ICSR PERSPECTIVES ICSR Fractional Authorship & Publication Productivity Highlights Authors divide their research output across publications, contributing via research collaborations The trend is for authors to produce more publications per year (increased fractionalization) but for the overall number of publications per author to decrease We suggest that the effort required to participate in research collaborations is a factor in the decrease in publications per author AUGUST 2019 AUGUST Are authors collaborating more in response to the pressure to publish? Growth in the number The “publish or perish” research reasons; for instance, to gain access of scholarly publications culture provides incentives for to samples, field sites, research each year has been well researchers to have long publication facilities, or patient groups. lists on their CVs, especially where Researchers wishing to study topics documented (e.g., Bornmann those publications appear in high- outside their own expertise require & Mutz, 2015, Figure 1). But impact journals (Tregoning, 2018). interdisciplinary collaborators or may how has that growth been By examining authorship trends, we simply look to find co-authors whose achieved? Is it purely due aim to understand if researchers are skills and knowledge complement to increasing investment in responding to the pressure to publish their own. Evidence suggests that research, resulting in a greater by fractionalizing themselves across diverse research teams are more more papers and whether this leads likely to be successful at problem number of active researchers? to more publication outputs overall. solving (e.g., Phillips, Northcraft, & Or is each researcher Does increasing collaboration enable Neale, 2006) and that publications producing more publications? each researcher to be involved with, by collaborative teams benefit from To investigate these questions, and produce more, research? a citation advantage (e.g., Glanzel, we build on Plume & van 2001). International collaboration has Researchers are motivated to also been shown to drive publication Weijen’s (2014) previous work. enter into collaborations for many growth (Adams, 2013). authorships authors CAGR: .% publications CAGR: .% CAGR: .% Figure 1: Trends of publications, unique authors, and authorships per publication year 2008–17 (millions). Compound Annual Growth Rate (CAGR) is shown for the same period. 2 Growth in publications and authors is PERSPECTIVES ICSR outstripped by growth in authorships Using Scopus, we examined under five authors, compared to publications per author declined. This authors contributing to just under four authors in 2008, indicates that, in terms of research documents published between suggesting that authors are writing efficiency (as measured by publication more collaboratively. Some of this output), authors are collaborating 2008 and 2017. Collaborative change is attributable to the rise of more and increasing their personal writing can be measured in papers with very long author lists. productivity, but the net output per terms of co-authorships—the In 2017, 1,249 papers had more than active researcher has decreased. number of authors listed as 100 authors, compared to 526 in We see the same trend across all 27 contributing to a document. 2008. Many papers with the longest subject areas studied (not shown, see This contrasts with the count lists of contributors are the result of supplementary data); therefore, we large research collaborations, such don’t believe that the trends can be of unique authors, where each as the Large Hadron Collider (e.g., attributed to any possible changes author is only counted once Aad et al., 2015) or the Reduction in the balance of content indexed in per year in which they appear of Atherothrombosis for Continued Scopus across different subject areas listed on any publication. Health (REACH) Registry (Eisen et over time. Figure 1 shows that the number al., 2016). This interpretation assumes of authorships has grown at a Publications have more authors that authors are included on much faster rate than both the contributing on average, but how publications where they have made growth of publications and of does that look for each researcher? a genuine contribution and not as unique authors. Are individual researchers writing “gift” authorship (where authors more articles every year, or are more are included despite little or no Perhaps the most straightforward authors just writing collaboratively? participation) or “ghost” authorship measure of author behavior is the (where authors are not included The number of authorships per number of authors who contributed despite making a significant author remained very stable from to each publication (authorships per contribution), and that the rates 2008-17, meaning that each author publication). Over this decade, the of gifting and ghosting have not contributed to a similar number average number of authors on each substantially changed over time. publication increased (see Figure 2), of papers per year; however, the which is consistent with previous number of publications per author Not all fields have the same studies (Mallapaty 2018; Plume & van (total publications divided by total authorship patterns. For example, Weijen 2014). The rise in the number unique authors) decreased from the prevalence of single-authored of authorships per publication over 0.57 in 2008 to 0.48 in 2017. Whilst papers varies by field, constituting the past ten years means that, on collaboration (authorships per the greatest share of Arts and average, a publication now has just publication) rose, the number of Humanities publications (66% in Publications have more authors contributing on average, but how does that look for each researcher? Are individual researchers writing more articles every year, or are more authors just writing collaboratively? 3 2017) and the least in Immunology and energy required to work with Our results imply that full and and Microbiology publications (3% a team (and the effort to get each fractional counting approaches in 2017). The trend of decreasing document published) means that offer complementary perspectives publications per author while each researcher will produce fewer on publication output, as they each authorships per publication increases publications overall. So, while authors expose different aspects of research is seen across all fields. are responding to the pressure to practice. Furthermore, it highlights publish by fractionalizing themselves the need to employ both metrics to shows that fields with more Figure 3 across more publications, they fully understand how publication collaborative authors produce fewer become less productive (in terms of output reflects productivity. publications per author. This is the number of publications), though consistent with the trends observed Given the trend of increasing author this does not impact all fields equally. across all publications, where authors lists on publications, it has become This may be of more concern to collaborate more but contribute to more difficult to discern author research institutions or funders who slightly fewer publications in total. contribution from that list alone. This look at total research outputs counted emphasizes the need for improved This finding suggests that there is a across groups of researchers, rather credit assignment, particularly for the productivity penalty to collaboration than individual researchers who are purposes of evaluating productivity —while collaborating allows authors successfully using this strategy to and research leadership. to appear on more articles, the time grow their own publication lists. authorships per publication CAGR: .% authorships per author publications per author CAGR: . % CAGR: . % Figure 2: Trends of authorships per publication and per author and publications per author 2008–17. 4 ICSR PERSPECTIVES ICSR PUBLICATIONS PER More publications per unique AUTHOR ­ author, shorter author lists . Arts & Humanities . Economics, Econometrics & Finance Social Sciences Mathematics . Business, Management Computer Science & Accounting Engineering Psychology . Decision Sciences Materials Physics Energy Earth & Planetary Sciences Chemistry Environmental Science Medicine . Dentistry Nursing Agricultural & Biological Sciences Health Professions Veterinary Biochemistry, Genetics & Molecular Biology Chemical Engineering . Pharmacology, Toxicology Immunology & Microbiology & Pharmaceutics Neuroscience .­ Fewer publications per unique Multidisciplinary author, longer author lists AUTHORSHIPS PER . PUBLICATION ­ ­ Figure 3: Scatter plot of authorships per publication vs. publications per author in 2017 per All Science Journal Classication (ASJC) subject area. Our results imply that full and fractional counting approaches offer complementary perspectives on publication output, as they each expose different aspects of research practice. 5 Method & Data Sources This report uses bibliometric data from Scopus. Scopus is Elsevier’s abstract and citation database of peer-reviewed literature covering 75 million documents published in over 23,500 journals, book series, and conference proceedings. Data for this report was accessed in April 2019. Document types selected for the analysis were: journal articles, journal reviews, and conference proceedings. For each document in the analysis, the count of authorships is the total number of authors listed. Within Scopus, articles are assigned