Open methodology
How the Democracys Score is calculated
A plain-language guide to what the score asks, how each part counts, why weak areas matter, and where the evidence is uncertain.
THE SHORT VERSION
Five equal tests. Weak links matter.
Each test contributes 20%. First, we calculate a familiar average. Then we calculate a weak-link average that falls more sharply when one part of democracy is very low. The published score blends the two.
A score of 72.0 does not mean a country is “72% democratic.” It is a relative estimate on this published scale.
THE INPUTS
Five public questions
We use five published V-Dem measures. Each answers a different question about who holds power and how that power is limited. We use one electoral index and four component measures rather than averaging five headline indices that repeatedly incorporate electoral democracy. This reduces a specific source of repeated weighting; it does not establish that the inputs are statistically independent.
Free and fair elections
Can people choose their leaders through broad voting rights, honest elections, free political groups, and open public debate?
20%Rights and checks on power
Are basic freedoms protected, laws applied fairly, courts independent, and leaders prevented from abusing power?
20%Public participation
Can people shape public life through civic groups, direct voting, and meaningful local and regional government?
20%Public reasoning
Do leaders explain decisions, consult the public, answer opposing arguments, and consider the common good?
20%Political equality
Do different social groups have equal rights, access, resources, and a real chance to influence political power?
20%CALCULATION
Reproduce the score
Start with one place and one year
Collect all five source estimates for that place and year. If one is missing, we do not invent it and no overall score is shown.
Give every test equal weight
Each of the five tests contributes one-fifth of the score. No single test is treated as the whole meaning of democracy.
Blend, scale, and round
Average the two means, multiply by 100, then round only the published result to one decimal place.
Rank within one observation year
Rank only places with complete 2025 inputs. Rank is descriptive; it does not establish that nearby scores are meaningfully different.
UNCERTAINTY
Expert-coded evidence is estimated, not exact.
The source publishes lower and upper estimates for each test. We apply the score formula separately to the five lower bounds and the five upper bounds to create a “source-bound range.” This range has no established coverage probability. It is not a 68% or 95% credible interval for the composite, because we do not propagate the joint posterior distribution or model dependence between the inputs.
Do compare broad patterns, durable gaps, and multi-year trajectories.
Do not treat a one-rank gap or a few tenths of a point as settled fact.
Do inspect the five democracy tests and original source documentation.
Do not infer a cause from a score change alone.
BOUNDARIES
What the score does not measure
- It does not judge a population, culture, religion, or national worth.
- It does not reward a political ideology or a particular policy outcome.
- It does not privilege presidential over parliamentary, federal over unitary, or proportional over majoritarian systems.
- It does not include GDP, happiness, military strength, or geopolitical alignment.
- It does not silently replace missing values with zero or splice predecessor states into current countries.
EXPLAINING CHANGE
Separate the calculation from the political story.
Country profiles compare exact observations 1, 5, 10, or 25 years apart. They identify which of the five areas rose or fell, whether recent movement differs from the longer trend, and the three largest adjacent-year changes within that window. Missing years are never bridged to invent an annual change.
Approximate point contributions are calculated by replacing the five starting inputs with ending inputs in every possible order and averaging each input’s effect on the score. This is a Shapley allocation of the formula’s change, not an estimate of the causal effect of a law or political event. Because the displayed input values are rounded, the contributions can differ slightly from the change between published scores.
The political-context note is a separate, dated summary of a selected Freedom House country assessment. The 2025 report generally describes 2024 and the 2026 report describes 2025. Some notes describe continuing conditions; Zanzibar’s limited historical coverage is explicitly labeled. One selected account cannot explain every movement across a century of observations. These AI-assisted summaries are not independently expert-reviewed, and source disagreement should be examined rather than hidden.
VERSION HISTORY
Method changelog
Country interpretation layer added
Dated, source-linked political context, model-change contributions, and measured-shift timelines added to country profiles. No score, ranking, underlying observation, weight, or scoring formula changed.
1.0 beta — initial public method
Five equal-weight V-Dem v16 inputs; 50/50 arithmetic-geometric blend; annual series calculated from 1900 through 2025; source archive fingerprint recorded.
This beta has not yet received external peer review. It should be cited as an experimental transformation, and the full historical series will be recalculated—never spliced—when a future source version is adopted.
Method sources
V-Dem v16 Codebook ↗V-Dem measurement methodology ↗V-Dem Country-Year Dataset v16 DOI ↗Download the derived data and provenance record →OPTIONAL TECHNICAL APPENDIX
For researchers reproducing the score
Most readers can stop above. The exact notation is kept here for researchers, reviewers, and people checking the calculation.
Show the exact research formula
DPS = 100 × [0.5 × AM(E,L,P,D,Q) + 0.5 × GM(E,L,P,D,Q)]E, L, P, D, and Q are inputs on the source’s 0–1 scale. AM is their equally weighted arithmetic mean; GM is their equally weighted geometric mean. A zero input makes GM zero. Missing inputs do not become zero.
| Democracys test | Source variable |
|---|---|
| Free and fair elections | v2x_polyarchy |
| Rights and checks on power | v2x_liberal |
| Public participation | v2x_partip |
| Public reasoning | v2xdl_delib |
| Political equality | v2x_egal |
Worked illustration, not a country: inputs 1.0, 0.8, 0.6, 0.4, and 0.2 have arithmetic mean 0.6 and geometric mean approximately 0.5210. Their blended score rounds to 56.1 out of 100.
Download the small, dependency-free calculation script. Run it with Node.js and five source-scale inputs. It reproduces the formula, not V-Dem’s underlying expert-coding model.