Portal de Periódicos da CAPES

Impact factor 2.0: a name can be everything

2018; Elsevier BV; Volume: 110; Issue: 1 Linguagem: Inglês

10.1016/j.fertnstert.2018.05.022

1556-5653

David R. Meldrum,

scientometrics and bibliometrics research

The impact factor (IF 1.0) hasn't been updated in the more than six decades since it was first suggested in 1955. Today there is a broad consensus that it is an inaccurate and misleading indicator for its intended purpose to rank journal quality. This Inklings argues for a long overdue update to IF 2.0 by using existing calculations and a simple conversion to familiar digits. Anyone who has been publishing in recent years has noticed the extreme proliferation of medical journals. Not a day seems to pass without a few or several email invitations to submit an article or review a paper from newly formed journals with little if any scientific reputation. A cardiologist documented online that during a two-month period he was assaulted by 26 heart-related journals and 75 non-heart-related journals (http://www.ajconline.org/article/S0002-9149(15)02213-4/abstract)! As the number of journals has skyrocketed, virtually anyone with reasonable status, word processing software, and an internet connection can publish their work or a review article. Unfortunately, IF 1.0 adds all citations to the numerator, each having an equal effect on the final calculation of IF (total citations divided by scientific papers published), whether a citation was from an obscure journal which very few read or respect, or from Nature, Science, or the New England Journal of Medicine. Giving equal weight to citations from a recently formed enterprise as would be given to these esteemed journals is clearly inappropriate and misleading. There are available calculations, of which the Eigenfactor (EF) score, is one of the earliest, that appropriately corrects for this perversion of scientific value by assigning more weight to citations from journals with higher influence. The problem is that the term impact factor, by describing itself perfectly, is such a powerful name that it has been almost impossible to displace from its position of prominence in everyone's minds and jargon. The EF score (and the Article Influence Score which is based on EF) appeals to the intellect, but the complexity of its methodology is much more difficult to understand, the calculated result is a small fraction of 1, and it totally misses the mark as a name. Solution #1 is simple: rename EF as IF 2.0. We're on iOS 11.3 for the Apple iPhone and OS 13.3 for Apple computers. We would not for a nanosecond consider regressing to versions 1.0. It's shameful that in this narrow aspect of science we still are stuck on IF 1.0, lacking any comprehensive update since it was first proposed in 1955. The ability of poor quality journals to have impact that is identical to journals whose prominence has been hard-earned over decades by some of the best researchers and clinicians in the world is now, with the proliferation of low quality journals, a more urgent problem than ever before. This is a further instance of where information technology can be a disruptive influence that continually demands action and is ignored only at considerable peril. There are other items that chip away at the impact factor's validity that seem, at first glance, to be more difficult to remedy. The effect (and abuse) of self-citations has been a major concern, and the more references each paper has, the greater the likelihood of its own journal being cited. There is monitoring of the proportion of self-citations, with journals having a high percentage being subject to review and being mandated to take corrective action. However, that threshold of 20% of references is high enough to not be an adequate solution, and there is no correction for the number of self-citations per paper. The logical way to simply eliminate that variable and even the playing field is to remove all self-citations from the numerator, which EF (and therefore impact factor 2.0) does as part of the calculation. Also, the more members/subscribers of the sponsoring society and its journal, the more chance that as authors, they will be familiar with papers published in that journal. They may also do part of their literature review using that journal's search function, therefore bringing up only papers published in that journal. The paper they are writing may be published in a different journal, thus escaping any self-citation oversight. That variable can have a very large impact that will increase as more fledgling journals become established. I have not seen discussion of this important influence, which gives an undue advantage to journals with more subscribers. By removing self-citations and giving less weight to citations from less established and respected journals, IF 2.0 would make a limited correction for this inappropriate perturbation. Another advantage of EF (and therefore IF 2.0) is that it bases the denominator on a five-year rather than the two-year period of publications with IF 1.0, thereby reflecting a more broad, lasting, and stable impact of a journal, akin to the five-year impact factor. Otherwise, publication of a highly cited paper can skew the results for a particular year. A flash in the pan should not earn a leading position. IF 1.0 does have a clear intuitive advantage. The calculation results in a number from 1 or less for a low-quality journal to higher single digits as quality increases, to double digits for very high-quality journals gathering a large number of citations. Solution #2: simply multiply the EF score by 100 to normalize IF 2.0 to digits more akin to IF 1.0. Table 1 illustrates IF, EF, and the normalized IF 2.0 for four good quality reproductive medicine journals and four very high-quality journals for 2016 (1InCites Journal Citation Reports. Journal titles ranked by impact factor.http://jcr.incites.thomsonreuters.comDate accessed: May 16, 2018Google Scholar). One can see that rankings do change (as they should) as the playing field is evened by removing self-citations, weighting citations according to the quality of the journal of origin and extending the denominator to 5 years.Table 1Impact factor (IF), Eigenfactor (EF) score, the normalized impact factor 2.0, and comparative rankings are given for four good quality reproductive medicine journals and four very high-quality journals for 2016 1InCites Journal Citation Reports. Journal titles ranked by impact factor.http://jcr.incites.thomsonreuters.comDate accessed: May 16, 2018Google Scholar.JournalIF 1.0 (rank)EFIF 2.0 (rank)Reproductive medicine Am J Obstet Gynecol5.23 (1)0.050325.032 (3) Fertil Steril4.44 (4)0.050345.034 (2) Hum Reprod5.02 (3)0.0414.1 (4) Obstet Gynecol5.22 (2)0.0565.6 (1)Very high quality Nature40.1 (3)1.43143 (1) Science37.2 (4)1.15115 (2) N Engl J Med72.4 (1)0.6969 (3) Lancet47.8 (2)0.4040 (4) Open table in a new tab As if the previously mentioned perversions of impact factor 1.0 were not enough, today promotions in many research and academic settings are influenced by the individual candidate's IF 1.0. Consequently, we have all seen instances, whether we recognized them or not, of papers being submitted to a journal with a higher IF when they would have been most appropriate for a journal with a lower IF within the discipline generally covering the substance of the paper. Those papers are then less often read and appreciated by those who would, together with their patients, benefit the most. To make matters worse, it perpetuates any differences among IF's that result in such misdirecting of manuscripts to game academic promotions. Academic and scientific organizations should not assign any merit to an inaccurate measure like IF 1.0 when other, more appropriate indices are available. Being based on the prior 5 years of publications, IF 2.0 should be quite suitable for promotions. But can we, after using an obsolete but ingrained index for decades, actually change? That is where the rubber meets the road. Einstein was quoted to have said, “The measure of intelligence is the ability to change,” not, “A measure,” but, “The measure”. With the plethora and complexity of the many bibliometric indices that have been proposed (almost 40), it will require concerted leadership by motivated, well-respected individuals to forge a consensus. Each journal's editors naturally will tend to favor an index, even IF 1.0, which favors their ranking position, making such an effort more difficult. In summary, it's a disgrace to academics and science that we are still using and misusing the grossly misleading IF 1.0. Film, and its hold-out, Kodak, lost their leading positions in the world of photography to the digital revolution and IF 1.0 has far outlived its usefulness. Entities still clinging onto this ancient index are at risk of having their own Kodak moment. This Inklings proposes the adoption of a modified EF score, with correction to yield a more familiar digit, thereby giving it a new lease on life as impact factor 2.0. All that is needed is a new column in the Journal Citation Reports. A name can be everything.