How we rate senators and candidates.

1. Overview and design principles

TheMajority.us rates federal politicians and candidates by answering a single question: Where do they stand, and can we prove it?

The system has five interlocking layers:

• Spectrum position — where they sit on a 7-point left-right axis
• Majority Alignment Score (MAS) — how often votes align with positions 60%+ of Americans support
• Word-vote gap — distance between public statements and actual votes (incumbents only)
• Voter gap — distance between constituents' positions and actual votes (incumbents only)
• Cross-ideology scorecards — validation from sources across the political spectrum

Five principles guide everything

• Every claim has a source. No data point appears on the site without a verifiable link to its origin.
• Estimates are labeled, not hidden. When data is partial, scores are marked as estimates with the basis explained.
• Cross-ideology by default. Every rating includes scorecards from both left and right, so users can see ideological convergence (or disagreement) on the same record.
• Descriptive, not prescriptive. We don't tell you who to vote for. We show you where they stand and where the public stands, so you can decide.
• Open to correction. Every rating page has a “Report an issue” link. We treat user corrections as a feature, not a nuisance.

2. Majority Alignment Score (MAS)

MAS is the foundation score: how often a senator's roll call votes align with positions roughly 60%+ of Americans support, measured across our four pillars.

How it's calculated

For each pillar, we identify the votes that match well-polled policy positions. For each vote:

1. Find the relevant policy issue (e.g., “Medicare drug price negotiation”)
2. Average support across multiple polling sources for that issue
3. If average ≥ 60%, the “majority position” is YES; if ≤ 40%, the “majority position” is NO
4. If between 40% and 60%, the vote is excluded — there's no clear majority
5. Compare the senator's vote to the majority position; one point per match

The pillar score is matches divided by total votes considered, multiplied by 100. The overall MAS is the average across the four pillars.

Reading the score

3. The four pillars

MAS is calculated across four policy domains we call pillars: Environment, Healthcare, Education, and Immigration. We chose these because they're consistently top-ranked priorities across both Republican and Democratic voters in nearly every major poll, and because each has a substantial and identifiable record of congressional votes.

Each pillar contains 5–15 specific policy issues with clear majority positions. For example:

• Healthcare: Medicare drug price negotiation, prescription drug importation from Canada, capping insulin costs, prior authorization reform
• Environment: Solar and wind tax credits, methane emissions standards, Keystone XL approval, fuel efficiency standards
• Education: Federal student loan refinancing, Pell Grant expansion, charter school funding, school nutrition standards
• Immigration: DACA protection, employer verification (E-Verify), border security funding, refugee admissions ceiling

The full list of issues per pillar is published and updated as new policy debates emerge. We never add an issue to the list because of its political valence — only because polling shows a clear majority position and it has been or is likely to be voted on by Congress.

4. What “majority” means

An issue qualifies as “majority” if multiple major pollsters consistently show 60% or more national support (or opposition). This 60% threshold is deliberately high — it filters out polarized issues where the public is genuinely divided.

For each qualifying issue, we cross-check against pollsters with different methodologies and ideological reputations. If Pew and Gallup show 76% and 72% support, but Rasmussen shows 38%, that's a red flag and the issue is excluded pending review. When all sources cluster within roughly 10 points, we treat it as a stable majority position.

We also break out polling by party affiliation. When 66% of Republicans and 78% of Democrats agree on an issue, it qualifies as a “Republican-majority” issue specifically — useful for evaluating whether a Republican senator votes with their own party's voters.

5. Political spectrum (DW-NOMINATE)

The 7-bucket spectrum (Far-left, Progressive, Center-left, Center, Center-right, Conservative, Far-right) is a plain-English label sitting on top of an academic measure called DW-NOMINATE.

DW-NOMINATE is a statistical method developed by political scientists Keith Poole and Howard Rosenthal that places every member of Congress on a left-right axis based on their roll-call voting behavior compared to all other members. It produces a score from roughly −1.0 (far left) to +1.0 (far right). It is the most widely used measure of congressional ideology in academic political science.

We translate the DW-NOMINATE score into our 7 buckets:

The bucket boundaries are reviewed annually to ensure they remain calibrated as Congress as a whole shifts.

6. Word-vote gap (incumbents only)

The Word-vote gap measures the distance between what a politician says publicly and how they actually vote. Some politicians have remarkable consistency between their rhetoric and their record. Others say one thing and do another. Both patterns matter, and we surface both.

For each major issue in a senator's pillars, we identify:

• What they say: A direct quote from their campaign site, official press releases, congressional floor speeches, or public statements (with a date and source link)
• What they voted: The roll call number, vote, and date (with a Senate.gov source link)
• The alignment status: Aligned, Partial, Mismatch, or No record

The overall gap is calculated as the percentage of analyzed issues where votes contradict statements, weighted by the magnitude of contradiction. Gaps are bucketed as Narrow, Moderate, Wide, or Very-wide.

7. Voter gap (incumbents only)

The Voter gap measures the distance between what the senator's constituents want and how the senator votes. This is different from the Word-vote gap because constituents may want something the senator never publicly addressed.

For each major issue, we compare:

• The senator's vote on that issue
• State-level polling on that issue (when available)
• National polling on that issue (as a fallback)

State-level polling is sparse for many issues. When we rely on national polling as a proxy, the rating is marked as a medium-confidence Voter gap rather than high-confidence. We disclose this clearly on the rating page.

The Voter gap is particularly useful for understanding whether a senator represents their state. A Republican senator from a deep-red state who votes in line with their state's Republican voters has a narrow Voter gap. The same senator voting against the state's Republican voters on, say, drug pricing has a wider Voter gap.

8. Cross-ideology scorecards

Every incumbent rating page shows scorecards from organizations across the political spectrum. The point isn't to endorse any of these organizations — it's to demonstrate convergence. When organizations with opposite ideologies arrive at consistent rankings, our rating gains credibility no single source could provide.

Each score is sourced directly from the organization's published scorecard. We never modify the scores. When an organization hasn't published its current-Congress scorecard yet, we mark the score as estimated and explain the basis (typically the senator's lifetime score from that organization).

The scorecards row is shown as a horizontal strip of small score chips on each candidate card. A senator scoring 96 from Heritage Action and 0 from LCV is unambiguously conservative. A senator scoring 50 from both is unambiguously moderate. The chips do the explaining without us needing to.

9. Incumbents vs. challengers: a critical distinction

Our rating system relies heavily on voting records. Incumbents have voting records. Challengers do not. This means we cannot rate the two groups identically without misleading users about the strength of our evidence.

How we rate incumbents

For sitting senators and representatives, we have:

• A complete roll call voting record from Congress.gov
• A DW-NOMINATE position derived from those votes
• Cross-ideology scorecards based on those votes
• An MAS score calculated from those votes against polling data
• Publicly recorded statements that can be compared against votes (Word-vote gap)
• Constituent-level polling that can be compared against votes (Voter gap)

The full rating system applies. Every score is derived from observable behavior, not inference.

How we rate challengers

For challengers — anyone who has not previously held the office — we have:

• The candidate's own campaign website
• Press interviews and public statements
• Endorsements they have received or sought
• Their party affiliation
• Their professional background
• Sometimes: voting records from different offices (state legislature, governorship, etc.)

From these sources, we infer:

• A spectrum position based on stated platform across our four pillars. This is shown on the spectrum line with a dashed/hollow marker instead of a solid one — the visual cue that this is self-described, not behaviorally verified.
• Stated positions on major issues sourced directly from their own materials (with links).

For challengers, we deliberately do not show:

• An MAS score (no votes to score)
• A Word-vote gap (no votes to compare against words)
• A Voter gap (same reason)
• Cross-ideology scorecards (Heritage, ACU, etc. don't score non-incumbents)

What this means for users

When you compare an incumbent and a challenger on the same race page, the visual treatment is intentionally different:

• The incumbent's card shows the full rating system — five components, all with sources.
• The challenger's card shows the spectrum (dashed marker) and a curated list of their stated positions on key issues, each with a link to the original source.

This makes the difference in evidence clearly visible. You can read both cards side by side and know exactly what kind of judgment each is asking you to make.

What happens when a challenger wins

If a challenger is elected to Congress and begins voting, their rating transitions from challenger-style to incumbent-style as soon as they have enough roll call votes for the metrics to be meaningful. Typically this is around 50 votes, or roughly 6 months into their first term. Their original challenger-era stated positions remain visible as historical context, and we explicitly compare their early voting record to those stated positions — generating a Word-vote gap from day one of their voting career.

10. Polling sources we use

Every claim about majority opinion is grounded in actual polling. We deliberately use sources with different methodologies and ideological reputations so that no single pollster's house effects drive our scores.

When pollsters disagree by more than ~10 points on the same question, we exclude that issue from our calculations until additional polling clarifies the picture. We never average a single pollster's number into our ratings without independent corroboration.

11. AI and human review

Some content on this site — particularly statement summaries for the Word-vote gap and the auto-generated race briefs — is initially drafted by AI and then verified before publication. We disclose this clearly on every page where it applies.

Our process:

1. An AI system surfaces candidate statements and matches them to specific votes
2. The drafted content is published with an “AI-generated” badge and a “review pending” status
3. Our editorial team and reader reports verify the content over time
4. Once verified by a human, the badge changes to “Verified”
5. If a reader reports an inaccuracy via the “Report an issue” link, the content is reviewed within 48 hours

The AI never makes the final ideological judgment. It surfaces sources; humans confirm the alignment status. We use AI for scale, not for opinion.

12. Update cadence

Real-time alerts (candidate withdrawals, primary results, MAS shifts of 10+ points) are pushed to members within hours of detection.

13. Report a problem with a rating

Every rating page on this site has a “Report an issue” link. We encourage you to use it whenever you spot:

• An incorrect statement attributed to a politician
• A vote categorized incorrectly
• A source link that's broken or wrong
• A scorecard score that doesn't match the source organization's published value
• A candidate's stated position that doesn't match their actual platform
• A name spelling, party affiliation, or biographical detail that's wrong

Reports are reviewed within 48 hours. We respond to every report, whether or not we end up making a change. Treating user corrections as a feature is part of our methodology — no rating system is bias-proof, but a system that responds to corrections gets closer to honest over time.

You can also email us directly at corrections@themajority.us.

14. Changelog

Material changes to the methodology are logged here. Score recalculations triggered by methodology changes are noted on individual senator pages so users can see when and why a score moved.