Methodology

We cover the U.S. House and Senate from the 117th Congress (January 2021) onward. On each roll-call vote, a member’s position is one of Yea (the House records “Aye” on some votes, which we treat the same), Nay (or “No”), Present, or Not voting.

We sort every roll call into one of five types:

Quorum calls and the election of the Speaker of the House aren’t counted in any statistic. Quorum calls only record who is present, and Speaker elections are votes for a person rather than yea or nay. We still record them: for a Speaker election, that’s who each member voted for.

Which types count toward each statistic:

A member is only counted on votes held in their chamber while they held the seat, from the day they were sworn in to the day they left. A member who changes parties is counted with the party they belonged to on the day of each vote. Independents are counted with the party they caucus with.

We use the definition Congressional Quarterly has used since the 1950s. A party-unity vote is a roll call where a majority of voting Democrats voted one way and a majority of voting Republicans voted the other way. Only Yea and Nay votes decide the majorities. If either party splits exactly in half, it isn’t a party-unity vote.

A member’s party unity score is how often they sided with their own party on those votes:

Party unity = party-unity votes where the member voted with their party’s majority ÷ party-unity votes where the member voted Yea or Nay

Present and Not voting are left out of both numbers, so missing a vote doesn’t raise or lower the score. All five vote types count, as in CQ’s scores; on member pages you can narrow the score to one type. Each score shows how many votes it’s based on.

We report attendance as the share of votes a member missed:

Missed votes = votes where the member was Not voting ÷ votes the member was eligible for

Voting Present counts as voting, not as a miss. In the House during the 117th Congress, votes cast by proxy count as votes, because the Clerk records them as the member’s position.

The official record doesn’t say why a member missed a vote, so absences for illness, family leave or official travel count as misses like any other. By tradition, the Speaker of the House rarely votes, so the Speaker’s missed-vote rate is shown with a note saying so.

For each issue area, such as healthcare or defense, we count how a member voted on roll calls about bills in that area:

Yea rate = Yea votes ÷ votes the member was eligible for (and likewise for Nay, and for Present or Not voting)

The main figure uses final passage votes only. Amendment, cloture, motion to recommit and procedural votes are shown separately, broken down by type, because a Yea on them doesn’t mean support for the bill. A Yea on a motion to recommit or a motion to table, for example, is often a vote against the bill.

A bill can be in more than one issue area, and its votes count in each, so the numbers across issue areas don’t add up to a member’s total. Votes not tied to a bill, such as most nominations, have no issue area and aren’t included here, though they still count toward party unity and attendance.

These numbers describe votes, not positions. A Nay on a healthcare bill isn’t a vote against healthcare; members vote against bills for many reasons, including that a bill doesn’t go far enough.

A member’s votes can’t measure bipartisanship separately from party unity: on a party-unity vote, siding with the other party is the same as breaking with your own. So we measure it through the bills members put their names to, with two numbers:

Cross-party cosponsoring = bills the member cosponsored that were introduced by a member of another party ÷ bills the member cosponsored

Cross-party support = bills the member introduced that have at least one cosponsor from another party ÷ bills the member introduced

Both count bills and joint resolutions introduced in each Congress. Original and later cosponsors count alike, and cosponsorships that were withdrawn don’t count. Each member is counted with the party they belonged to on the day they put their name on the bill, and independents with the party they caucus with. Each number links to the bills it was calculated from.

Two members can be compared when they served in the same chamber in the same Congress, since only then did they vote on the same roll calls. How often they voted the same way is:

Voted the same way = votes where both voted Yea or both voted Nay ÷ votes where both voted Yea or Nay

As with party unity, Present and Not voting are left out of both numbers: a vote where either member was Present or didn’t vote counts as neither agreement nor disagreement. All five vote types count, and the comparison page can narrow the number to one type. It lists every vote where one member voted Yea and the other Nay, and shows each member’s other numbers side by side.

Agreement describes votes, not views. Members of the same party usually vote the same way far more often than members of different parties, so the number is most telling next to party unity.

New votes usually appear the day after they’re held. We keep a copy of every file we download, with the address it came from and when we fetched it, and every record on the site can be traced back to the file it came from. As we add sources, such as campaign finance and state legislatures, they’ll be listed here.

The Library of Congress’s Congressional Research Service (CRS) gives each bill one policy area and a list of more specific legislative subjects. We map those onto our own, shorter list of issue areas, such as the environment, education, healthcare, the economy, defense and immigration. Votes on a bill, including votes on its amendments, take the bill’s issue areas.

A bill gets the issue area its policy area maps to. Some subjects add a second issue area when a bill crosses into it: a veterans’ health bill has the policy area Armed Forces and National Security, which maps to defense and veterans, and its subject Veterans’ medical care adds healthcare. This is the full mapping:

These policy areas are too narrow or too broad to say much about a member’s votes, so they don’t give an issue area on their own: Arts, Culture, Religion; Emergency Management; Private Legislation; Social Sciences and History; Sports and Recreation. A bill in one of them can still get an issue area from its subjects.

CRS analysts often add subjects weeks after a bill is introduced, so issue areas for recent votes may fill in later.

Later, for bills that CRS doesn’t cover, such as state legislation, we plan to use AI to suggest issue areas. A person will review every suggestion before it’s published, and those issue areas will be labeled as AI-suggested and reviewed. Every issue area shows where it came from: CRS, another named source, AI-suggested and reviewed, or our own editors.

We never rate members as for or against a cause, and we don’t give grades. Ratings like that appear on this site only when they come from a third-party scorecard, such as those published by advocacy groups, and they’re always labeled with:

• the organization that made the scorecard, and the year it covers;
• a link to the organization’s own explanation of how it scores members;
• the score exactly as the organization published it.

We only show scorecards that publish their method and list the votes they scored. Their ratings are never mixed into our own statistics. Where we show scorecards on an issue, we include organizations with different points of view when they exist.

Plain Votes is free and has no ads. Readers who find it useful can support this site with a contribution, which pays for hosting, data and development.

Contributions don’t buy any say over what we publish or how our statistics are calculated. Every number on the site follows the definitions on this page, whoever has contributed.

If you think something on the site is wrong, use the “Report an error” link on any member or vote page, or the error report form. Tell us which page, what you think is wrong, and, if you can, where you found the correct information. The same form works for any other question about the site.

We check every report against the official record. If our copy is wrong, we fix it and add a dated note to the page it affected saying what changed. If the official record itself is wrong, we keep showing the official record and add a note explaining the problem. When a member states in the Congressional Record how they meant to vote, for example, we note it but don’t change their recorded vote.

Every correction, and every change to this methodology, is listed here with its date.

Corrections and changes

None yet.