How VoteSee Calculates Its Numbers

VoteSee turns official congressional records into profiles, report cards and election pages for every member of Congress. This page explains where each number comes from and how it is computed, so you can check any figure you see on the site.

VoteSee does not endorse candidates or parties. Every score on the site comes from recorded votes, bill records and public filings, never from a member's statements.

Last reviewed Oct 13, 2026.

Roll-call votes

House roll-call votes come from the Clerk of the U.S. House (clerk.house.gov). Senate roll-call votes come from the U.S. Senate (senate.gov). VoteSee checks for new votes every 6 hours.

Each vote page shows how every member voted, a party-by-party breakdown, and the members who voted against the majority of their own party on that vote.

Bills, sponsors and summaries

Bill titles, sponsors, cosponsors, status, subjects and official summaries come from the Library of Congress bill status data published on govinfo.gov, the same records behind congress.gov. VoteSee refreshes this data daily.

Official summaries are written by the Congressional Research Service (CRS). Some vote pages also show a short "In plain English" summary. That summary is generated by an AI model (Google Gemini) from the CRS summary only, and it is labeled as a plain-English summary.

Attendance

Attendance is the share of roll-call votes a member took part in during the current Congress: votes cast (Yea, Nay or Present) divided by votes cast plus votes missed. A vote counts as missed when the member is recorded as Not Voting.

The votes counted are roll calls on bills that received a passage or suspension vote, including amendment and cloture votes on those bills. Nomination votes and quorum calls are not counted.

Attendance grades use fixed cutoffs: A is 98% or higher, B is 95% or higher, C is 90% or higher, D is 85% or higher, and F is below 85%. Attendance is recalculated nightly.

Votes with party and bipartisanship

For each passage, amendment and suspension vote, VoteSee finds how the majority of each party voted, counting only that party's Yea and Nay votes. Votes where a party split evenly are skipped.

A member's "votes with party" rate is the share of their own Yea and Nay votes that matched their party's majority. Absences and Present votes are left out. Independents are compared with the party they caucus with. The rate covers every such vote VoteSee holds for the member, not only the current Congress. Bipartisanship is 100% minus that rate.

The "broke with their party" lists on member pages show recent votes, from 2025 onward, where the member voted against their party's majority.

Legislative effectiveness

Effectiveness is the share of bills a member introduced as primary sponsor that were enacted into law, using the bill status records described above.

Member pages also show bills sponsored or cosponsored and how many of those became law. Those counts include cosponsorships.

Report card grades and rankings

Every current voting member with at least 100 recorded votes is ranked against all other current voting members, House and Senate together, on attendance, bipartisanship and effectiveness. Each ranking becomes a percentile: the share of ranked members who scored worse.

The overall report card is a weighted average of those percentiles: attendance 40%, effectiveness 35% and bipartisanship 25%. If a member is missing one ranking, the other weights are rescaled. Percentile grades: A is the top 20%, B the next 20%, then C, D, and F for the bottom 20%.

Non-voting delegates from D.C. and the territories are not ranked on attendance or bipartisanship, so their overall grade reflects effectiveness only. The same is true for members with fewer than 100 recorded votes, such as newly sworn-in members, until they reach 100 votes. Member summaries and election pages only quote an overall grade for members ranked on attendance or bipartisanship.

Issue positions (the sliders)

Each issue, such as guns or immigration, has two ends of a spectrum. Bills are assigned to issues using the Congressional Research Service's subject terms and policy areas.

Bills are placed along each issue's spectrum by comparing pairs of CRS bill summaries with an AI model (Google Gemini) and updating an Elo-style rating, like a chess ranking. The AI only places bills on the spectrum; it does not decide which bills belong to an issue.

A member's score on an issue averages the scores of the bills they voted Yea on at final passage or sponsored (sponsorship counts three times as much). Scores are then standardized each year across all members and shown on a 0 to 100 scale. Nay votes do not move the score. Bill scoring is currently paused, so recent bills may not have a position yet.

Party alignment chart

The analytics chart labeled "Majority Alignment" shows the same votes-with-party rate described above: how often a member votes with the majority of their own party. It includes members with at least 50 recorded votes, so it can rank a few members the report card does not.

Stock trades and trade-conflict flags

Stock trades come from the periodic transaction reports members file under the STOCK Act: House reports from the Clerk of the House and Senate reports from the Senate's electronic financial disclosure system. VoteSee reads these filings daily. Handwritten or scanned paper filings are not included.

VoteSee flags a trade for three reasons. Committee: the company is in an industry overseen by a committee the member sits on, for trades since January 3, 2025. Related vote: the member voted on a bill in that industry's policy area within 30 days before or after the trade (passage, cloture and veto-override votes only). Late filing: the trade was reported more than 45 days after it happened, past the STOCK Act deadline.

Companies are matched to industries using GICS sectors for S&P 500 companies and SEC industry codes for others. Treasuries, municipal bonds and broad funds are excluded from the committee and vote checks. A flag is not an accusation: many trades are made by financial advisers or by a spouse. Flags are recalculated weekly.

Campaign donors and money by sector

Donor data comes from Federal Election Commission bulk files and covers contributions from PACs and party committees to a member's campaign. Itemized contributions from individuals are not included.

FEC filings do not classify donors by industry. VoteSee groups PACs into industries by matching keywords in their names, and PACs that match no keyword are left unclassified. A member's top industry and industry breakdown count only classified PAC money for the cycle shown, and leave out leadership and joint-fundraising PACs. The "Money by Sector" analytics use the same groups.

2026 election pages

Candidate lists for the Nov 3, 2026 general election come from state certified candidate lists as compiled on Wikipedia, checked against Federal Election Commission filings. Where Wikipedia has no list yet, VoteSee lists FEC-registered candidates who have raised at least $5,000.

Money raised and cash on hand come from each campaign committee's latest FEC report, and every figure shows the date that report covers. Incumbents' records (report card grade, attendance and votes with party) come from the same calculations described above. Election data refreshes daily.

The ZIP code and address lookup uses the Geocodio geocoding service to find the 2026 congressional district.

2026 polls and results

Polls come from the Polling sections of each race's Wikipedia article, refreshed daily. VoteSee keeps only general-election polls whose candidates are all on the ballot, so primary and hypothetical matchups are left out. Each poll links to its source; see each pollster for full results.

The poll average is a simple mean of polls whose fieldwork ended in the last 30 days, one per pollster (its most recent). When a pollster reports both likely voters and registered voters, the likely-voter figure is used. With fewer than two such polls, VoteSee lists the polls without an average. Averages describe past polls only. They are not forecasts or probabilities.

The national generic congressional ballot figure is the mean of the polling averages listed on Wikipedia that were updated in the last 30 days, since Wikipedia does not list the individual polls. Presidential approval uses the same 30-day, one-per-pollster average as races.

Election results come from Wikipedia's race tables, which editors update as races are called, checked every 15 minutes on election night. A race shows a winner only once it is called, and VoteSee can correct a call by hand. Louisiana's Nov 3 House vote is an open primary with a Dec 12 runoff when no one wins a majority. Georgia holds a Dec 1 runoff under the same rule. California and Washington use top-two primaries, so both November candidates can be from one party.

Quiz matches

The quiz compares your answers with members' issue scores. Your match percentage with a member is 100 minus the average distance between your answer and the member's score on each issue you answered. Members need a score on every issue you answered to appear.

Members of Congress

Member names, terms, districts, party and identifiers come from the public congress-legislators dataset maintained by the @unitedstates project, which is built from the Biographical Directory of the U.S. Congress. Committee assignments come from the same project. Photos come from the Biographical Directory or Wikipedia.

District pages use U.S. Census Bureau data for the current (119th Congress) boundaries: the 2024 cartographic boundary file for the map and neighboring districts, 2020 Census population for counties and communities inside each district, and American Community Survey 2020-2024 estimates for demographics. The list of past members covers everyone who held the same seat number. District lines are redrawn after each census, so older holders may have represented a different area.

Known limits

Party is each member's current party, applied to all of their past votes. A member who switched parties is compared with their current party throughout.

"Data as of" dates on member pages show the newest roll-call vote VoteSee has loaded for that chamber. Former members' pages show their last day in office.

If you find an error, email [email protected] and we will correct it.

Sources