Collaborative Research: Revisiting the Dynamics of Scientific Influence: Measures and Applications

Universities, states, and the federal government invest millions of dollars in scientific research, and are eager to assess its impact. Researchers who study science are also interested in explaining how and why some research is particularly influential, and how scientific research evolves over time. Meeting these goals requires detailed yet scalable measures of scientific influence. Building on recent developments in network science, this project will develop and evaluate a novel measure of scientific influence -- the CD index. The CD index characterizes how scientific works (e.g., published journal articles) alter the use of the knowledge upon which they build. Some articles build on and therefore consolidate (C) previously published research, whereas other articles depart from and therefore destabilize (D) existing knowledge. This measure offers a new, systematic way of identifying breakthrough scientific discoveries that open up novel fields, and differentiating them from more incremental, but also important, contributions that develop established areas. Results from this project should be useful to individuals involved in research policy development and implementation. For example, university administrators are moving away from a focus on sheer quantity of scientific output toward a greater appreciation for quality and impact. Results from this project will allow administrators to not only identify research that is influential in different (consolidating or destabilizing) ways, but to foster the conditions (e.g., an optimal team composition, or interdisciplinary approaches to scholarship) that promote such output.

Prominent theories in the economics, philosophy, and sociology of science suggest that scientific influence is a complex and multidimensional phenomenon, with different types of works potentially having different types of effects on the course of science. However, despite growing availability of data and computing power, efforts to incorporate these important theoretical insights into empirical work have been scarce. Most large-scale studies rely on quantitative bibliometrics, using journal impact factors and citation counts to proxy the amount of scientific influence exerted by a given paper or patent. But such proxies are crude and are unable to distinguish between qualitatively different types of influence. Methodologically, the measurement approach developed and evaluated in this project will more closely align empirical and theoretical work on scientific influence. Substantively, the project argues that better, more nuanced measures of scientific influence will help provide clearer and more robust answers to pressing science policy questions. To date, much of the scholarship on scientific influence is quite conflicted. Are women or women dominant teams more likely to produce influential science? Do interdisciplinary efforts facilitate transformative, influential research? The literature is rife with contradictory answers to these important questions. Applying the novel CD index to two entire corpora (JSTOR and the Web of Science)-- as this project will do -- should help resolve them.

This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.