Single-Author Papers: A Waning Share of Output, But Still Providing the Tools for Progress
In 2012, ScienceWatch last revisited the topic of multiauthor papers—the trend toward scholarly publications listing authors whose numbers, in some cases, now reach into the thousands. In this report, we turn to the opposite end of the spectrum: papers credited to a single author, and how their fate has progressed during the concurrent rise of global “mega-authorship.” For this analysis, ScienceWatch turned to Thomson Reuters Web of Science and its extensive store of information on more than a century of published scholarly literature, including listings of authors. This survey concentrates primarily on the last three decades.
As was mentioned in the multiauthor report, the average number of authors on papers indexed since 1981 by Thomson Reuters has edged steadily upward. Graph 1 tracks this progression, from 2.48 authors in 1981, through a comparatively gradual phase in which an average of 3 authors in the early 1990s transitioned into 4 by the early 2000s, proceeding in fairly short order to exceed 5 in 2012—a doubling of the 1981 figure.
As the average number of authors rose, the proportion of single-authors was following a decidedly opposite trajectory. Graph 2 below records the percentage of single-author papers, out of all Thomson Reuters-indexed papers, from 1981 to 2012. In 1981, more than 30% of papers listed a single author. By 2012, the percentage had shrunk to 11.
This progression suggests factors that echo those discussed in the report on multiauthor papers: a pronounced increase in collaboration in recent decades, likely with a strong international component.
One note: Graph 2, as well as those that follow, excludes papers coded by Thomson Reuters as reviews (along with letters, abstracts, proceedings, and other indexed items apart from journal articles). This restriction on reviews constitutes an effort to capture original works of research and scholarship, as opposed to retrospective summaries or progress reports—which are frequently written by a single author—pertaining to a given topic or field. (Even though reviews, needless to say, are crucially important, highly useful, and, very often, highly cited.)
Notably, the annual number of single-author, non-review papers themselves, as tracked since 1981, has remained largely consistent in the course of the three decades. Thomson Reuters indexed 145,585 such papers in 1981, and 144,106 in 2012. In between, the number went as high as 166,046 (in 1996) and as low as 125,691 (2007), while mostly staying in the range of 140,000 to 150,000.
Meanwhile, even as the number of single-author papers was largely unchanging, overall scientific output was expanding apace: from 439,902 Thomson Reuters-indexed non-review articles in 1981 to 1.3 million in 2012. And the abundantly evident trend was away from single authorship.
THE SOCIAL SCIENCES HOLD OUT
To gauge the current state of single authorship by field, Graph 3 captures the number of lone-author papers as a percentage of all papers in each of 21 main subject areas, as of 2012.
As the graph makes plain, a few fields have maintained a fairly high percentages of single authorship—notably, the general Social Sciences, in which nearly 40% of papers list one author name. Also prominent, with both exceeding 25%, are Mathematics and Economics & Business. At the other end of the graph, indicating much higher levels of collaboration, is a group of life-sciences or biomedical fields, all of them below 5% in their shares of single-author papers. Lowest of all is Microbiology, which barely exceeds 1.50% for 2012.
To show the progression in these fields over the last three decades, Graph 4 below contrasts the 2012 percentage of single-authored papers in each field against the level recorded in 1981. (The fields are ordered according to the 1981 percentage.)
Once again, the main field of Social Sciences dominates, according to both the 1981 and 2012 measurements, with Economics & Business and Mathematics following closely. Each of those fields, in 1981, registered single authorship on upwards of 70% of its total papers.
By 2012, as Graph 4 makes plain, the rate of single authorship had fallen drastically in all the fields, with the greatest percentage drop in Economics & Business (42 points), and comparable reductions in Computer Science (40 points), and Mathematics (38). Social Sciences, from a high of 72% in 1981, also recorded a 30-points-plus decline to its 38% of single-authored papers in 2012.
As in the previous graph, the life- and biomedical sciences gravitate to the far end, demonstrating that, for both time periods, single-author papers constituted a distinct minority.
HIGHLY CITED METHODS, TOOLS
For a selection of highly cited single-author papers of comparatively recent vintage, ScienceWatch turned again to Web of Science. The table below collects the most-cited single-author papers published since 2002. Again, the intention was to skip conventional or standard review articles in favor of work more reflective of original effort (although, admittedly, the distinction can be difficult to draw).
Highly cited single-author papers published since 2002
(Listed by citations, and excluding standard review articles)
|G.M. Sheldrick, “A short history of SHELX,” Acta Crystallograph. A, 64: 112-22, 2008. [U. Gottingen, Germany]||31,280|
|A.L. Spek, “Single-crystal structure validation with the program PLATON,” J. Appl. Crystallography, 36: 7-13, 2003. [U. Utrecht, Netherlands||9,413|
|F.H. Allen, “The Cambridge Structural Database: a quarter of a million crystal structures and rising,” Acta Crystallograph. B, 58: 380-8, 2002. [Cambridge Crystallog. Data Ctr., UK]||6,867|
|D.G. Lowe, “Distinctive image features from scale-invariant keypoints,” Int. J. Computer Vision,” 60(2): 91-110, 2004. [U. British Columbia, Vancouver, Canada]`||6,495|
|R.C. Edgar, “MUSCLE: multiple sequence alignment with high accuracy and high throughput,” Nucleic Acids Res., 32(5): 1792-7, 2004. [drive5, Mill Valley, CA]||5,333|
|M. Zuker, “Mfold web server for nucleic acid folding and hybridization,” Nucleic Acids Res., 31(13): 3406-15, 2003. [Rensselaer Polytechnic Inst., Troy, NY]||4,400|
|D.L. Donoho, “Compressed sensing,” IEEE Tr. Information Theory, 52(4): 1289-1306, 2006. [Stanford U., CA]||3,102|
|A. Stamatakis, “RAxML-VI-HPC: Maximum likelihood-based phylogenetic analyses with thousands of taxa and mixed models,” Bioinformatics, 22(21): 2688-90, 2006. [Swiss Federal Inst., Lausanne]||2,909|
|D. Posada, “jModelTest: Phylogenetic model averaging,” Molec. Biol. and Evolution, 25(7): 1253-6, 2008. [U. Vigo, Spain]||2,388|
|S. Grimme, “Semiempirical GGA-type density functional constructed with a long-range dispersion correction,” J. Computational Chem., 27(15): 1787-99, 2006. [U. Münster, Germany]||2,372|
|W.J. Kent, “BLAT – The BLAST-like alignment tool,” Genome Res., 12(4): 656-64, 2002. [U. Calif., Santa Cruz]||2,331|
|S.M. Smith, “Fast robust automated brain extraction,” Human Brain Mapping, 17(3): 143-55, 2002. [U. Oxford, UK]||1,905|
|Z.H. Zhang, “PAML 4: Phylogenetic analysis by maximum likelihood,” Molec. Biol. and Evolution, 24(8): 1586-91, 2007. [Univ. College London, UK]||1,761|
|J.D. Storey, “A direct approach to false discovery rates,” J. Royal Statistical Soc. Ser. B, 64: 479-98, 2002. [Stanford U., CA]||1,558|
|SOURCE: Thomson Reuters Web of Science|
Most immediately striking is the top-listed paper, “A short history of SHELX,” a description of a suite of programs for determining molecular structure, written by its principal developer, George M. Sheldrick of the University of Gottingen, Germany. Despite its comparatively recent publication date of 2008, the paper has already been cited more than 30,000 times. According to Essential Science Indicators, this is the most-cited paper published in the last 10 years. Analysis of the Web of Science Research Areas in which the citing papers are categorized demonstrates that, although the SHELX programs have principally found application in Crystallography (nearly 21,000 of the current 32,000 citing papers), the paper is also cited in the journals of Chemistry, Biochemistry/Molecular Biology, Materials Science, and Spectroscopy.
The field of Crystallography, in fact, accounts for the three most-cited reports in the table, with the second- and third-listed papers attracting more than 9,400 and 6,800 citations, respectively.
The remainder of the list is similarly dominated by papers describing tools and methods. “Distinctive image features from scale-invariant keypoints,” for example, with nearly 6,500 citations, falls under Computer Science and discusses a method for image and object recognition. Other papers involve methods for gene sequencing, protein alignment, and other aspects of automated genomic analysis.”Fast robust automated brain extraction,” with roughly 1,900 citations, provides an automated method for segmenting magnetic resonance head images into brain and non-brain tissue, while “A direct approach to false discovery rates,” cited more than 1,500 times, discusses an approach to statistical testing of multiple hypotheses.
Even this small selection of papers attests that, although the proportion of single-authored papers may be endangered in many fields, such reports still carry substantial weight in providing the methods and means for advancing research.
The data and citation records included in this report are from Thomson Reuters Web of ScienceTM. Web of ScienceTM is a registered trademark of Thomson Reuters. All rights reserved.