Once upon a time, journals were made of paper and ink. However, we left the dark ages of dead woods behind us and moved forward to an age in which authors don’t need to publish in journals (but still want to). There’s an increasing decoupling between the individual article and its publishing journal, created by search engines and electronic repositories. “Are elite journals declining?” is the title of a new article by Larivire, Lozano and Gingras which was published, how appropriate, in ArXiv. An earlier study by the same authors found evidence that the relationship between the number of an individual article’s citations and the Impact Factor of its publishing journal has been declining since 1991. They also found that the proportion of best-cited articles in journals with high IF has been declining as well since that time (Lozano, Larivire & Gingras, 2012).
This time they wanted to see whether these changes apply specifically to top-tier journals rather than being just a general trend. They looked at seven “elite” journals as well as at a group of six “emerging” journals. The “elite” journals in the sample had to be considered as “prestigious” to have had a high Impact Factor in 2011, not be review journals, publish a high number of articles every year and exist for at least several decades. They chose three multidisciplinary journals: Science, Nature, PNAS (included for its prestige rather than its IF) and four biomedical journals: Cell, Lancet, New England Journal of Medicine (NEJM), and the Journal of the American Medical Association (JAMA). The “emerging” group journals were chosen because they have had the highest growth in their proportion of top cited papers over the past 40 years. The group includes PLoS One, Nano Letters, Advanced Materials, Nature Materials, Journal of Clinical Oncology and Chemical Reviews. Of these journals PLoS One, Nature Materials and Nano Letters have been founded in the 21st century, Chemical Reviews is older by several decades (1925) and the Journal of Clinical Oncology and Advanced Materials have been published since the 1980s.
But, before we get into the decline of the journals, let me mention that the “Highly cited” threshold is going up. When the authors ranked the top articles for each year since 1970, they found that to be included in the 1% and 5% best-cited, articles have to receive about twice as many citations as they did 30-40 years ago.
Back to the journals; In comparison with 20-25 years ago, all the elite journals studied publish now a smaller proportion of highly cited articles. Also, they all publish a larger proportion of the highly cited 1% articles than of the highly cited 5% articles. PNAS, for example, went from publishing almost 9% of the 1% highly cited and over 4% of the 5% highly cited in the mid-1980s to 2.7% and 2.2% respectively in 2010. Nature and Science used to publish 20-25 years ago about 7% and 6% of the top 1% articles, but now publish only 4% and 3% respectively. The NEJM’s share of articles in the top 1% has been decreasing since the 1980s, while the Lancet managed to keep it relatively stable. However, they both went from having about 1% of the 5% top articles in the 1970s to about 0.5% in 2010. JAMA increased its proportion of articles in the 5% until the late 1990s and its proportion of articles in the 1% until the early 2000s, but both proportions went down again.
Since the number of articles in most journals has increased since the 1970s, the authors wanted to make sure the changes in proportions weren’t merely a by-product. They calculated a yearly normalized top 1% index for each journal. It measures the number of the top 1% articles published in a journal in a specific year in comparison with the number of top 1% articles it was supposed to publish if journals published top articles at random. As the authors put it “for a given year, it is the relative number of top papers in a given journal (top papers / total papers) divided by the proportion of top papers published in all journals.” If the journal publishes the exact number of top articles we expect it to publish by chance, the coefficient will be 1. Naturally, the coefficient of high-impact journals is a lot higher than one - that is, they publish way more top articles than can be expected by chance. However, even by the normalized index, high-impact journals publish less top articles than they used to (Figure 2).
The emerging journal group also publishes more top 1% articles than can be expected by chance (Figure 3), and Chemical Reviews and Nature Materials are now in the 40-50 coefficient range. PLoS One fans would have a bit of a disappointment for here, since its coefficient is in the “low single digits” as the authors put it (Figure 3) and that is probably because of the large number of articles it publishes per year.
The digital age means it’s easier to create and publish new journals, which are bound to decrease the market share of the old ones. Some older publishers try to deal with the problem by creating their own new specialized journals. For example, the Nature Publishing Group now publishes almost 40 journals under the “Nature something” title, which all enjoy a bit of the “Nature” prestige. The proportion of the top 5% articles of all the Nature journals combined went up from 3% in the 1970s to about 5% today. The authors point out that while the Nature Publishing Group is for-profit, the non-profit organization American Association for the Advancement of Science (AAAS), which publishes Science, only owns three peer-reviewed journals.
Are top-tier journals in decline? My suggestion is not to cancel the Nature subscription yet. The top journals still publish many highly cited articles. Journals, in general, still matter – even though this paper was first published in ArXiv, it’ll be published in a journal as well, to receive its “stamp of approval.” However, the decline in the share of top articles published by older elite journals means that first, articles can be judged more independently than they used to and second, that the hierarchy of journals might be changing.
Vincent Lariviere, George A. Lozano, & Yves Gingras (2013). Are elite journals declining? arXiv: 1304.6460v1
George A. Lozano, Vincent Lariviere, & Yves Gingras (2012). The weakening relationship between the Impact Factor and papers'
citations in the digital age arXiv: 1205.4328v1