Global Institutional Responses to Generative AI in Higher Education

This document contains the data collected during the manuscript's data collection process, entitled “Global Institutional Responses to Generative AI in Higher Education.” The initial synthesized data are provided in the data_pre.xlsx file. The data produced by Validator 1 and Validator 2 are provided in the data_v1.xlsx and data_v2.xlsx files, respectively. The cross-check data are provided in the data_cek.xlsx. The final extracted dataset is provided in the data_final.xlsx file. Dataset overview.This dataset documents institutional responses to Generative AI (GenAI) in higher education, with a focus on assessment-related governance and policy architecture. It contains structured codes extracted from publicly available, official institutional documents (e.g., policy PDFs, webpages/FAQ, academic integrity hubs). The dataset supports analyses of how institutions articulate rules for GenAI use in assessment, disclosure expectations, evidence/process requirements, detector positioning, and broader governance coverage (privacy, IP/copyright, tool/vendor approval, training/support, and review/update mechanisms). Coverage. Units of analysis: 40 higher-education institutions Regional sampling: 5 regions (coded C1–C5), 8 institutions per region (balanced) Source type: official URLs to the institutional document(s) used for coding File structure (Excel).The dataset is provided as an Excel workbook (data_pre.xlsx) with two sheets: Sheet1 — Document-level metadata + full field coding (24 columns).Includes region, country, institution, document title/type, issuing authority, year/last updated (as stated), architecture class & depth, assessment model, default stance on AI use, disclosure rules, process-evidence requirements, detector positioning, governance coverage score, five governance components (binary), official links, and short notes quoting explicit evidence. Sheet2 — Compact coded table for analysis (19 columns).A cleaned/condensed view of key analytic variables used for statistical summaries and visualization: ArC (architecture class; categorical) ArD (architecture depth; 0–3) AsM (assessment model; categorical) DeS (default stance; categorical) DiR (disclosure rule; categorical) PrE (process-evidence requirement; categorical) DeP (detector positioning; categorical) NRu (normative rules explicitly stated; Y/N) EMe (implementation/enforcement mentioned; Y/N) PrM, IcA, VeA, TrS, ReM (governance components; Y/N) GcS (governance coverage score; 0–5) AsR (assessment regime label; A–E), a categorical summary used for grouping institutional stances. Key distributions (Sheet2). GcS (0–5): 0(2), 1(7), 2(8), 3(10), 4(10), 5(3) — n = 40 AsR (A–E): A(6), B(13), C(7), D(9), E(5) — n = 40 Reuse notes/limitations.This dataset represents a time-stamped snapshot of institutional documents; policies may change after data collection. “Year issued/last updated” is recorded as stated on the source and may be missing or qualitative in some cases. Coding is based on explicit textual evidence from the official document(s) linked in the dataset. Related publication.Global Institutional Responses to Generative AI in Higher Education (manuscript uses this dataset).