Methodology

The population basis

Voting-age population (VAP)

Everyone 18 and older, from the Census Bureau’s American Community Survey (ACS). Not total population, and not registered voters. A state with a large non-citizen or non-registered population is credited with everyone 18 and up, because the figure measures who a district or state’s seat represents, not who voted. Voting-age population is the Census Bureau’s term; everywhere a reader meets a figure, the site says “U.S. adults”, the same population in plain words. Who is included lists who that population counts and who it leaves out.

Which year’s population

Every vote uses the ACS release for the year the vote happened, or the most recent release available for that year. A vote from 2005 is computed against the 2005 ACS one-year estimates (about 214.8 million adults in the fifty states), and a vote from 2026 against the 2024 ACS one-year estimates (about 266.6 million). Each vote page states its own vintage, so two votes from different years are never sliced against the same population figure.

Data window

The dataset covers 2005 to the present, which is where the official sources begin to align. api.congress.gov’s House vote data goes back to 2017; the Clerk of the House’s XML archive extends that to 2005; senate.gov’s roll-call archive starts in 1989; and the Census Bureau publishes no congressional-district population by age, at ACS quality, before 2005. Every figure in the window uses the same method. No earlier decade is estimated on rougher data.

District map vintage

District populations come from the ACS release used for the vote, which follows whichever congressional district map was current when that release was published. A district redrawn after that map (redistricting after a census, or a court-ordered map) shows an approximate population until the next release catches up. Each vote page states which Congress’s district map its figures used.

How the House figure is computed

Every voting House district is assigned to the side its representative voted with. The House figure for a side is the sum of the adult population of every district that voted that way, divided by the adult population of the fifty states.

A representative who was absent or voted “present” puts their district’s population on neither side. That population is reported as unrepresented on the vote rather than folded into either side, since assigning it would inflate that side’s share.

How the Senate figure is computed

Each state’s adult population is split evenly between its two Senate seats. If both senators vote the same way, that side receives the state’s full population. If they split, each side receives half, regardless of the margin within the state, because a Senate seat carries no population weighting of its own.

A senator who was absent or voted “present” puts half their state’s population on neither side, by the same rule as the House.

Territories

Washington, D.C., Puerto Rico, Guam, the U.S. Virgin Islands, American Samoa, and the Northern Mariana Islands sit outside both the numerator and the denominator of every figure. Their delegates cannot cast a vote that counts toward final passage, so there is no side for their population to be assigned to. Including them would either add population with no vote to the denominator or require assigning a side to a vote that was never cast.

When the smaller number wins

Some votes need more than a simple majority. Cloture, the motion to end debate in the Senate, needs 60 of 100 votes; a motion can get 55 yea against 45 nay and fail, and the side that won is the 45. The winning side is determined from the chamber’s official result wording (“Passed,” “Cloture Motion Rejected,” and so on), not by comparing the tallies, which would invert every supermajority vote. A vote decided by a threshold other than a simple majority carries a note saying what was needed, on its card and on its page.

What counts as a vote here

“Vote” on this site means one entry in the list: a measure the House, the Senate, or both took a recorded position on. Where both chambers voted on the same bill, that is one entry with two chamber figures behind it, not two entries.

The unit the government publishes is the roll call, a single chamber’s recorded vote on a single question. A bill can collect many of them. This site keeps the most recent roll call per chamber, the one that decided the measure in that chamber, so the 12,391 votes here sit on top of a larger number of roll calls, each linked from the vote page it belongs to.

Roll calls that never carried a title of their own, mostly procedural motions such as approving the Journal, are excluded from the list and from that count.

Two ways a vote can be unrepresentative

These are different claims and the site keeps them apart. The first is that the winning side represented under half of U.S. adults, which is true of 2,334 of the 12,391 votes here. The second, and stronger, is that the winning side represented fewer people than the losing side, which is true of 1,556. Every vote in the second group is also in the first; the reverse does not hold.

The difference is absence. A member who does not vote puts no population on either side, so both shares can fall short of a majority at once. A side can win holding 45% of the country while the side it beat held 40%: under half, and still the larger of the two.

The first condition is what the red margin star on a vote card marks, and it is the “won with less than a majority” switch in the list’s controls. The second is not shown as its own mark; the two figures on every card and page carry it.

When the two chambers disagree

A measure voted on by both chambers receives a combined figure: the average of what one side represented in the House and in the Senate. When both chambers picked the same winner, that side is the winner and the average is straightforward.

When the chambers disagree, the combined view follows the yea side, averaging what the yea side represented in each chamber against what the nay side represented in each. It does not average the House’s winner with the Senate’s winner, because on a measure the House passed and the Senate rejected those are opposite sides and their average describes nothing. The combined number is a measure’s support across Congress, not a floor outcome. Each chamber’s own figure is available on the vote page through the chamber toggle, and that figure is never an average.

How to check the figures

Every vote page links to the chamber’s own official record for that roll call, on senate.gov or clerk.house.gov, which is the document this site read. If a name, a position, or a result differs from what is shown here, the official record is correct and this site has an error worth reporting.

Every vote page also states the population vintage and the district map its figures used, so a figure can be recomputed from the same Census release. The full list of sources, with the exact endpoints and releases, is on the Sources page.

Where the data does not support a claim, no claim is made. A member whose seat was vacant is counted as a vacancy, not as an absence; a bill with no official summary shows no summary; a measure decided by a supermajority threshold says so; and no figure is extrapolated before 2005, where the sources stop.

Independence

Spoken For is an independent reading of the public record. It is not affiliated with, endorsed by, or produced by Congress, the Government Publishing Office, the Census Bureau, or any other government body. The typographic conventions of official printing are used as a setting; no seal, insignia, or official title is used.

What this site owes to GovTrack

GovTrack.us has published the Senate-side version of this calculation (each state’s population split across its two senators) on its Senate roll-call pages for years, and has written about the structural finding this site is built around. GovTrack has no House-side equivalent, so it shows neither the House figure, the combined-chamber view, nor a chamber toggle. This site uses none of GovTrack’s data, API, or visual language; every figure here is computed independently from api.congress.gov, senate.gov, and Census ACS data, as described above.

A disclosure

I used Claude for some of the data gathering and for designing the experience. The rules above, and the figures they produce, are checkable against the official records linked from every vote page.