freedomrankings.comHow It Works
Methodology
How FreedomRankings measures and scores freedom across all 50 states
How it works
Methodology
How we measure and score freedom across all 50 states. DC & US territories are tracked separately.
The numeric ranking covers the 50 US states. The District of Columbia and the five inhabited US territories (Puerto Rico, Guam, US Virgin Islands, Northern Mariana Islands, American Samoa) are governed under a different constitutional architecture and aren't directly comparable to states on several of our ten categories.
We still track DC on all ten categories using the same methodology, but keep it out of the 1–50 ranking so including or excluding DC never shifts state ranks. The five territories are covered on our territories & federal district page, which explains the constitutional-status story (Insular Cases, Territory Clause) and the freedom-relevant laws in each.
Each state is evaluated across 10 categories of freedom, each scored on a scale of 0 to 10. The overall freedom score (0–100) is a weighted average of all category scores, multiplied by 10.
By default, all categories are weighted equally. Users can adjust category weights on the Rankings page to prioritize the freedoms that matter most to them.
How much does that default actually decide?
A fair amount, and we publish the number rather than leaving you to wonder. Ranked under four different defensible weightings, the average state moves about ten places. Because seven of the ten categories are personal-freedom categories, equal category weights implicitly put roughly 70% of the score on personal freedom — weighting the three dimensions equally instead puts it at 33% and reorders the table substantially. The full comparison, per state, is on the sensitivity analysis page.
Data sources & transparency
Hybrid (default): Each of the 10 categories is sourced from an annually-updated third-party publication (see the per-category table below). Two categories — 1st Amendment speech and religious liberty — have no strong annual state-level source and are marked editorial. This source is the recommended default going forward because it stays current as each upstream publisher releases a new edition.
Cato: Derived from the Cato Institute "Freedom in the 50 States" 2023 edition, which evaluates 230+ state policies. Cato has not published a newer edition, so this source is frozen at 2023 and is retained for historical continuity.
Editorial: FreedomRankings's own opinion-based assessment. Not research-backed; clearly labeled with a caution banner and should be treated as opinion, not fact.
Per-category sources (Hybrid)
| Category | Source | Cadence | Confidence |
|---|---|---|---|
| 1st Amendment | editorial | Editorial | |
| 2nd Amendment | annual | Stale edition | |
| 4th Amendment | periodic | Current | |
| Economic Freedom | annual | Current | |
| Criminal Justice | annual | Current | |
| Drug Policy | annual | Current | |
| Property Rights | periodic | Current | |
| Religious Liberty | editorial | Editorial | |
| Education Choice | annual | Current | |
| Regulatory Burden | annual | Stale edition |
"Stale edition" means the most recent publisher release is older than one full cadence cycle. "Editorial" means no third-party index publishes this category and the score reflects FreedomRankings's judgment. Hover a source chip for the full publication name, edition year, and vintage note.
What is editorial / estimated
- Net worth — midpoint of the bracketed asset ranges on financial disclosures, not an exact figure. The range is always shown alongside it
- Policy positions for governors — editorial summaries of public actions
- Two category scores (1st Amendment, religious liberty) — no annual third-party state index exists for either
Changed: approval ratings used to appear on this list and have been retired outright — an estimate presented with a percentage sign and a progress bar reads as a measurement no matter what the caption says. The editorial “integrity score” has been retired for the same reason and is no longer published anywhere: not on profiles, not in the directory, and not as a sort. It was the share of an official’s votes that matched a direction we assigned to each bill by hand, without a citation or a reviewer — twelve such calls, covering 2017 to 2023, aggregated into a percentage and a letter grade attached to a named living person. The only score an official now carries is the Accountability Index, which is computed from filing records. Key votes are still published; we show what a member did and link the official record, without grading it.
What is verified from official sources
- Congress member data — from Congress.gov official API
- Campaign donor data — from FEC.gov official filings
- Congressional salaries — public law ($174,000 base)
- Governor salaries — public record by state
- Cato freedom scores — published academic index
Officials used to carry an editorial “integrity score”: the share of curated policy positions where we judged they had followed through. Where we had no roll-call record it fell back to party-line assumptions, which meant the grade was largely a restatement of party affiliation dressed up as a measurement. It has been removed.
In its place is the Accountability Index, which measures one thing: disclosure compliance. It does not measure whether an official's politics are good, whether they kept campaign promises, or whether the public approves of them. Every component is a fact recorded in a government filing — a date on a Periodic Transaction Report, a position in the Congressional Record, a receipt in an FEC filing.
| Component | Weight | What it counts & why it carries that weight |
|---|---|---|
| Disclosure timeliness | 30% | Share of stock transactions reported within the 45-day statutory deadline. A late PTR is a black-and-white failure to meet a statutory deadline, visible on the face of the filing with no interpretation required. It is the single most defensible accountability signal available, so it carries the largest weight. Source: House Clerk / Senate PRD Periodic Transaction Reports (STOCK Act) |
| Vote participation | 25% | Share of recorded roll-call votes on which the member cast Yea, Nay, or Present rather than not voting. Showing up to vote is the core duty of the office and the record is complete and unambiguous. Weighted below disclosure because excused absences (illness, bereavement, official travel) are not distinguished in the public record. Source: Congress.gov / House Clerk / Senate roll-call records |
| Disclosure filing record | 20% | Whether an annual Personal Financial Disclosure is on file for the most recent required year. Annual PFDs are mandatory under the Ethics in Government Act. A missing filing is a compliance fact. Weighted below timeliness because filing extensions are routinely granted and are not always reflected in the public index. Source: House Clerk / Senate Public Records annual PFD filings |
| Committee conflict exposure | 15% | Share of disclosed trades in sectors overseen by a committee the member sits on. A trade inside your committee's jurisdiction is a disclosed structural conflict, not proven misconduct — nothing here asserts the member acted on non-public information. Weighted modestly to reflect that it flags exposure, not wrongdoing. Source: PTR filings cross-referenced against committee jurisdiction (House/Senate rules) |
| Funding concentration | 10% | How concentrated itemized campaign receipts are among the largest donors. Broadly funded campaigns answer to more people than narrowly funded ones. This is the most interpretive component in the index and is deliberately weighted least; it is reported as a disclosed characteristic, not a violation. Source: FEC itemized receipts (Schedule A) |
Absence of a record is not a failing
An official who has never traded a stock has no STOCK Act filing history. That is no data, not a zero, and scoring it as a zero would penalize members for not doing the thing the component exists to detect. Components with no underlying records are dropped and the remaining weights are rescaled to 100%. Where under half the nominal weight is available we suppress the headline number entirely and show only the components, because a composite resting on one input is not a composite.
The same rule applies to thin samples, not just missing ones. Vote participation is only reported when a member has at least 20 roll-call votes in our records. A member serving a full session votes on hundreds; a single-digit denominator means our ingestion is incomplete for them, not that they had five chances to vote. Reporting “100% attendance” off five votes would be a number that looks like a measurement and isn't one.
Methodology version 1.0.0. The statutory filing deadline used is 45 days (STOCK Act of 2012, Pub. L. 112-105 § 6; 5 U.S.C. app. § 105(d)(1)). We score against the 45-day outer bound rather than the 30-day notice clock because the date of notice is not present in the public filing and scoring against it would require guessing.
Every component, with its underlying counts, is exposed on the API and in the bulk data export, so the composite can be reproduced or disputed without scraping the site. Disagreements go in the corrections log.
We intend to score how each legislator's votes affect each of the ten freedom categories. The full coding rubric is published before any vote has been coded and before any score exists.
That order is deliberate. A system that grades named legislators on “freedom” is only defensible if its rules were open to challenge before anyone had a number attached to them; publishing the rules afterwards invites the reasonable suspicion that they were fitted to the result. The rubric states its own premises, the strongest objection to each one, what it refuses to score and why.
Read the coding rubric →Each federal official's page carries counts taken from the roll-call and sponsorship record: votes cast and missed, bills sponsored and cosponsored, committee seats, and years in the current term. These are counts, not judgments — there is no composite and no grade.
Party unity, and what it is not
Party unity is the share of votes where a member sided with the majority of their own party. It counts Yea and Nay only — “Present” takes no side, and counting it as breaking ranks would penalize something that isn't. Roll calls where a member's party splits evenly are excluded, because there is no majority to agree with. It is only reported once a member has at least 20 qualifying votes, and always beside its denominator.
It measures voting alongside your party. That is not ideology, and it is not DW-NOMINATE. We publish no ideology score: DW-NOMINATE is a specific published scaling procedure, so putting its name on a home-grown number would misrepresent it — and an unnamed “ideology” figure would be exactly the editorial scoring this site removed. A member can score low on party unity for opposite reasons depending on which direction they broke, and the number alone cannot tell you which.
The ten category scores are currently derived from published third-party indices — the rank tables below. That is a defensible way to start, but it has a real limitation: if you ask why a state scores what it does, the honest answer today is “because that is where another organization ranked it”.
We are rebuilding onto individually documented policy variables so the answer resolves to a statute instead. 54 of 250 target variables are defined so far, each with a published unit, direction, weight, and weight rationale.
No state has been scored on these variables yet
The catalog is definitions only. Coding values means reading the statute in each jurisdiction, and each coded value will carry its own citation and retrieval date. We publish the catalog and its weights first, before any number depends on them, so the methodology can be argued with on its merits — and so no one has to wonder whether the variables were chosen after someone saw which states they favored.
Download the full catalog as CSV.
Version & change history
Every published score is identified by two things: the dataset version, a content hash of the numbers themselves, and the methodology version, which describes how they were computed. The hash changes if and only if a published figure changes, so if you cite a number you can check later whether it still means what it did.
US State Freedom Index 2026
- Dataset version :
- 5efe4d2c6d30
- Methodology version :
- v2.0
- Jurisdictions :
- 51
- Categories :
- 10
Overall score is the weighted mean of the ten category scores (each 0-10), multiplied by 10 and rounded to one decimal place. Default weights are equal (1 each); the ranking recomputes live when a reader changes them.
Cite as: FreedomRankings State Freedom Index 2026 (methodology v2.0, dataset 5efe4d2c6d30)
International Freedom Index 2026
- Dataset version :
- 3362448ea184
- Methodology version :
- v2.0
- Jurisdictions :
- 197
- Categories :
- 10
Overall score is the sum of the ten universal category scores (each 0-10), giving a 0-100 scale. Countries with no source coverage remain on the 5.0 editorial placeholder and are flagged as data-pending.
Cite as: FreedomRankings International Freedom Index 2026 (methodology v2.0, dataset 3362448ea184)
hybrid v2.0, 2026-01-01
Replaced the editorial baseline with named annual third-party sources for eight of the ten categories: Fraser EFNA (economic freedom), EdChoice ABCs (education choice), Institute for Justice (criminal justice, property rights), Vera Institute (incarceration), NORML (drug policy), Mercatus (regulatory burden), Guns & Ammo (2nd Amendment). 1st Amendment and Religious Liberty remain editorial because no annual state-level index covers them.
Effect on published scores: Every state score changed. Scores under v2.0 are not comparable to v1.0 figures; a state that moved between editions may have moved because the source changed, not because its policy did.
cato v1.0, 2023-01-01
Cato Institute "Freedom in the 50 States", 2023 edition, ingested unchanged as a selectable historical source.
Effect on published scores: Frozen. Cato has published no newer edition, so these figures cannot change.
freedomscore v1.0, 2025-01-01
The original FreedomRankings editorial assessment: every category scored by our own judgement rather than from a published index. Retained as a selectable source for continuity with figures cited before the hybrid index existed.
Effect on published scores: These are opinion, not research output, and are labeled as such wherever they appear. They are not comparable to hybrid v2.0 scores, which are mostly third-party sourced — a state can differ substantially between the two without any policy having changed.
The 10 categories
Freedom of speech, religion, press, and assembly. Measures how well a state protects expressive freedoms and limits government restrictions on speech and protest.
Source note. No single published index ranks 1st Amendment speech protections state-by-state annually. Scores are an editorial synthesis of press-shield laws, anti-SLAPP coverage, and transparency-law rankings.
Factors considered
Right to keep and bear arms. Evaluates firearm regulations, carry laws, and the degree to which the state respects gun ownership rights.
Source note. Guns & Ammo has not published a publicly-indexable 2024 or 2025 edition of "Best States for Gun Owners". The 2023 ranking is retained until a newer edition is confirmed.
Factors considered
Protection against unreasonable searches and seizures. Measures privacy rights, surveillance restrictions, and civil asset forfeiture protections.
Source note. IJ's periodic (not strictly annual) grading of each state's civil-asset-forfeiture statutes.
Factors considered
Taxes, business climate, and economic regulations. Assesses the overall burden of state government on economic activity and individual financial freedom.
Source note. EFNA 2024 edition. Scores 50 states on government spending, taxation, and labor-market freedom.
Factors considered
Fairness and proportionality in the criminal justice system. Evaluates incarceration rates, police accountability, and the balance between public safety and individual rights.
Source note. Vera's 2024 incarceration tally across all 50 states; captures the carceral-intensity component of criminal-justice freedom.
Factors considered
Drug policy liberalization. Measures marijuana legality, decriminalization efforts, and the treatment of drug offenses.
Source note. Composite of NORML state pages, DISA legality map, and NCSL medical-cannabis tracker. No single publisher; each source cross-checks the others.
Factors considered
Protection of private property from government taking and excessive regulation. Evaluates eminent domain protections and land use restrictions.
Source note. IJ Castle Coalition legislative tracker for eminent-domain protections and post-Kelo reforms.
Factors considered
Protection of religious exercise and conscience rights. Measures state-level religious freedom restoration acts and accommodation policies.
Source note. Built from state RFRA presence, conscience-protection statutes, and faith-based adoption laws. No aggregated third-party index exists for this category.
Factors considered
Freedom to choose educational options outside the public school system. Evaluates school choice programs, homeschool regulations, and charter school access.
Source note. "ABCs of School Choice" 2025 edition. Scores reflect voucher / ESA / tax-credit programs per state, weighted by enrollment eligibility.
Factors considered
Overall regulatory burden on individuals and businesses. Measures occupational licensing requirements, business entry barriers, and bureaucratic overhead.
Source note. RegData State Edition 2024. Counts regulatory restrictions -- occurrences of binding terms ("shall", "must", "may not") -- in each state administrative code. The count is RAW and NOT normalized: it is not adjusted for code length, population, or economic output, so a state with a longer administrative code ranks worse whether or not its rules bind harder. Read it as a measure of code volume rather than of regulatory burden as experienced. Covers 48 states; West Virginia and Arkansas are not published and retain editorial scores.
Factors considered
9–10: Most free
7–8: Above average
5–6: Average
3–4: Below average
1–2: Most restrictive