Methodology & Sources

A category with 4 tracked items in the current session:

Score = (1.0 + 0.0 + 0.3 + 0.0) ÷ (1.0 + 1.0 + 1.0 + 0.5) = 1.3 ÷ 3.5 = 37.1% → C−

Cosponsorship timing

• Early cosponsor (within 90 days of introduction): 0.5 — signing on early signals genuine commitment
• Late cosponsor (after 90 days): 0.25 — joining late earns half credit on the item, reflecting that late sign-ons carry less weight than early commitment

When a cosponsored bill reaches a floor vote

The cosponsorship converts to the full vote weight (1.0). Even a late cosponsor who votes the right way gets full credit for the vote; a member's grade improves when bills they cosponsored reach the floor.

Sponsor bonus

The original sponsor of a tracked bill gets 1.0 weight, or 1.5 if the bill reaches a floor vote, recognizing the effort of introducing legislation. The bonus also applies when a sponsor votes correctly on their own bill on the floor.

When Congress considers several closely related items at once (such as a trio of CRA resolutions on the same topic, or a week of floor votes on Iran sanctions), we group them into a batch and split the weight evenly so the batch totals roughly 1.0.

Without batch weighting, a single legislative push that produces three near-identical votes would triple the weight of that topic relative to other categories. Batch normalization keeps the scoring balanced: one legislative event equals approximately one unit of weight, regardless of how many procedural votes it generates.

Example

Three CRA votes on California vehicle-emission rules, all scored the same week:

Total batch weight ≈ 1.0 instead of 3.0, preventing the batch from overwhelming the category score.

Committees from chambers of commerce, business associations, and manufacturers that oppose union-organizing legislation and minimum-wage increases.

Committees funded by the cryptocurrency and digital-asset industry, including its super PACs (the Fairshake network) and connected corporate PACs.

Committees from defense and aerospace contractors active on military spending, weapons procurement, and arms-sales policy.

Pro-democracy, voting-rights and campaign-finance-reform advocacy committees.

Environmental and climate-protection advocacy committees. Distinct from fossil-fuel industry committees.

Committees from oil, gas, and coal companies that oppose stricter climate and environmental regulation.

Gun-rights advocacy committees.

Gun-violence-prevention advocacy committees.

Healthcare-provider committees — hospitals, physicians, dentists, nurses and allied providers. Distinct from the pharmaceutical and insurance industry.

Organized-labor committees representing union workers.

Committees from police unions, law-enforcement associations, and private-corrections companies active on criminal-justice, policing, and immigration-detention policy.

LGBTQ-rights advocacy committees.

Committees from pharmaceutical manufacturers and health-insurance companies that oppose drug-price controls and a public health-insurance option.

Political action committees that primarily support pro-Israel policy positions in Congress, including direct contributions and independent expenditures.

Committees from real-estate developers, homebuilders, and property-industry groups that oppose rent regulation and tenant-protection legislation.

Committees supporting abortion rights and reproductive-healthcare access.

Committees opposing abortion or supporting abortion restrictions.

Veterans and military-service advocacy committees focused on war and foreign policy. Distinct from defense-contractor committees.

Committees from securities firms, investment banks, hedge funds, and private-equity firms that oppose stricter financial regulation.

• Only itemized contributions (over $200) are individually tracked by the FEC. Under-$200 totals combine FEC’s Individual_Unitemized_Contribution figure with the gifts under $200 that a campaign lists line by line on its own reports.
• Hybrid PACs and leadership PACs that span multiple issue areas are categorized by their primary purpose.
• 527 organizations are not tracked; FEC data covers only registered PACs.
• Bundled contributions are attributed to the earmarking PAC, not the original individual donor.
• Independent expenditure data is available from 2010 onward; earlier election cycles do not include Schedule E records.

Lobbying data comes from LD-2 quarterly activity reports filed with the Senate LDA database. These filings are required under the Lobbying Disclosure Act of 1995 for any lobbying firm or organization that meets registration thresholds.

Our coverage spans 2025–2026 filing years and includes only House bills tracked in our reference list. Data is refreshed periodically from the LDA API; it is not real-time. Currently, 33% of tracked bills have at least one matching lobbying filing.

Dollar figures shown on bill pages are combined quarterly totals for all lobbying activity by each firm; they are not bill-specific spend. An LD-2 filing covers a firm’s entire quarterly lobbying portfolio. If a firm lobbies on 50 bills, the income or expense figure covers all 50.

We show these totals as a measure of the lobbying resources deployed by firms active on the bill, not the cost of lobbying on that specific bill.

Income is reported by lobbying firms (registrants) and represents what they were paid by clients. Expenses are reported by organizations that lobby in-house and represent what they spent on their own lobbying operations. A single filing reports one or the other, never both. Our “combined quarterly activity” figure sums both types across all filings for a bill.

LD-2 filings require lobbyists to disclose prior “covered positions” in government: roles in Congress, the Executive Branch, or senior staff positions held within the previous 20 years. We surface this as a transparency indicator: it shows which lobbyists working a bill previously held government roles.

This is public record and legal under most circumstances. It is not an accusation of wrongdoing. We only count lobbyists listed on activities that specifically reference the bill, not all lobbyists on the filing.

• Bills are matched by searching filing activity descriptions for the exact bill number (e.g., “H.R.3421”). Filings that reference a bill by name only or use alternate formatting may not be captured.
• Dollar figures are quarterly firm-wide totals, not per-bill spend.
• Only the top 10 clients are displayed per bill, ranked by total reported activity.
• “No lobbying data” means no matching filings were found in the LDA database; it does not necessarily mean zero lobbying occurred on that bill.
• Data is refreshed periodically; the cache may not reflect the most recent filings.
• Only House bills are currently tracked; Senate bill lobbying is excluded.
• Filing count reflects the number of firms filing, not lobbying intensity; one filing from a large firm may represent more activity than ten from small ones.

Each bill page places every currently-serving member’s vote next to the money and the local reality around that bill’s issue area(s), so the cross-reference is visible in one place. For a weapons bill, that means a member’s vote alongside the pro-Israel PAC money they received, the defense stocks they traded, and the travel they took on that issue. Every figure is computed at build time from the same disclosures shown on member profiles: PAC contributions (FEC), privately funded travel (House Clerk), and stock trades (STOCK Act PTRs), joined to each bill’s issue area. Nothing on a bill page changes any letter grade; it is context, not scoring.

The “Campaign money in context” panel compares the side that voted with the bill’s scored position against the side that voted against it, by average PAC dollars per member and the share of each side that took money from the relevant industry’s PACs. We report this as correlation, never causation, and we say so plainly when money does not track the vote; for many bills, the same industry funds large majorities on both sides, and the panel states that directly rather than implying a pattern that isn’t there.

Reading the numbers honestly: each member’s own top donor committee is named in their row, not just a category total. Disclosed stock trades are reported transactions, not a current portfolio; absence means non-disclosure, not non-ownership. District and state indicators (uninsured rate, rent burden, incarceration, maternal mortality, and so on) are neutral constituent context shown next to each member with a plain above/below-national marker; they are not a scorecard and carry no “good/bad” coloring. A $0 in a covered issue (a member who refuses that industry’s PAC money) is shown distinctly from an issue we don’t track money for at all.

Coverage varies by issue. Some issue areas have no industry PAC category or no disclosed trades; on those bills the money comparison is replaced by a national/state context note rather than empty zeros. The roster table’s columns adapt to what data actually exists for the bill, and a district map offers the same record shaded by vote, issue grade, or PAC dollars.

Members of Congress must disclose securities transactions within 45 days under the STOCK Act. We show each transaction’s asset, ticker, type (purchase, sale, exchange), date, and owner (self, spouse, joint, or dependent child).

The STOCK Act requires disclosure in dollar ranges, not exact amounts; we display the range exactly as filed without estimating values. Each transaction links to the original filing on the Clerk of the House website.

Limitations: Transactions below $1,000 are not required to be disclosed. The 45-day reporting window means recent trades may not yet appear. Ticker symbols may be inferred from asset descriptions by AI. Verify against the linked original filing PDF.

Floor speeches come from the Congressional Record via GovInfo. We include substantive remarks and filter out procedural items (adjournments, motions, bill introductions, etc.). Speeches under 50 words are excluded.

Each speech is classified into 1–3 of 21 topic areas using AI analysis, which also generates a short summary and flags whether the speech is substantive or commemorative. The 21 topics are: Palestine & Israel, Immigration, Climate & Environment, Healthcare, Housing, Labor, Reproductive Rights, LGBTQ+ Rights, Criminal Justice, Democracy & Voting, Finance & Money, National Security, Education, Economy & Inflation, Veterans, Technology, Agriculture & Food, Gun Policy, Infrastructure, Foreign Policy, and Commemorative.

The Rhetoric Tracker counts how often members use specific keywords across 35 issue areas: 20 topic-based areas (e.g., Palestine & Israel, Immigration, Climate & Environment, Healthcare, Housing, Labor, Reproductive Rights, LGBTQ+ Rights, Criminal Justice, Democracy & Voting, Spending & Fiscal, National Security, Education, Economy & Trade, Veterans, Technology, Agriculture & Food, Gun Policy, Infrastructure, Foreign Policy) and 15 framing/narrative areas (e.g., Affordability & Cost of Living, War & Peace, Healthcare Access & Cost, Populism & Anti-Establishment, Culture War Rhetoric, Social Safety Net, Corruption & Accountability, Immigration Framing, Energy Policy Framing, China & Great Power Competition, Economic Framing, Bipartisanship & Unity, Children & Families, Religion & Faith, Executive Power & Checks). To compare fairly between frequent and infrequent speakers, all counts are normalized per 10,000 words of total speech text.

Limitations: AI topic classification handles most speeches well but edge cases exist; topics are assigned algorithmically, not reviewed by humans. Keyword matching counts occurrences, not sentiment; a member arguing against a policy will still register the keyword. Some remarks are submitted for the record rather than spoken aloud.

We show privately funded travel filed with the Clerk of the House: sponsor organization, destination, dates, and cost breakdown (transportation, lodging, meals, other). Filings include both member travel and staff travel — each trip is labeled accordingly. Filing links are provided where available.

Limitations: Only privately funded travel is disclosed; official government travel (congressional delegations) is not included. Cost figures reflect what the sponsor reported; some filings may not have extractable cost data and are marked “N/A.” Cost breakdowns are extracted using AI-assisted document reading (OCR) from scanned PDF filings.

Individual donor data comes from FEC bulk data files (Schedule A) covering all available election cycles from 2004 to present. We break contributions into five size tiers: under $200, $200–$499, $500–$999, $1,000–$1,999, and $2,000+.

The “small donor percentage” is the share of donations under $200, combined across both FEC cycles of the current congressional session (2024 + 2026 for the 119th Congress). A higher percentage indicates more grassroots funding; a lower percentage indicates reliance on large-dollar donors. We also show average donation size, tenure-to-date totals, and per-election-cycle breakdowns. Chamber averages provide context.

Limitations: Under-$200 totals combine FEC’s Individual_Unitemized_Contribution figure in candidate summary reports with the gifts under $200 that a campaign lists line by line on its own reports. Members who have not yet filed for the current cycle will have their grassroots percentage based on the prior cycle only. PAC contributions are tracked separately in our PAC Money Tracker.

Data handling: FEC bulk files retain both an original report and its later amendment. Where a committee amends a report we keep the amended version of a transaction and drop the original, applying that only within a single report and only where the two filings agree on at least half of the transactions they share. We include direct (transaction type 15) and earmarked (15E) contributions per FEC Form 3 line 11(a)(i), excluding refund disbursements, candidate self-funding, conduit-side legs, and special-account contributions (Convention, Headquarters, Recount). Rows flagged by FEC as in-flight refunds or pending reattributions are also excluded. Line 11(a)(i) covers “individuals and persons other than political committees,” and federal law defines a person to include an organization, so the total also carries giving from organizations and from tribes. That money is shown as its own segment. The donation-size breakdown and the under-$200 share below are calculated on money from people only.

We track every bill a member sponsors using Congress.gov data and classify each into the highest stage it has reached: Introduced, Passed Committee, Passed Chamber, or Enacted. Chamber averages provide context; most bills never make it past committee.

Limitations: This measures only sponsored legislation. A member who works through amendments, coalition-building, or cosponsorship may appear less effective by this metric. Resolutions (ceremonial or procedural) are counted alongside substantive legislation.

Using roll call records from the Clerk of the House, we calculate each member’s attendance rate: votes cast divided by total roll calls. Any recorded participation (yea, nay, present) counts as attending. Rates are shown per congressional session with chamber averages for comparison.

Limitations: This only captures roll call votes; voice votes, unanimous consent, and committee work are not reflected. Members may miss floor votes for legitimate reasons (committee hearings, district events, medical). We present the data without judgment about reasons for absences.

Committee and subcommittee data comes from the @unitedstates/congress-legislators project. We map each committee to a related PAC funding category to surface potential conflicts of interest: for example, Financial Services maps to Wall Street PAC funding, and Natural Resources maps to Fossil Fuel.

Limitations: The committee-to-PAC mapping is editorial. Some committees touch multiple funding areas; we chose the single most relevant category for each.

Community Project Funding data comes from the House Appropriations Committee. We show each project’s recipient, description, dollar amount, and appropriations subcommittee.

Limitations: Amounts shown are requested, not necessarily final appropriated amounts. Not all members participate in the earmark process.

We show population, median household income, median rent, poverty rate, uninsured rate, and racial composition for each congressional district using U.S. Census Bureau American Community Survey 5-year estimates. This provides context about the communities each member represents.

Limitations: ACS data is based on survey sampling, not a full count; margins of error exist but are not displayed. Political composition (voter registration, partisan lean) is not available from the Census and is not included.

Each House office runs on a yearly allowance that pays for staff, office rent in the district and in Washington, travel, mail, printing, supplies, and equipment. The House publishes every payment in its quarterly Statement of Disbursements as two files: one lists each transaction, the other gives each office’s totals by category. We read both, add up the transactions for every office by program and category, and hold a refresh unless every member office’s sums match the totals file to the cent. The card shows the latest quarter the House has published for the office and the running total for that year. Session figures also include bills posted after a session ended, which the House books to the year the allowance was for. Paid internships come from a separate allowance and are listed on their own line. The House publishes each staff member’s name and pay; we show the total and the number of people paid. The full transaction detail for each office can be downloaded from the card, with staff names withheld.

District Reality indicators are informational context; they do not affect the member’s letter grade. They show the lived conditions in each congressional district so constituents can evaluate whether their representative’s voting record reflects the community’s needs.

Census ACS Demographics (CD-level)

Source: U.S. Census Bureau ACS 2023 5-year estimates.

• Population, median income, poverty rate, uninsured rate, rent burden, education — direct CD-level estimates
• Racial composition — Black, Hispanic, white, Asian, other (non-Hispanic breakdowns)
• SNAP recipients — % of households receiving SNAP benefits
• Homeownership rate, foreign-born %, citizenship %

CDHD Health Indicators (CD-level)

Source: Congressional District Health Dashboard, NYU Langone Health. 435 districts.

• Life expectancy — 2010–2015 USALEEP modeled estimate (latest available at CD level; county-level 2020–2022 data from County Health Rankings is being evaluated as a replacement)
• Uninsured rate (total & children) — includes racial breakdowns
• Cardiovascular death rate — per 100K, with racial breakdowns. Source: CDC WONDER via CDHD
• Primary care shortage — % of district population in a federally-designated HPSA. Source: HRSA via CDHD
• Child poverty, teen births — with racial breakdowns
• Opioid overdose deaths — per 100K, age-adjusted, with racial breakdowns. Source: CDC via CDHD
• Neighborhood segregation index — racial/ethnic residential segregation score

Crosswalk Indicators (county-to-CD)

County-level data crosswalked to congressional districts using HUD population-weighted allocation ratios.

• FEMA National Risk Index — composite natural disaster risk score (floods, wildfires, extreme heat, hurricanes). Source: FEMA
• Infant mortality — deaths per 1,000 live births. Source: CDC WONDER
• Drinking water violations — health-based violations from EPA SDWIS. Excludes undated records
• Police violence — rate per 100K residents killed by police. Source: Mapping Police Violence
• Voter turnout — % turnout in most recent House election. Source: MIT Election Data + Science Lab
• Incarceration rate — combined jail + prison per 100K (ages 15–64). County of commitment methodology: counts incarcerated people at the county where they were sentenced, not the prison location. Includes racial breakdowns (Black, white, Latinx, Native). Source: Vera Institute of Justice, 2019 (latest year with combined data in Vera’s public release; newer federal NCRP data is restricted-access)
• Dark money / outside spending — independent expenditures for or against each member, cumulative across their tenure (the same figure as the member Outside Spending card). Source: FEC Schedule E
• Asthma prevalence — % of adults with current asthma. Source: CDC PLACES 2023 release (2021 BRFSS survey data)

State-Level Indicators

• Maternal mortality — deaths per 100K live births. Source: CDC WONDER
• Felony disenfranchisement — estimated citizens denied voting rights. Source: The Sentencing Project (2024)
• Immigrant tax contributions — state & local taxes paid by undocumented immigrants (in millions). Source: ITEP 2024
• Anti-LGBTQ+ legislation — bills passed into law per state (2025 session). Source: ACLU Legislative Tracker

National Context Indicators

• LGBTQ+ youth suicide risk — 4x multiplier nationally. Source: CDC Youth Risk Behavior Survey (YRBS)
• Gaza casualties — Euro-Med Human Rights Monitor
• U.S. military aid to Israel — Congressional Research Service
• Post-9/11 war costs — Brown University Costs of War Project

Pending Indicators

The following indicators are configured but not yet populated with data:

• Anti-LGBTQ+ hate crimes — FBI UCR data not yet imported; sporadic voluntary reporting limits coverage
• Eviction rate — Eviction Lab national estimates end at 2018; LSC Civil Court Data Initiative (1,250 counties in 30 states) is being evaluated as a replacement

Data Transparency Note

Some indicators use older data because newer data is not publicly available. Incarceration data is from 2019 because the federal government restricts access to current county-level prison commitment data (NCRP); researchers must apply through a university portal. Life expectancy data at the CD level is frozen at 2010–2015 because no agency has updated sub-county estimates. Meanwhile, private prison companies like GEO Group and CoreCivic file detailed quarterly revenue reports with the SEC. The prison industry has full financial transparency to its shareholders and zero data transparency to the public whose communities are affected.

Because this is an issue area where the gap between established law and congressional behavior is uniquely stark. The Leahy Law (22 U.S.C. § 2378d) prohibits military aid to foreign security units credibly accused of gross human rights violations. The ICJ issued provisional measures in January 2024 requiring Israel to prevent acts that could constitute genocide under the Genocide Convention. Amnesty International, Human Rights Watch, Médecins Sans Frontières, and multiple UN bodies have documented systematic violations. Yet Congress continues to authorize unconditional military aid, while over $554M in combined contributions and independent expenditures has flowed from pro-Israel PACs to House members of both parties, including $32M in direct contributions during the 2024 election cycle alone (56 tracked PAC committees in our database).

The combination of clear legal standards being overridden by documented financial influence is exactly what an accountability platform should make visible. No other single issue area has this specific convergence of international law, PAC money, and policy disconnect.

Because they track fundamentally different institutional failures. Justice & Civil Liberties covers the criminal legal system: policing, incarceration, sentencing, surveillance, and due process. The documented disparities here are about how the state treats people after they encounter the system. Democracy & Voting Rights covers who gets to participate in self-governance: ballot access, election integrity, gerrymandering, and disenfranchisement. The documented disparities here are about who gets to shape the system in the first place.

A representative can score well on voting rights while scoring poorly on criminal justice reform, or vice versa. Collapsing them into one category would hide those divergences, and hide from voters a meaningful distinction in their representative’s record.

Our scoring is grounded in constitutional law, federal statute, international treaty obligations, and documented empirical evidence, not in progressive ideology. Examples:

• When we score “Vote No” on a bill that mandates detention based on charges rather than convictions, that’s grounded in the 5th Amendment’s Due Process Clause.
• When we score “Cosponsor” on the Leahy Law Enforcement Act, that’s grounded in existing federal law (22 U.S.C. § 2378d).
• When we score “Vote No” on Medicaid cuts, that’s grounded in CBO projections showing millions would lose coverage.

We acknowledge that some scoring decisions involve policy judgment where reasonable people disagree, and we’re transparent about which those are. Where a scoring position rests on policy judgment rather than hard legal or empirical grounding, it is labeled as such in our bill records.

These 12 represent the issue areas where all three conditions hold:

1. Sufficient legislative activity in the current Congress to produce meaningful grades (generally 13–32 tracked items per category)
2. Established advocacy organizations or legal frameworks provide a basis for identifying correct positions (e.g., ACLU scorecard for Justice, LCV scorecard for Climate, IMEU Policy Project for Palestine)
3. The gap between documented public interest and actual congressional behavior is large enough to warrant tracking

We considered and excluded categories where legislative activity was too sparse to grade, or where the policy landscape was too nascent to establish clear standards. The category list is reviewed each congressional session and may expand.

PAC contributions are presented as context, not as part of the grade calculation. We don’t claim that money causes votes; we show them side by side and let visitors draw their own conclusions.

However, the academic evidence supports paying attention to this relationship: Gilens & Page (2014) found that policy outcomes in the U.S. correlate far more strongly with the preferences of economic elites and organized interest groups than with the preferences of average citizens. Showing the money next to the votes is the simplest way to make that pattern visible at the individual representative level.

Our pro-Israel PAC funding figures come from FEC Schedule A filings: the official federal disclosure records for contributions received by candidate committees. What we label “PAC Funding” on the site includes two types of money from these filings:

1. Direct PAC contributions — money a PAC sends from its own treasury to a candidate’s campaign committee
2. Earmarked individual contributions — donations from individual donors that are routed through a PAC to a specific candidate

We also track independent expenditures (Schedule E — outside spending for or against candidates) separately on the site, but these are not included in the “PAC Funding” total.

TrackAIPAC uses a broader scope. In addition to direct PAC donations and independent expenditures, their totals include individual contributions from donors who have also made large contributions to pro-Israel PACs, even when those individual donations were not routed through a PAC. They publish their methodology on their site.

Every dollar in our PAC Funding figures is directly tied to a PAC filing. You can verify any figure against the candidate’s own records at FEC.gov.

We track what we can score rigorously. Adding a category requires sufficient legislative activity, a defensible basis for correct positions, and reliable data sources. Education, gun policy, and tech regulation are all areas where future expansion is possible as legislative activity and scoring methodology mature.

We’d rather have 12 well-grounded categories than 20 thin ones. A category that produces unreliable grades does more damage to credibility than simply not having it.

Those organizations publish single-issue scorecards that rate members on their specific policy area. We use their bill selections as methodology inputs (they help us identify which legislation matters in each issue area), but we compute our own grades using our own formula. We do not reproduce their grades.

We also track across all 12 issue areas simultaneously and overlay campaign finance data, which single-issue scorecards don’t do. Think of it this way: LCV tells you how a member votes on climate. We tell you how they vote on climate, and show you the fossil fuel PAC money, and show you that their district has high asthma rates and air quality violations.

Yes, and that’s why we don’t use it as our primary standard. Our scoring is grounded in law and evidence, not in popularity.

Where we display district-level data (uninsured rates, poverty, life expectancy), that’s factual context about the community a representative serves, not an opinion poll. We believe voters deserve to know both what the law says and what their community looks like. How they weigh those factors is up to them.