Data for: Dual-use research under scrutiny
Data files
Jun 04, 2026 version files 60.52 MB
-
dataset.dta
60.50 MB
-
README.md
7.10 KB
-
stata_codes.do
13.70 KB
Abstract
Recent U.S. policies strengthening oversight of dual-use research (DUR) raise questions about their implications for both scientific advancement and national security. Analyzing 600,000 research articles published between 1981 and 2005, we document the prevalence of DUR and its significant scientific importance. The data reveal that an increasing proportion of DUR emerges without U.S. federal government involvement, thus constraining national authorities’ capacity to apply ex-ante oversight mechanisms due to jurisdictional and territorial limitations. These results indicate that enhanced U.S. security oversight may entail costs to cumulative scientific progress by delaying diffusion of scientifically important discoveries, while offering diminishing institutional utility given the evolving, globally distributed DUR landscape. This could in turn compel U.S. security authorities to rely increasingly on costlier ex-post oversight measures.
Dataset DOI: 10.5061/dryad.7wm37pw7p
Description of the data and file structure
README: Data for “Dual-Use Research Under Scrutiny”
Overview: The Stata dataset used for the analysis (dataset.dta) and the Stata code (stata_codes.do) have been submitted to facilitate the reproduction of the findings reported in the article of Kwon (2026, Science) and its supplementary materials.
- data.dta: Main analysis dataset.
- stata_code.do: Stata do-file for reproduction.
Data Sources: The dataset was constructed by combining the following five databases,
- SciSciNet V2 (https://github.com/kellogg-cssi/SciSciNet) [SciSciNet]
- Lin, Z., Yin, Y., Liu, L., & Wang, D. (2023). SciSciNet: A large-scale open data lake for the science of science research. Scientific Data, 10(1), 315.
- Clarivate’s Web of Science Core Collection (https://www.webofscience.com/wos/woscc/) [WOS]
- WoS-derived variables are categorizations computed by the author; no raw Web of Science data is redistributed.
- Patent-Citation-to-Science data ( https://relianceonscience.org/patent-to-paper-citations) [PCS]
- Marx, M., & Fuegi, A. (2020). Reliance on science: Worldwide front‐page patent citations to scientific articles. Strategic Management Journal, 41(9), 1572-1594.
- Marx, M., & Fuegi, A. (2022). Reliance on science by inventors: Hybrid extraction of in‐text patent‐to‐article citations. Journal of Economics & Management Strategy, 31(2), 369-392.
- USPTO patent examination dataset released in 2022 (https://www.uspto.gov/ip-policy/economic-research/research-datasets/patent-examination-research-dataset-public-pair) [PatEx]
- Other USPTO patent metadata (patentsview.org) [PatView]
Citation: Please cite the following article when using this data: Kwon, S. Dataset: Dual-Use Research Under Scrutiny. Dryad.
Description of Data and File Structure
Variables in data.dta (Origin of the variable is indicated in brackets [ ])
- paperid: OpenAlex ID of the paper [PCS & SciSciNet]
- year: Publication year of the paper [SciSciNet]
- c5: 5-year window forward citation count [SciSciNet]
- c10: 10-year window forward citation count [SciSciNet]
- hit1_5pct: Dummy variable taking the value of 1 if the paper’s field-year normalized citation count is in the top 5% of the distribution [SciSciNet]
- cf1: Field-year normalized paper citation count [SciSciNet]
- ncite_patent: Total number of patent citations received [PCS & PatView]
- ncite_patent15: Number of patent citations received within 15 years of publication [PCS & PatEX & SciSciNet]
- dual: Dummy variable taking the value of 1 for Dual-Use Research (DUR) papers, 0 otherwise [PCS & PatEx]
- dual_secret: Dummy variable taking the value of 1 for DUR papers where the citing sensitive patent was “once under secrecy order”, 0 otherwise [PCS & PatEx]
- sponly: Dummy variable taking the value of 1 for papers cited only by sensitive patents, 0 otherwise [PCS & PatEx]
- nsponly: Dummy variable taking the value of 1 for papers cited only by non-sensitive patents, 0 otherwise [PCS & PatEx]
- is_usfed_funded: Dummy variable taking the value of 1 for papers acknowledging funding from U.S. federal government organizations [WOS]
- is_foreign_funded: Dummy variable taking the value of 1 for papers acknowledging funding from non-U.S. government organizations [WOS]
- ncite_patent_in: Total number of in-text patent citations received [PCS & PatView]
- ncite_patent15_in: Number of in-text patent citations received within 15 years of publication [PCS & PatEx & SciSciNet]
- is_author_us_fed: Dummy variable taking the value of 1 for papers with author(s) affiliated with U.S. federal organizations [WOS]
- is_author_foreign: Dummy variable taking the value of 1 for papers with author(s) affiliated with non-U.S. organizations [WOS]
- is_foreign_involve: Dummy variable taking the value of 1 for papers acknowledging non-U.S. funding or with authors affiliated with non-U.S. organizations [WOS]
- is_usfed_included: Dummy variable taking the value of 1 for papers with either is_usfed_funded = 1 or is_author_us_fed = 1 [WOS]
- is_anyus_included: Dummy variable taking the value of 1 for papers with is_usfed_included = 1, acknowledging any funding from U.S. institutions, or having author(s) affiliated with any U.S. organizations [WOS]
- is_foreign_only: Dummy variable taking the value of 1 for papers with is_anyus_included = 0 and is_foreign_involve = 1 [WOS]
- time2sens: Duration between the publication date and the first citation from a sensitive patent (days) [SciSciNet & PatView]
- time2nosens: Duration between the publication date and the first citation from a non-sensitive patent (days) [SciSciNet & PatView]
- dur15: Dummy variable taking the value of 1 for DUR papers defined using only a 15-year window of PCS information, 0 otherwise [PCS & PatEx & SciSciNet]
- sponly15: Dummy variable taking the value of 1 for papers cited only by sensitive patents using only a 15-year window of PCS information, 0 otherwise [PCS & PatEx & SciSciNet]
- nsponly15: Dummy variable taking the value of 1 for papers cited only by non-sensitive patents using only a 15-year window of PCS information, 0 otherwise [PCS & PatEx & SciSciNet]
- field_: Macro-level research field [SciSciNet]
- t: Time trend variable (years counting from 1981) [SciSciNet]
- paper_type: String variable indicating category of paper: “dur_secret” (DUR cited by secrecy order patents), “dur_nosecret” (other DUR), or “nondur” [PCS & PatEx]
- dual_intxt: Dummy variable taking the value of 1 for papers categorized as DUR using only in-text patent-paper citation information [PCS & PatEx]
- dur15_intxt: Dummy variable taking the value of 1 for papers categorized as DUR using only a 15-year window of in-text patent-paper citation information [PCS & PatEx & PatView]
- funding_ack_exist: Dummy variable taking the value of 1 for papers acknowledging funding support as recorded in the WoS Core Collection [WOS]
- dualxusfed: Interaction term between dual and is_usfed_included [PCS & PatEx & WOS]
- field_subfield: Subfield (SciSci Field) [SciSciNet]
Citation
You are free to use, modify, and redistribute it without permission. However, I strongly request that anyone using this dataset cite the dataset itself and the source databases listed above, in keeping with academic norms for data reuse:
-
Kwon, S. Dataset: Dual-Use Research Under Scrutiny. Dryad. https://doi.org/10.5061/dryad.7wm37pw7p
-
(plus the source references listed under Data Sources)
Code/Software
- Stata code: stata_code.do (Stata do-file)