The selective use of physics knowledge in policy: how interdisciplinary physics bridges subfields and shapes policy influence

TL;DR

The paper investigates how physics knowledge enters policy and why interdisciplinarity matters. It introduces a linked Overton–APS dataset and a suite of methods (PACS-based subfield classification, fractional weighting, topic modeling, backbone extraction, community detection, and regression) to separate policy visibility from downstream influence. Key findings show policy demand centers on General and Interdisciplinary Physics, with interdisciplinary areas serving as structural brokers, yet visibility does not guarantee policy impact; Geophysics, Astronomy, and Astrophysics can yield higher downstream influence via synthesis reports in climate governance. The work highlights a fundamental mismatch between scientific visibility and policy influence and offers a framework for more nuanced assessment of science-to-policy translation with implications for how researchers and institutions communicate and package knowledge for policy audiences.

Abstract

Scientific knowledge has become central to policymaking as societies face challenges related to technological change, climate risk, and public health. Despite the growing emphasis on evidence-based policy, a systematic understanding of how science is selectively used in policy, specifically which forms of knowledge are preferred and which scientific citations translate into influence, remains limited. We address these questions by constructing a novel dataset that links policy documents from the Overton database with publications from the American Physical Society, enabling an analysis of how physics knowledge enters and circulates in policy discourse. Using subfield classifications, we provide quantitative evidence for a gap between scientific communities and policymakers. First, we find that policy documents draw on broad and interdisciplinary areas of physics, such as General Physics and Interdisciplinary Physics, rather than mirroring the structure of physics research production. Second, we identify substantial institutional heterogeneity with systematic differences in subfield preferences across policy producing organizations and topics. Third, network analysis reveals that interdisciplinary areas of physics act as a central bridge connecting specialized subfields. Finally, regression analysis reveals a clear separation between policy visibility and policy influence. While interdisciplinary areas facilitate entry into policy discourse, it does not necessarily increase downstream policy influence. Conversely, documents citing geophysics are associated with approximately 24 percent higher policy influence, likely driven by the political salience of climate change policy. Our findings underscore the distinction between scientific visibility and policy influence, contributing to a deeper understanding of the complex relationship between scientific communities and policy system.

Figures (14)

• Figure 1: Schematic of the data linkage and classification pipeline. Policy documents from the Overton database (left) are linked to scientific articles in the American Physical Society corpus (center) via Digital Object Identifiers (DOIs). This linkage captures the structure of policy demand, where a single policy document may reference multiple scientific papers. Each cited paper is then characterized by its scientific content (right) using the first-level categories of the Physics and Astronomy Classification Scheme (PACS 0–9). Interdisciplinary areas of physics—specifically Category 0: General Physics and Category 8 :Interdisciplinary Physics and Related Areas of Science and Technology—are highlighted in bold.
• Figure 2: Distribution of scientific supply versus policy demand. The blue bars represent the full APS corpus (scientific supply), while the red bars denote the subset of papers cited in policy documents (policy demand). Frequencies are fractionally weighted to account for papers assigned to multiple PACS categories. The vertical dotted line marks the expected proportion under a uniform null model, representing the baseline where all ten categories are equally represented. The label for Interdisciplinary areas of physics is highlighted in bold. While Category 7: Condensed Matter Physics dominates overall academic production, policy documents exhibit a substantial preference for Interdisciplinary areas of physics, indicating a structural divergence between scientific supply and policy demand.
• Figure 3: Institutional heterogeneity in the use of physics subfields. The heatmap displays citation intensity normalized by the number of cited papers and assigned PACS codes per policy document (See Materials and Methods). The x-axis represents institutional affiliations, displaying broad sectors (Public Sector vs. Think Tanks) on the left and a detailed breakdown of public sector entities on the right. The y-axis lists the first-level PACS code categories. The color scale indicates relative prominence: a value of 0.1 (white) represents the expected average distribution, while red color denotes higher-than-average citation intensity and blue color indicates lower-than-average citation intensity. Think Tanks exhibit a distinct preference for interdisciplinary areas of physics, whereas government documents show a relative under-representation of these interdisciplinary domains.
• Figure 4: Thematic preferences for physics subfield in policy documents. The heatmap displays citation probabilities across six policy topics identified via LDA analysis: Global Security, Complex Systems, Energy and Climate, Industrial Finance, Global Economy, and Health and Nuclear. The x-axis lists first-level PACS code categories, while the y-axis represents the extracted topics. Each cell represents the probability that a given topic cites a specific first-level PACS category, where a value of 0.1 (white) corresponds to the expected baseline and red hues indicate higher citation intensity. Distinct disciplinary signatures emerge: Energy and Climate, a domain heavily regulated by government agencies, draws primarily on Condensed Matter Physics. In contrast, Complex Systems, a topic frequently associated with think tanks, relies predominantly on interdisciplinary areas of physics. These thematic distinctions mirror the institutional preference patterns observed in Fig. \ref{['fig:Institution']}.
• Figure 5: Co-occurrence network of two-digit PACS codes in policy documents. Nodes represent physics subfields, and links connect subfields cited together within the same policy document. Node size reflects the log-scaled frequency, while link thickness indicates co-occurrence intensity. To retain systematic structural patterns, the network is filtered using the disparity filter ($\alpha = 0.1$). We identify five distinct communities corresponding to major disciplinary clusters: Interdisciplinary Areas, Condensed Matter, Atomic, Molecular, & Optical Physics, Particle & Nuclear Physics, and Earth & Planetary Sciences. The topology reveals a hub-and-spoke structure where the Interdisciplinary areas of physics communities (containing central nodes like 05 and 89) (see Table.S17)form the integrative backbone, bridging the more specialized technical clusters.