney/