Skip to main content

Advertisement

Impact factor, H index, peer comparisons, and Retrovirology: is it time to individualize citation metrics?

Article metrics

Abstract

There is a natural tendency to judge a gift by the attractiveness of its wrapping. In some respect, this reflects current mores of measuring the gravitas of a scientific paper based on the journal cover in which the work appears. Most journals have an impact factor (IF) which some proudly display on their face page. Although historically journal IF has been a convenient quantitative shorthand, has its (mis)use contributed to inaccurate perceptions of the quality of scientific articles? Is now the time that equally convenient but more individually accurate metrics be adopted?

I surmise that a common question posed to an editor of a new journal is "What is your impact factor?" Based on my experience, in the majority of instances as the conversation evolves, it becomes evident that the questioner misunderstands what impact factor means. IF is a useful number. However, its limitations must be clearly recognized. Given the pervasive (if not obsessive) interest in IF, Retrovirology, as a new journal entering its fourth year of publication, has necessarily mined the citation databases and calculated IF numbers for 2005 (2.98) and 2006 (4.32) [1]. After having captured those numbers, it is perhaps instructive to consider some factual denotations and frequently misinterpreted connotations of IF. Indeed, as science and medicine march to a more personalized approach, one might further ask if it is time to embrace highly accessible technology in order to complement/supplant generic IF with individually precise citation metrics?

Impact Factor --- what does it (not) say?

In the 1960s, the Institute for Scientific Information (ISI), a component of Thomson Scientific, a division of The Thomson Corporation (a publicly traded company engaged in financial services, healthcare sectors, law, science and technology research, and tax and accounting services) devised the "impact factor" (IF), a number developed for the purpose of comparing different journals [2].

IF gauges the standing of a journal for a specified year. Hence, IF can be viewed as the mean number of citations that occurred in a specific year to articles published in a journal during the two previous years. In common vernacular, IF reflects the number of times an "average article" in a journal has been cited per year in the two immediately preceding years. A poorly-understood nuance to this definition is that IF disproportionately favors citations made during the first two years subsequent to a paper's publication, and does not accurately capture the paper's "value" over a longer time. Hypothetically, let's consider two papers that receive the same total number of citations (e.g. 100 times) over a 10 year period. Paper A is cited 80 times in its first two immediate calendar years after publication, and then 20 times over the subsequent eight years. Paper B is cited 20 times over the first two years and 80 times over the next eight years. Paper A fits the profile of an average article published by the journal "FlashyStuff"; while paper B is a usual paper in the journal "ReliablySolid". Within the context of this example, FlashyStuff would sport an IF of 40 while ReliablySolid would have an IF of 10. Intriguingly, the rather impressive 4 fold difference in impact factor belies common sense --- that over ten years, a FlashyStuff article is cited no more frequently than a ReliablySolid paper (both exactly 100 times).

The above discussion briefly spotlights what IF in part does and does not convey. With that disclaimer, how is Retrovirology doing IF-wise as the journal enters its fourth year? Employing the algorithm that IF derives from the number of citations to a journal (by other ISI tracked journals) divided by the "citable items" published in the journal, in 2005 Retrovirology had a calculated IF of 2.98 based on ISI data. For 2006, Retrovirology's IF is calculated using the following parameters : N (numerator) = the number of times Retrovirology articles published in 2004 and 2005 were cited by other journals during the single 2006 year, and D (denominator) = the total number of citable Retrovirology articles published in years 2004 and 2005. Retrovirology's 2006 IF is then N divided by D = 4.32.

How does 4.32 stack up against numbers from other journals? Figure 1 compares Retrovirology's 4.32 IF with recent IF numbers of other virology journals that frequently publish basic research on retroviruses. When measured against Journal of Virology, Journal of General Virology, Virology, Current HIV Research, AIDS Research and Human Retroviruses, and Intervirology, Retrovirology, the only Open Access publication amongst the seven, indeed rates very competitively (Fig. 1).

Figure 1
figure1

A comparison of Retrovirology's calculated 2006 impact factor with selected journals that publish retrovirus research papers. Values for other journals are from ISI journal citation report for 2005.

Now for something completely different?

In the late 1970's when I began graduate school, large bulky word processing machines were just being invented, and small personal computers did not exist. This was a period when if one wished to learn what was being published, one had to reach for the weekly/monthly periodicals (that often meandered through the postal service sometimes, if it is a foreign journal, arriving months after publication) which were displayed on reading shelves in libraries. PubMed, other electronic databases, email, keyword e-alert, and instant table of contents notification were science fiction. In that era, it was laborious and time consuming to assess individually a journal's or a colleague's citation records. Hence, back then, judging a "book by its cover" or rating a paper based on the journal's IF would seem excusable simply because there was little other practical recourse.

In 2007, one can do much better. In the fifty years since the advent of IF, a couple of salient shortcomings to this index have been noted. First, an inherent quirk to IF definition allows the numerator to contain citations to "non-citable" items that are not counted in the denominator [3]. In general, "non-citable" items are the short newsy/opinion/commentary "front matter" pieces written by professional writers which appear frequently in many "high impact" journals. Citations to "front matter" pieces are tallied in the numerator for a journal's IF without a commensurate "penalty" added to the denominator. Hence, venues that publish numerous "non-citable" front matters have "inflated" IFs relative to counterparts that publish only "citable" articles (e.g. original research papers). Second, a journal's IF is a poor surrogate indicator of individual articles published in that journal. There is a statistical pattern to citations that on average 15% of the articles in a journal accounts for 50% of all citations to that journal, and the top 50% of articles in a journal garner 90% of citations to that journal [4]. Thus, a top 50% article can be cited 10 times more frequently than a bottom 50% article in the same journal [4]. Given the likelihood of a 10 fold difference in actual citations, why should colleagues assume that one Cell/Science/Nature paper has remotely the same value as another? Hence, even if one accepts citation frequency as a reflection of quality, there is little reason to adopt a journal's overall IF as a reliable touchstone for the gamut of papers published in that journal. The umbrella-like use of IF as a general quality tag seems all the more unnecessary since there are so many rapid and accessible options for tracking article-specific citations (e.g. Google Scholar, Scopus, and Web of Science [5]).

A couple of days ago, I read a remarkable news headline. "James Watson of DNA fame gets his own genome map". I recollected that when the first human genome sequencing was being done generic anonymity of that initial DNA was important. However, time has changed, and today James Watson (and you too) can have an individual genome sequenced rapidly and inexpensively. Is now not also the time that scientists should move to personalized measurements of citations? Aren't individual citation frequencies more thoughtful reflections of one's scientific corpus than the answer to the oft-bantered generic query "How many Cell/Science/Nature papers has he/she published?"

Today, one's individual citation frequency is easily accessible to all who have a few minutes to spend and internet access to databases. Google Scholar, Scopus, or Web of Science can each fully provide such information. Of the three, I found Scopus [6] to be the most user-friendly in its data organization and searchability. Hence, yesterday when I had a spare hour, I used Scopus to tabulate individual citation frequency and H index of 45 members of Retrovirology's editorial board (Table 1). {The H index is another way to quantify a scientist's quality and quantity of scholarly output [7, 8]. This index attempts to combine and balance the effect of "quantity" (number of publications) and "quality" (citation rate) in a specific way.} I should point out that I did this data collection quickly (each person's numbers took no more than 1 minute), and as with all databases and human entries there can be errors (apologies to colleagues if I made mistakes with your numbers). Hence, please take table 1 to be illustrative rather than factually literal. Nevertheless, this elementary exercise echoes the words of a past US president, "You can run, but you cannot hide." Like it or not, use it or not, each author's personalized citation number and H index are there for all to compare.

Table 1 Citation frequency and H index for selected Retrovirology Editorial Board members (data collated on June 11, 2007 from Scopus).

Gilt by association?

How then should one choose where to publish one's manuscript? It has been raised that many scientists employ the "gilt by association" [9] approach, first sending their papers to high-visibility, high-IF journals, perhaps hoping that the "free ride" hypothesis works [4] and some of a journal's sheen would direct attention to and rub off on the work. However, there is no factual evidence that publishing a paper in a highly touted journal adds "free citations" to a paper other than those achieved by its content [4]. Indeed, this point seems to make intuitive sense. For example, a paper is the same paper if it were initially declined at Cell and then published in the Journal of Virology than if the paper were quickly accepted at Cell and did not have to make the rounds to the Journal of Virology. If the paper remains the same, should one sequence of events confer higher inherent quality to the same paper over another?

In the past, to support the interest of equal access to knowledge by scientists and students in developing economies who cannot afford subscription-based journals, I have argued that we have a responsibility to support Open Access publishing [10]. From a principled point of view, not to do so is poorly defensible. On the other hand, if one formulates decisions using a self-interest citation frequency driven perspective, evidence similarly supports that in head-to-head comparisons Open Access articles are cited more frequently than non-Open Access counterparts [11]. In today's publishing world, there are important roles to be played by both subscription and Open Access journals. However, as Open Access journals ascend in quality and visibility and globalization brings us closer to previously distant strangers, scientists confident in the inherent content and value of their papers might ask if they can tolerate the "guilt associated with not supporting egalitarian access"?

The above is a difficult question that each individual has to ponder. I am grateful for the answer that many authors and editorial board members of Retrovirology have provided to this question.

References

  1. 1.

    Jeang KT: Retrovirology: 3 at age 2. Retrovirology. 2006, 3: 30-10.1186/1742-4690-3-30.

  2. 2.

    Garfield E: The history and meaning of the journal impact factor. JAMA. 2006, 295: 90-93. 10.1001/jama.295.1.90.

  3. 3.

    Dong P, Loh M, Mondry A: The "impact factor" revisited. Biomed Digit Libr. 2005, 2: 7-10.1186/1742-5581-2-7.

  4. 4.

    Segelen PO: Why the impact factor of journals should not be used for evaluating research. BMJ. 1997, 314: 498-502.

  5. 5.

    Bakkalbasi N, Bauer K, Glover J, Wang L: Three options for citation tracking: Google Scholar, Scopus and Web of Science. Biomed Digit Libr. 2006, 3: 7-10.1186/1742-5581-3-7.

  6. 6.

    Scopus. [http://www.scopus.com/scopus/home.url]

  7. 7.

    Braun T, Glanzel W, Schubert A: A Hirsch-type index for journals. The Scientist. 2005, 19: 8-

  8. 8.

    Hirsch JE: An index to quantify an individual's scientific research output. Proc Natl Acad Sci USA. 2005, 102: 16569-16572. 10.1073/pnas.0507655102.

  9. 9.

    Grivell L: Through a glass darkly. EMBO Rep. 2006, 7: 567-570. 10.1038/sj.embor.7400718.

  10. 10.

    Jeang KT: Libya, HIV, and open communication. Retrovirology. 2006, 3: 99-10.1186/1742-4690-3-99.

  11. 11.

    Eysenbach G: Citation advantage of open access articles. PLoS Biol. 2006, 4: e157-10.1371/journal.pbio.0040157.

Download references

Acknowledgements

I thank Ben Berkhout, Andrew Lever, Matt Cockerill and Robert Brumfield for critical readings of this editorial. The opinions expressed in this editorial are personal to the author and do not necessarily represent the views of his employer, the NIH. Work in KTJ's laboratory is funded by intramural funds from the NIAID, NIH.

Author information

Correspondence to Kuan-Teh Jeang.

Authors’ original submitted files for images

Below are the links to the authors’ original submitted files for images.

Authors’ original file for figure 1

Rights and permissions

This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

Reprints and Permissions

About this article

Cite this article

Jeang, K. Impact factor, H index, peer comparisons, and Retrovirology: is it time to individualize citation metrics?. Retrovirology 4, 42 (2007) doi:10.1186/1742-4690-4-42

Download citation