Sharp D. "As we said it..." Lancet 364(9436). p.744, August 28, 2004.

[Eugene Garfield]

TITLE :  As we said ... (Editorial Material)

AUTHOR:    Sharp D

SOURCE:   Lancet 364(9436). p.744, August 28, 2004

Author Address : D. Sharp, The Lancet, London NW1 7BY, England

REPRINTED WITH PERMISSION FROM ELSEVIER (THE LANCET, 2004, 364(9436) PAGE
744.  http://www.lancet.com


As we said...

Any author or editor who looks closely at citation statistics will find
temptations for self-advancement. Stick to review articles or methods in the
hope of striking occasional gold; the most cited Lancet paper (hereby
acquiring yet another hit) is a method,(1) and second and third prizes go to
publications of this sort too. Why not cite another research group more
often than science or common decency demands, provided of course they do the
same for you? Editors could encourage journal self-citation or tease out the
advantages to be had from the different bases for an impact factor's
numerator and denominator.(2) If no one else wants to refer to your published
efforts you had better do so yourself lest they  sink without trace, the
fate of about a quarter of mainstream science.(3) and (4) Apoor Gami and
colleagues(5) recently estimated the frequency of author self-citation but
we do not know what proportion, if any, is unworthy. George Bernard Shaw
almost said: "I often quote myself. It adds spice to my career prospects."

For 266 papers on diabetes published in clinical journals in 2000 and
receiving one or more citations by April, 2003, Gami and colleagues5 found
an average self-citation rate of 18% (median 7%). In the infancy of modern
citation analysis, the self-citation rate was 8% for first authors
alone(6)and a massive study on Norwegian data revealed self-citations
running at 17% for clinical medicine,(7) which is very close to what Gami
and colleagues found for diabetes.

A ban on self-citation is clearly ridiculous as it would remove much of the
"why?" of a study and would lead to wasteful repetition of the "how?" if the
detailed methodology had been published already. Garfield and
Welljams-Dorof(8)thought that excessive self-citation would be corrected by
 the editorial and peer-review process. The potential for such correction is
there-but the reality? Asknes7 wants "Normative limits for how often one can
cite oneself". Yet no one knows what normal or excessive mean in this
context. The detailed table linked to Gami and colleagues' paper5 suggests
that calling a guillotine into play when self-citations rise above, say, one
in five would not make a lot of difference, and self-citations can be
eliminated from literature searches.9 Similarly, a recent essay10 points to
only a weak correlation between journal self-citation and impact, and any
editor succumbing to the temptation to urge authors to cite his or her own
journal could be in for a disappointment.
The Selfcite 2.0 and Egocite programs11 and 12 remind us that the subject
can be taken too seriously. No one would worry much, and studies such as
Gami and colleagues' might not be needed were it not for the fact that,
despite severe criticism13 and some attempts to do things differently,
citation indices are still used to judge science and scientists. How did
anyone get appointed or promoted before citation analysis expanded from the
1960s?

I write a regular column for the Thomson ISI publication Science Watch.

References
1 JM Bland and DG Altman, Statistical methods for assessing agreement
between two methods of clinical measurement, Lancet 1 (1986), pp. 307-310.

2 C Jennings, Citation data: the wrong impact?, Neuroendocrinol Lett 20
(1999), pp. 7-10.

3 D Hamilton, Research papers: who's uncited now?, Science 251 (1991), p. 25.

4 D Pendlebury, Science, citation, and funding, Science 251 (1991), pp.
1410-1411.

5 AS Gami, VM Montori, NL Wilczynski and RB Haynes, Author self-citation in
the diabetes literature, CMAJ 170 (2004), pp. 1925-1927.

6 E Garfield and IH Sher, New factors in the evolution of scientific
literature through citation indexing, Am Documentation 14 (1963), pp.
195-201.

7 DW Aksnes, A macro study of self-citation. Seventh Nordic Workshop on
Bibliometrics and Research Policy
(www.nifo.no/bibliometri/Aksnes%20Self-citation.pdf) (Oct 11-12, 2002)
(accessed July 13, 2004).

8 E Garfield and A Welljams-Dorof, Citation data: their use as quantitative
indicators for science and technology evaluation and policy-making, Sci Publ
Policy 19 (1992), pp. 321-327.

9 National University of Singapore Libraries, How to search WOS: eliminating
self-citation search
(http://www.lib.nus.edu.sg/lion/slb/d/wos/wsselfcite.hml) (accessed July 8,
2004).

10 ME McVeigh, Journal self-citation in the Journal Citation Reports(r)-
Science edition (2002): a citation study from the Thomson Corporation
(http://www.isinet.com.ezproxy.library.drexel.edu/essays/journalcitationrepo
rts/8236411/) (accessed Aug 12, 2004).

11 N Craddock, MC O'Donovan and MJ Owen, Introducing Selfcite 2.0-career
enhancing software, BMJ 313 (1996), pp. 1659-1660.

12 PJ Morrison, Making the most of self-citation, BMJ 314 (1997), p. 832.

13. PO Seglen, Why the impact factor of journals should not be used for
evaluating research, BMJ 314 (1997), pp. 498-502.