How we measure democracy – method, data and how to cite

HistoryGate shows how democracies rose and fell. This page explains where every number comes from, what we changed, what we cannot measure – and how you can check, reuse and correct our work.

What we measure

We look at democracy through eight pillars: elections, separation of powers and courts, rights, free press, a shared space of facts, civil society, mutual toleration (low polarisation), and social and economic security. For each pillar a lesson shows where it broke and where it held.

Where the numbers come from

Modern cases (from 1789): V-Dem. The values come from the Varieties of Democracy project, version 16 (Coppedge et al. 2026, https://doi.org/10.23696/vdemds26), as processed by Our World in Data. V-Dem combines ratings from several thousand country experts in a statistical model, country by country and year by year. We use V-Dem’s best estimate exactly as Our World in Data publishes it (three decimals). Only pillars 2 and 7 are changed – see below.

Our World in Data names the V-Dem code only for polarisation; we matched the other codes through the V-Dem codebook.

Case years: Germany 1932, Spain 1935, France 1938, Britain 1936, Finland 1932, Belgium 1937.

Pre-modern civilisations (before 1789): HistoryGate assessment. No openly licensed dataset covers all four features we need for societies before 1789. We therefore rate them ourselves – who chose the rulers, whether a council could check them, whether law bound them, and how many adults had a say – each on a scale of 0 (absent), 1 (partial) or 2 (strong), with a confidence level and at least one scholarly source. For participation the steps are under 5 %, 5–49 % and 50 % or more of all adults, women included. Without a reliable source we write “unknown”, never 0. We rate the central state in a stated period. Where none existed, we show a short sourced text instead of numbers. Every rating names who was left out. We never add up or rank civilisations. For the law feature we cross-check against the Seshat Global History Databank (public data, CC BY-SA 4.0).

What we changed – and why

• Pillar 2 is the average of two V-Dem indices. Both original values are in the download.
• Pillar 7 is not a V-Dem index. V-Dem’s polarisation score has no fixed range. To show it on the same 0–1 radar we convert it with 1/(1+eˣ): 0.5 is roughly the average of all countries and years in V-Dem’s data, and higher means less polarised. Its scale is not directly comparable with the other pillars.
• France is shown for 1938, before wartime restrictions began.

What we cannot measure

• Pillar 5 (shared facts) and pillar 8 (social and economic security) have no comparable historical measurement. We say so on the radar instead of filling the gap with a stand-in measure.
• V-Dem’s historical scores are assessments made today, with hindsight.
• Six cases cannot prove cause and effect. The radar shows a pattern; it does not prove why a democracy fell.
• Pre-modern ratings are judgements, not measurements. We mark them as such.

How we check our work

Before publication, a separate reviewer – currently an AI system, not the author – checks each lesson claim by claim against its sources. Pre-modern ratings are cross-checked by four AI models from different makers. Where models disagree and cite a real source, we re-check that source. AI output is never used as a source. None of this replaces expert review – see the invitation below.

Download the data

All HistoryGate data on this page is free to reuse under CC BY-SA 4.0.

• hg_democracy_radar.csv – one row per case and pillar: country, year, pillar, value, V-Dem variable, and the original V-Dem values before averaging or rescaling; empty for pillars 5 and 8.
• hg_democracy_trail_modern.csv – V-Dem values per civilisation and year from 1789 (available later in November 2026).
• hg_democracy_trail_premodern.csv – our four-feature ratings with confidence, sources and “who was left out” (from 8 December 2026).

When you reuse the data, credit V-Dem (via Our World in Data), Seshat where used, and HistoryGate, and share your work under the same licence.

How to cite

HistoryGate (2026). Democracy radar and democracy trail, version 1.0. https://historygate.eu/en/lessons/method. Data based on V-Dem v16 (Coppedge et al. 2026, https://doi.org/10.23696/vdemds26), processed by Our World in Data. CC BY-SA 4.0.

@misc{historygate_democracy_2026,
  author = {{HistoryGate}},
  title  = {Democracy radar and democracy trail},
  year   = {2026},
  url    = {https://historygate.eu/en/lessons/method},
  note   = {Version 1.0. Based on V-Dem v16 (Coppedge et al. 2026, doi:10.23696/vdemds26), processed by Our World in Data. CC BY-SA 4.0}
}

Historians wanted

Are you a historian or political scientist? We are looking for experts to review our ratings of early democracies – Athens, the Roman Republic, the Aztec Empire, Han China and many more. Reviewers are named on this page. Write to us: expertboard@historygate.eu

Changes

Every change to data or method is listed here with a date. Version 1.0 · 3 November 2026.