State forms dataset: integrated, imperial, civilizational and nation-states, 2026.
Authors/Creators
Description
This record contains the dataset and analysis code for the article Beyond nation-states: a global taxonomy of integrated, imperial and civilizational state forms.
The dataset covers the 193 member states of the United Nations and its two observer states, the Holy See and the State of Palestine. Each state is scored on seven dimensions: entrenched (legally protected) substate autonomy, shared rule at the centre, equality of citizenship, exit, civilizational legitimation, centre–periphery hierarchy and cultural plurality. The scores are coded from constitutions, citizenship law, treaties, court rulings and official doctrine, and compared with comparative datasets. Fixed rules assign each state to one of four forms: imperial, civilizational, integrated, or nation-state, the default when none of the other three conditions is met. The Holy See is recorded but not classified. Every score that departs from the default carries its documentary basis.
The dataset also records membership of regional unions and defence alliances, external coercion in 2024–26 and limited territorial control. It includes the external indicators used for comparison: V-Dem v16, the Regional Authority Index, the fractionalization indices of Alesina et al. and Fearon, Ethnic Power Relations 2021, Freedom House and UN World Population Prospects 2024. Scores describe the situation on 1 November 2026.
Contents:
- data/state_forms_dataset.xlsx: the dataset (sheets Units, Coding_notes, Codebook and Sources), with CSV copies of the Units and Coding_notes sheets.
- code/classify.py: recomputes every classification from the scores, including the sensitivity analyses, and checks the result against the dataset.
- code/exposure_tests.py: contains the lists of unions, alliances, security guarantees and coercive events, with sources. It recomputes the membership and coercion columns and runs every statistical test reported in the article.
- results/: the tables written by the two scripts.
- README.md: run instructions (Python 3.10 or later with pandas, numpy, scipy and openpyxl).
The coded scores, bases and classifications are released under CC BY 4.0 and the code under the MIT licence. Columns taken from third-party datasets remain subject to their providers’ terms and should be cited to the original sources, listed in the Sources sheet.
All scores were assigned by a single coder, with computer-assisted compilation of sources; there was no independent second coding. The statistical tests are descriptive: the units are the whole population of states, and a single snapshot in time cannot identify causal effects.