Who admits the network? Regulatory openness and the reach of diaspora science diplomacy into military higher education in wartime Ukraine — replication package (v1.4)

Data, code, figures and supplementary material accompanying a manuscript under double-anonymised review at Science and Public Policy. The deposit is anonymised; author information will be added on acceptance. The study asks whether network-borne science diplomacy — the mechanisms of the Ukrainian Science Diaspora initiative — can reach Ukraine's military higher education, and locates the constraint in regulatory openness (θ): the share of a sector's regulatory acts that admit external academic contact. It combines deductive coding of 34 published interviews (221 question–answer units), a content analysis of 81 regulatory acts on military education, a bibliometric measure of revealed openness (OpenAlex, 2019–2025) and a scenario index (MEDRI). Contents (55 files, all in English):• Regulatory corpus: 81 acts with adjudicated coding (θ = 0.37), the authors' original codes, the working matrix (xlsx), the second coding of the NATO-integration category with clause-level justifications, and five institutional acts of the flagship defence university.• Interview corpus: codebook, machine-assisted coding of all 221 units with the authors' coding of a 45-unit subsample, intra-coder stability, a second blind machine pass (K2) and a third blind pass by an independent model family (Gemini 3.6 Flash) with prompts and agreement statistics; two exploratory passes for the EU 2025 category "diplomacy in science".• Model: MEDRI parameters and script (reproduces every number in the article), Sobol sensitivity analysis, alternative aggregations, θ by legal tier and by regulatory vertical, mechanism-redesign scenarios.• Revealed openness: OpenAlex query script and institution × year counts for eight military institutions.• De-identified snapshot of the platform's public directory (464 profiles).• Figures 1–5 and S3–S5 (PNG, SVG); Supplementary Data S0–S7 (docx).• Quotation register with page references, verified against the published corpus. Version 1.4 (3 November 2026) supersedes v1.3: θ disaggregated by tier and vertical (S7); mechanism-redesign scenarios; third machine pass and D4 passes; Supplementary Data S6.2 now reports agreement on the deposited pass-2 codes (κ = 0.30/0.40) alongside agreement under the rule as adopted (κ = 0.45/0.60); Table S7c reports partial derivatives of the index with respect to the dimension weights; the platform directory column oblast_origin re-cleaned (72 free-text entries removed); all materials translated into English. See README.md and CHANGELOG_v1.4.txt. The interview text is not redistributed: the deposit contains codes, page references and short evidence fragments only (source: Kostiuchenko, Maslenko & Shylo, Eds., 2025, 34 Snapshots: Science Diaspora for Ukraine). Platform data were de-identified before deposit. Generative AI was used as an additional coder under the conditions stated in Supplementary Data S6; all analytical decisions and adjudicated codes are the authors'. Licence: CC BY 4.0. Please cite the article and this record.