|
As Andrea Horgan noted in her July 6
news roundup (click
here to read the full text), "The annual release of the
Impact Factor is awaited with mixed feelings by many journal editors."
The release of the 2001 edition of the JCR®, which contains
the Impact Factor, prompted us to take a look at this influential
index - what it really is, some thoughts on how to use it, and how
it might be improved.
Since its inception in the 1960's, the ISI Impact Factor (IF) has
become the most widely-used commercial indicator of the standing
and influence of a journal. In fact it is the ONLY commercial indicator.
|
Based on pioneering research into citation patterns and journal
ranking by Gross and Gross in the 1920's and Brodman in the 1960's,
the IF is a measure of "the frequency with which the 'average
article' in a journal has been cited in a particular year or period."
(Please also see an essay on the Impact Factor by Dr. Eugene Garfield,
founder of ISI and Chairman Emeritus, by clicking here.)
The IF is based on a simple premise: researchers cite previously-published
articles in their own publications, and, if the "best research"
is published in the "best journals" - then these journals
will be cited most frequently. ISI creates the index using total
article counts to balance out frequency of publication and the number
of articles per issue.
|
| Although widely-used, it is astonishing
to note that, at least in our experience, most users cannot articulate
what the IF actually measures or how it is deprived.
At its essence, it is a citation-derived index of journal standing
and influence.
ISI calculates the IF as follows:
- From all journals in their database, ISI counts citations to
a particular journal in the current year, e.g., 2001, to articles
published in the two previous years, e.g., 1999 & 2000 = A
- Count total articles published in the journal during 1999 and
2000 = B
- Divide A (2-year cites) by B (total articles)
|
A few key points:
- The IF relates to a journal, not an article
- It is a retrospective analysis, published approximately 6 months
in arrears (the new 2001 Impact Factors have just become available
via the current issue of the JCR® - these IFs relate to citations
of articles from 1999 and 2000 in 2001 articles)
- Most users agree that it is best used to rank journals within
a discipline, as there is such a strong inter-discipline range
Why the wait? Why not factor 2001 cites? ISI has an index for this
too, the Immediacy Index®, which is also available via the JCR.
From the perspective of the IF, think for a moment of the normal
lead time for publication. It can easily be 9-18 months from submission
to publication, so it comes as no surprise that relatively few articles
actually get cited in the year that they were published.
|
|
Immediately you can see the wide range in both number of journals
monitored and the range of IF.
Our goal was to establish what represents a "good" IF.
Taking our analysis one step further, look at the percent of journals
with an IF of >=3, >=2, >=1, or >=0.
|
You can see that for oncology,
an IF of >=3 is clearly "good," whereas for Dermatology
this same IF range probably denotes "excellent"! Many users
adopt a rule of thumb that journals with an IF of 2 and above are
"good," at least as measured by this citation analysis tool.
Many also use the IF to place of their target journal in a rank of
all other journals in that specialty. This is also frequently used
by publishers: you will often see them refer to a journal as ranked
"15 out of 34 for (target specialty)." |
|
Perhaps the biggest flaws in the index are its:
- Inability to track the IF across a change in title
- The inordinately high rankings given to review journals
- The apparent bias toward (academic) research journals (perhaps
unavoidable, as this is a combination of ISI's collection development
policy and the origin of the index itself)
Using the IF alone can quickly confuse your perception of a journal's
true standing. For example, is published by the National
Cancer Institute, and is freely available to all US primary care
physicians. The journal publishes
|
updated incidence and prevalence data in its January issue each
year, which has resulted in it becoming the in all of biomedicine! A clear
24% higher than .
The literature is evenly balanced between the pros and the cons
of the IF. It is unquestionably useful, and as long as it remains
the only commercially available and quantitative rank, it will continue
to be widely used. It could, however, perhaps benefit from an overhaul
as it approaches "middle age."
|
