Source-linked AI summary
Team size matters: Collaboration and scientific impact since 1900
Vincent Lariviere, Cassidy Sugimoto, Andrew Tsou, Yves Gingras
TL;DR
The paper addresses the lack of a historical analysis of how collaboration relates to scientific impact and examines this relationship across 1900–2011. Using millions of papers and citations, it finds that greater collaboration is associated with higher impact beyond self-citations, although increasingly large teams are needed for comparable citation gains.
Problem
Existing studies linked collaboration with scientific impact, but none had examined these relationships from a historical standpoint or assessed self-citations across the period.
Method
The study analyzes 1900–2011 citation data using author counts, institutional addresses, and countries as collaboration indicators, with field- and year-normalized impact measures.
Results
Excluding self-citations, scientific impact increased with the numbers of authors, addresses, and countries, showing that self-citation did not fully explain collaboration’s citation advantage.
Takeaways & Limitations
As collaboration expanded, progressively larger teams were needed to achieve comparable citation impact, indicating diminishing citation returns over time.
Takeaways & Limitations
Institutional-address data were missing for approximately 30% of papers in 1975 and 8% later, limiting the completeness of that collaboration measure.
Abstract
from arXiv · showhide
This paper provides the first historical analysis of the relationship between collaboration and scientific impact, using three indicators of collaboration (number of authors, number of addresses, and number of countries) and including articles published between 1900 and 2011. The results demonstrate that an increase in the number of authors leads to an increase in impact--from the beginning of the last century onwards--and that this is not simply due to self-citations. A similar trend is also observed for the number of addresses and number of countries represented in the byline of an article. However, the constant inflation of collaboration since 1900 has resulted in diminishing citation returns: larger and more diverse (in terms of institutional and country affiliation) teams are necessary to realize higher impact. The paper concludes with a discussion of the potential causes of the impact gain in citations of collaborative papers.
Introduction
Scientific research has never been strictly individual, while increasing specialization has driven the rise of large collaborative teams. This paper addresses the lack of a historical analysis linking collaboration and scientific impact using 32.5 million papers and 515 million citations from 1900–2011.
- Motivation: Scientific research has never been a strictly individual enterprise, despite the enduring “lone genius” myth.Even Einstein collaborated on several papers.
- Background: Increasing specialization has accompanied the growth of large, well-funded collaborative teams and the decline of single-authored papers.The passage also notes a rise in international collaboration alongside the number of authors.
- Contribution: The paper fills a historical research gap by analyzing collaboration and scientific impact across 32.5 million papers and 515 million citations published or received over 1900–2011.Previous studies had connected impact with collaboration but had not examined these relationships historically.
Background
Increasing co-authorship has been a longstanding trend and subject of scientific study. Recent research confirms that co-authorship is becoming more common across all disciplines, consistent with Price’s 1963 prediction of increasingly many authors per paper.
- Background: Increasing co-authorship is a longstanding trend and has been studied since at least the 1960s.The passage cites Hagstrom’s 1965 study as an example of earlier work on collaboration in science.
- Background: In 1963, Price predicted that scholarly publications would move steadily toward an infinity of authors per paper.This prediction appears on page 89 of Price’s work, as quoted in the passage.
- Background: Recent studies confirm that co-authorship is becoming increasingly common across all disciplines.The passage attributes this finding to multiple studies published from 2003 through 2010.
The collaboration advantage
Collaboration is presented as a response to increasingly specialized research and resource shortages, bringing complementary skills together to solve problems. It is also positively associated with academic quality, including higher productivity that increases with team size.
- Collaboration suits increasingly narrow research focuses and may address shortages of necessary resources.Increasing specialization alone may not fully explain collaboration’s growth.
- Different and ideally complementary skills can be integrated through collaboration to solve a single problem.The integration is not always seamless and can cause friction.
- Collaboration is positively correlated with multiple academic-quality metrics, including productivity.
- Productivity increases as team size increases.
The geography of collaboration
Geographic proximity remains associated with stronger collaboration intensity despite easier communication. Cross-institutional and international collaboration can confer citation advantages, but these effects are not universal and have prompted initiatives to encourage geographically broader partnerships.
- The geography of collaboration: Collaboration intensity is inversely proportional to the distance between scientists.This proximity effect persists despite the falling cost and growing ease of communication among scientists.
- The geography of collaboration: Institutional and governmental initiatives have been introduced to incentivize collaboration across geographic boundaries.
- The geography of collaboration: Cross-institutional and international co-authorship has shown a citation advantage, but the effect is not universal.International collaboration may be more common in basic fields, with geography and other extra-scientific factors potentially shaping its prevalence.
Self-citations
The citation advantage of co-authored works has been challenged as a possible amplification of self-citation, which increases with the number of co-authors. Existing evidence is limited by disciplinary and temporal scope, motivating broader research on self-citation and impact.
- Self-citations: The citation advantage of co-authored works may reflect amplified self-citation, because self-citations increase with the number of co-authors.This challenge treats co-authored papers as receiving multiplied citations when each author self-cites similarly to a single author.
- Self-citations: Studies of collaborative citation impact often lack generalizability because they use small samples, focus on particular disciplines, or cover limited time periods.The passages call for large-scale research across time and disciplines to assess how consistently citation advantages prevail.
- Self-citations: Further research is needed to examine how self-citation interacts with impact across differences in subspecialty, researcher age, and gender.The passage reports evidence of self-citation differences by subspecialty and age, and to a lesser extent by gender.
Methods
The study analyzes publications and citations from major Thomson Scientific indexes spanning 1900–2011, with citation data normalized by publication year and specialty. It examines collaboration through authors, institutional addresses, and countries, alongside citation impact and self-citation effects.
- Data and measurement: The dataset comprises 28,160,453 NMS papers with 484,393,178 citations and 4,347,229 SSH papers with 30,587,347 citations from 1900–2011.Sources include SCIE, SSCI, and AHCI.
- Data and measurement: Citation counts run from publication year through 2011, while impact is reported through 2009 to ensure at least a two-year citation window.Citation data are normalized to the average for papers published in the same year and specialty.
- Evolution of collaboration: Single-author, single-address, and single-country papers declined over time in both natural and medical sciences and social sciences and humanities.Collaboration was grouped into classes by numbers of authors, institutional addresses, and countries.
- Evolution of collaboration: In 2011, papers with 4–5 and 6–10 authors represented about 29% and 27% of NMS papers, respectively.These shares reflect the increasing prevalence of larger author teams.
- Evolution of collaboration: Single-address papers fell from 70% in 1973 to slightly over 30% in 2011 in NMS and from 70% to 46% in SSH.In NMS, papers with three or more addresses continued increasing while the share with two addresses stabilized.
- Collaboration and impact: Impact increased with the numbers of authors, addresses, and countries, consistently across time and research areas, although higher impact increasingly required larger teams.For author counts, NMS impact rose until about 45 authors before flattening and oscillating, while remaining above average.
Discussion and conclusion
Since 1900, co-authorship and institutional and international collaboration increased in both NMS and SSH, while collaborative papers achieved higher citation impact. Impact gains remained after excluding self-citations, although citation returns diminished as contributor numbers rose.
- Collaboration trends: Single-authored papers fell from 87% in 1900 to 7% in 2011 in NMS and from 97% to 38% in SSH.These changes accompanied increasing co-authorship, inter-institutional collaboration, and international collaboration in both fields.
- Citation impact: Scientific impact excluding self-citations increased with the number of authors, addresses, and countries, at least during 2005-2009.This indicates that the citation advantage of collaboration is not explained solely by self-citation.
- Citation impact: Self-citation gains increased with author count but flattened at 10-12 authors in NMS and 25 authors in SSH.The self-citation gain was higher in SSH than in NMS.
- Citation impact: Collaborative papers were more heavily cited than sole-authored papers as early as 1900, but increasing contributor counts raised the number of authors needed to achieve a given citation level.The paper extends prior work by covering the first 50 years of the twentieth century.
- Possible mechanisms: Collaborative research may gain citations through greater epistemic value, specialized equipment, large-scale genomics, interdisciplinary problem-solving, and greater consensus.The passage presents these as possible explanations rather than a single established cause.