Methodology

How a name gets a place on the chart.

Leaders and presidents are placed from the public governing record. Sitting members of Congress are placed from roll-call scores. The 2026 chart uses party programs, not votes. The weights behind a reader’s own percent are not published.

Uncanted uses two axes for every chart. The horizontal axis is economics. Left is a larger public role in taxes, ownership, trade, and welfare. Right is more room for markets, private ownership, and trade. The vertical axis is authority versus liberty. Up is a greater willingness to use the state on borders, speech, policing, faith, or emergency power. Down is more restraint on those same questions. The scale runs from −100 to +100. The middle is zero. Neither direction is marked as the correct answer.

These two numbers are a compression. A war, a court, a tariff, and a civil-rights statute cannot all remain visible once they have been folded into a point. The charts exist so a reader can see direction. They do not exist so a reader can skip the record. Every written page on this site says what was used, and what was refused, for that name.

World leaders are placed by an editorial reading of the public governing record. The file is heads of government, plus a few people who never held the office but bent it. Each name has a short placement note: what the government did on the questions the axes can actually see. Where the site has it, a second note separates what was said in public from what was then done. The dot follows the governing record when those two pull apart. A slogan is not a placement.

That reading is editorial. It is not a secret formula, and it is not a scrape of every speech. Two people can disagree about how far left Clement Attlee sits and still agree that the National Health Service and the nationalizations belong on the public side of the economic axis. The page for each leader quotes the placement note, gives the two numbers, and names the nearest other figures on those numbers. Nearest means nearest on the chart. It does not mean the governments were allies.

A few names are marked as historical extremes. They are plotted so the corners of the scale are not empty. They are not recommendations. They stay off the default nearest list unless one of them is closer than every ordinary figure. The mark is a warning about the scale, not a thrill.

U.S. presidents are placed the same way: from the public governing record, one person once. Grover Cleveland’s split terms are one dot. The party color is the party while in office. Early presidents are not given today’s party names. Parties before the Civil War use a different color so the chart does not pretend that a Jeffersonian and a modern Democrat are the same organization.

Presidents are not placed with the roll-call score used for the sitting Congress. A nineteenth-century administration does not have a Voteview file that matches the 119th Congress, and Uncanted does not invent one. A president can sit left of someone from the other party. Theodore Roosevelt sits left of Grover Cleveland on economics. The color and the place are allowed to disagree, because that disagreement is part of what the chart is for.

Former members of Congress are a separate chart of one hundred better-known names who have left the House or the Senate. Anyone still seated is on the current Congress chart, not this one. Their dots are editorial, from the public career, in the same spirit as the presidents. They are not given a modern NOMINATE score from a Congress they are not in. Henry Clay’s American System is not a 2026 roll call. The page says so.

The current Congress is different, and it should be read differently. Sitting voting members of the 119th Congress are placed from Voteview’s DW-NOMINATE scores, snapshot October 2026. The first dimension, which runs from liberal to conservative in Voteview’s own description, is drawn left to right and scaled by one hundred so it shares a page with the other charts. The second dimension is drawn on the authority axis and scaled the same way. Voteview’s second dimension is not a pure measure of borders, speech, policing, faith, and emergency power. Uncanted uses the public number rather than inventing a cleaner one, and the member pages say that in plain words.

The source for a member’s place is the roll calls, not a news ranking and not the campaign site. Party is the color of the dot. In this Congress the two major parties do not cross on the first dimension: every Democrat’s score is still left of every Republican’s. The split worth looking at is the independents, and the members whose label and voting record are under pressure. Sanders and King vote with the Democrats. A member who leaves a party does not have their old votes erased. Nothing was removed to force the parties apart.

Some scores are marked estimated. A personal roll-call file was not solid enough, in the October 2026 snapshot, to treat the number as finished. Those pages say the score is provisional. They do not dress an estimate up as a completed career.

Where Uncanted has a sentence from the member’s own official House or Senate website, fetched on October 16, 2026, the member page quotes it and links the page it came from. Those are the member’s words. They are not rewritten, and they are not treated as the same evidence as the roll-call score. A website can feature a bill the votes do not. Where there is no such sentence, the page does not invent a biography, a committee story, or a list of key votes. The office, the party, the two scores, the rank inside the chamber, and the nearest other scores are the whole of the text.

Rank is a count, not a judgment. A representative who is 40th from the left in the House is more liberal on Voteview’s first dimension than the members to their right, and more conservative than the members to their left. The party median is the middle first-dimension score of that party in that chamber. A member can sit left or right of their own party without changing the color of the dot. The nearest names are the smallest distances on the two scores. They are not a slate, and they did not necessarily vote together on any one bill.

The 2026 federal-elections chart is not a roll-call chart. It keeps at most two names in each House or Senate race, preferring the Democrat and the Republican when both are on the public list. No polling percent is invented to choose them. A dot on that chart is the public program Uncanted assigned to the party line, not a personal voting record. Challengers who have never served do not receive a fake NOMINATE score. An incumbent who is also in the Congress file has a separate page, and that page is the vote-based one. The two charts answer two different questions. Mixing them would hide the difference between a promise and a record.

Country averages are the mean of completed quizzes from accounts that chose that country. Only the first finished quiz on an account is included, so a person cannot pay for an additional quiz and move the national dot. The average is not a national poll. It is not weighted to the census. It appears only after a reader has unlocked it. The free atlases can say what the average is. They do not give the coordinates away.

A reader’s own placement comes from the thirty-question test. The site does not publish the weights behind that placement or behind a similarity percent. Those are the product. What the reader can see without them is the direction of the two axes, the records the charts are built from, and the limits of a two-number summary. A high percent is not a moral likeness. A nearby candidate is not an instruction to vote.

The limits are the same on every page. Two axes drop issues that do not fit them. A single decisive question, a district, a court, or a war can matter more to a voter than the distance on this map. Editorial placements for leaders, presidents, and former members are judgments about the public record, and a reader who thinks a number is wrong can say so. Roll-call scores move when new votes are cast. The Congress snapshot is dated for that reason. Official-site quotations can go stale. None of this is a news article, and none of it is a substitute for the Congressional Record, congress.gov, or the government’s own papers.

The charts are on the atlases. A longer reading of each choice is in the articles.