Summer dreams

© 2003 Nature Publishing Group http://www.nature.com/natureimmunology EDITORIAL Summer dreams n early summer each year, ISI releases the previous year’s impact factors for scientific journals. These figures always generate a ‘buzz’ in the community, but often for the wrong reasons. Where do these numbers come from and how relevant are they? Impact factors are one of the indexes that ISI calculates from its collection of citation information. Every citation is recorded, along with the date the paper was cited. The data are then analyzed for citation trends, and comparisons are made between journals or authors. Impact factors are the average number of citations per paper in a given year and are calculated on two years’ worth of a particular journal’s reviews and research papers. For example, Nature Immunology’s latest impact factor, 27.868, is based on citations in 2002. It was calculated by dividing the citations in 2002 to Nature Immunology research articles and reviews published in 2000 and 2001 (5,490 of the 6,297 citations in 2002 to Nature Immunology) by the number of papers published in 2000 and 2001 (171 articles + 26 reviews = 197 papers). Dividing 5,490 by 197 gives the impact factor. Nature Immunology is placed first among primary immunology journals and second of all immunology journals. Although we are obviously happy with this result, we emphasize that impact factors require caution in their use and interpretation. Although Nature Immunology’s score was seventh of all journals ranked (with a higher impact factor than Science or Cell), that comparison does not carry the weight that many presume. Various fields traditionally have lower (or higher) rates of citation. Thus, without normalizing across disciplines, one can only draw conclusions within single fields or about a single journal’s trends. One can, however, take these numbers to estimate the degree with which a journal is accepted by its community, and on these grounds, Nature Immunology editors are reassured. The tenpoint leap from last year’s impact factor of 17.431 indicates continued validation of the journal in the eyes of immunologists. The editors gratefully acknowledge that this impact factor is a direct result of authors extending their trust in Nature, and referees extending their tradition of high standards, to this journal. Papers were selected based on their quality and with no thought of impact factor calculations. Our readers then rewarded this group effort by finding the papers useful enough to cite. Nature Immunology is proud to have become the medium through I NATURE IMMUNOLOGY VOLUME 4 NUMBER 8 AUGUST 2003 which some of the finest that immunology has to offer is communicated. Impact factors can be skewed by papers that receive extraordinarily high numbers of citations. Reviews are often targeted as likely suspects in pulling up otherwise lackluster impact factors, because a good review often receives more citations than an equally deserving primary paper. Therefore, to examine how well the Nature Immunology manuscript selection process has served our readers compared with the primary immunology journal that received the next-highest impact factor, Immunity, we tried to examine the effect of research papers alone. Because 2002 citations (the subset used to calculate 2002 impact factors) for each paper are not easily accessible, we instead used total citations accrued by each paper as of 20 June 2003 as our cruder gauge in this exercise. We found that about two-thirds of Nature Immunology’s research articles had more than 27 citations (our average number per paper from 2002), whereas only about half of Immunity’s research papers had the same. This indicates to us that although Nature Immunology does not publish the lion’s share of primary immunology, a large proportion of what we publish is solid and interesting enough to garner substantial numbers of citations. Concerns exist about the abuse of impact factors in the scientific community. Impact factors do not indicate an individual paper’s quality or an individual scientist’s potential, and yet we have all seen such misuse. Bad papers do appear in selective journals. Publication in a ‘top’ journal may be a rough surrogate for the acceptance of that particular piece of work by colleagues in the field, but is of little value as a surrogate for analyzing a scientist’s record and predicting future performance. Perhaps candidates should provide a graph of what could be called ‘annual CV factors’ (the number of citations received in a certain year on papers published the two previous years—the same formula used by ISI to calculate impact factors). This may be a better indication of a scientist’s changing stature in his or her subfield and, when used with other criteria, be of value in hiring and tenure decisions. However, comparisons of individuals in subfields are subject to the same caveats as mentioned above for comparisons of journals in different fields. Thus, impact factors for the most part should be treated in the spirit of the season from which they spring— good beach reading, but not to be taken too seriously. 715  (...truncated)