Abstract
A large number of published studies have examined the properties of either networks of citation among scientific papers or networks of coauthorship among scientists. Here, using an extensive data set covering more than a century of physics papers published in the Physical Review, we study a hybrid coauthorship/citation network that combines the two, which we analyze to gain insight into the correlations and interactions between authorship and citation. Among other things, we investigate the extent to which individuals tend to cite themselves or their collaborators more than others, the extent to which they cite themselves or their collaborators more quickly after publication, and the extent to which they tend to return the favor of a citation from another scientist.
Introduction
In the field of scientometrics, researchers typically analyze two distinct structures: citation networks (how papers link to one another) and coauthorship networks (how scientists collaborate). While both provide valuable insights into academic habits, they are rarely studied in tandem. This study addresses this gap by merging these networks into a hybrid model to explore the social mechanics behind scientific credit and knowledge distribution. By analyzing over 116 years of data from the American Physical Society (APS), the research tracks the transition from "Little Science"—characterized by individual researchers—to the era of "Big Science" involving massive international collaborations.
Methodology
The researchers employed a longitudinal network analysis spanning from 1893 to 2009, covering 462,090 papers and 235,533 unique authors. Key methodological steps included:
- Hybrid Network Construction: Treating authors and papers as interconnected nodes to track how social links (coauthorship) influence informational links (citation).
- Author Disambiguation: Developing a multi-stage process using institutional affiliations and coauthor overlaps to distinguish between authors with identical names.
- Statistical Modeling: Utilizing Lorenz curves and Gini coefficients to quantify productivity inequality among scientists.
- Time-Resolved Analysis: Breaking the data into five-year blocks to observe the evolution of publishing patterns over a century.
Key Results and Insights
Productivity Inequality and the 80-20 Rule
Scientific output demonstrates significant concentration. The study found that scientific productivity follows Lotka’s Law, where a small fraction of authors produces the majority of the work. Specifically, the top 20% of the most prolific authors are responsible for roughly 80% of the total papers. This is reflected in a Gini coefficient for productivity of 0.70, indicating a level of inequality higher than the wealth distribution in most nations.
The Social Dynamics of Citation
Despite the exponential growth of scientific literature—with the number of papers doubling approximately every 12 years—certain human behaviors have remained remarkably constant:
- The 20% Constant: The fraction of self-citations has remained steady at roughly 20% for over a century, suggesting that building on one's own work is a fundamental feature of the scientific process.
- Speed of Citation: Authors cite themselves and their collaborators significantly faster than they cite distant peers. The mean delay for a self-citation is 4.12 years, compared to 9.02 years for a citation from a stranger.
- Reciprocity: There is a strong "tit-for-tat" mechanism. If Author A cites Author B, there is a 13.5% chance B will eventually cite A. This probability jumps to 43.8% if they have previously coauthored a paper.
The Rise of "Big Science"
Around the mid-1970s, the data reveals a dramatic divergence in authorship patterns. While the average number of authors per paper remained relatively low for most subfields, experimental high-energy physics began producing papers with hundreds or even thousands of authors. This shift marks the emergence of massive international collaborations, which significantly skews traditional metrics of individual productivity.
Limitations and Future Directions
The primary limitation of this study is its reliance on the "APS Silo"; citations to or from journals outside the Physical Review family (such as Nature or Science) were not captured, likely underestimating the frequency of "distant" citations. Additionally, the author disambiguation process has a known error rate (3% false positives and 12% false negatives). Future research could incorporate geographical and institutional data to map how scientific ideas travel across borders and whether "citation favors" are more prevalent within specific regions or organizations.
Summary updated