Straight To The Point News

Political Rating Methodology

A political classification is not a measure of accuracy. A label describes a publication’s overall tendencies, never an individual article, and “Center” does not mean perfectly factual or objective.

The five classifications

STTPN uses five human-approved classifications, plus Unrated when the evidence doesn’t justify one. They are descriptive editorial judgments with limitations: STTPN’s opinion based on the evidence described below, not statements of fact about any publication or journalist.

LeftA strong, consistent preference in political story selection, framing, source choices or editorials associated with left-leaning politics in the publication’s market.
Lean LeftA noticeable but more moderate or less consistent preference in that direction.
CenterThe reviewed evidence doesn’t show a consistent enough left or right pattern to justify another classification. It is not a guarantee of neutrality or accuracy.
Lean RightA noticeable but more moderate or less consistent preference associated with right-leaning politics.
RightA strong, consistent preference associated with right-leaning politics in the publication’s market.

The five classifications appear on the Publications directory and each publication’s page. Coverage meters on stories and videos use three parts, so Lean Left counts as Left and Lean Right counts as Right there, and Related Coverage lists show the same three-part label.

What “Unrated” means

Unrated is not Center and not another political position. It means the evidence is insufficient, contradictory or out of date, the publication’s identity is uncertain, or a human review hasn’t been completed. Publications stay Unrated until the editor approves a classification; nothing is ever assigned automatically.

When STTPN changes its scale, earlier labels are not converted. On October 26, 2026 STTPN moved from three classifications to these five, and every publication returned to Unrated until it is reviewed again under this methodology.

How a review works

Confirm the publication. We verify its editorial identity, current websites and market.
Sample its coverage. A first review examines about 10–12 recent items from the past 90 days, across at least three political subjects. Only the publication’s own work counts; wire stories it republishes from other agencies don’t. News is recorded separately from opinion and editorials, and one unusual article is distinguished from a persistent pattern.
Weigh the evidence. We look at story selection and prominence, framing and word choice, the kinds of sources quoted, and repeated editorial positions, and we record contrary evidence and limitations.
Human approval. AI tools may help organize research, but the editor inspects the evidence and must explicitly approve every public classification.
Publish the profile. Rated publications get a public page with the classification, review date, a short evidence-based explanation and the next review date.

Ten to twelve items is a practical starting sample, not a statistically validated one. We add more when the evidence is mixed or a rating is challenged. No numeric bias scores or percentages are produced for a publication.

Reassessment and challenges

Classifications are reviewed at least once a year, and sooner after material editorial or ownership changes or a credible challenge. While a reassessment is under way, the last approved classification and its true review date stay visible until a new decision is approved, or it is withdrawn to Unrated. Every decision is kept in a dated, append-only private history.

Anyone can challenge a classification with supporting evidence. There is no public voting, and donations, sponsors and advertisers have no role.

How the coverage meter works

For each story, the editor selects related reporting from other news publications (“Related Coverage”). The meter on the story shows how those selected publications are classified:

Each distinct publication counts once, even if we link to several of its articles.
“X sources” is the number of distinct related publications, including Unrated ones. It is not a count of the sources STTPN used for its own reporting; those are linked within each story's text.
Percentages use only the rated publications: Left (including Lean Left) ÷ rated, Center ÷ rated and Right (including Lean Right) ÷ rated, rounded so the three always total 100%. Example: 4 Left, 3 Center and 3 Right gives 40% / 30% / 30%; one more Unrated publication makes it “11 sources” without changing the percentages.
Percentages appear only when at least three distinct rated publications are linked. Otherwise the story shows “Limited rated coverage”.
The meter always reads Left (blue), Center (white, outlined), Right (red), in every country, and each segment is labeled in text.

The meter is not a rating of STTPN’s article, not a measure of accuracy and not a survey of all media coverage. Which links the editor selects affects the result, so please read the linked reports and the publication pages.

How the 2026 midterms map works

Our limited-time midterms map colors each Senate and governor race by an average of recent independent polls. The polls come from the race’s article on Wikipedia, whose editors list each poll with its pollster, dates, sample and margin of error and link it to its source. Our server reads those lists about every two hours; the founder can hide a poll that’s wrong or add one that isn’t listed yet. Polls are snapshots of opinion at the time, not predictions by Straight To The Point News.

Which polls count: polls that finished in the last 45 days, one per pollster (its newest; likely voters preferred), up to the six newest. Polls Wikipedia marks as partisan, or that a campaign, party or PAC paid for, aren’t counted, and neither are very small polls (under 300 respondents, or a margin of error over 7 points).
Colors: a lead smaller than the polls’ typical margin of error is a toss-up. Beyond that, a race is Lean (a lead under 5 points), Likely (5–10) or Solid (10 or more) for the leading party. Gray means no recent independent poll; the lightest shade means no race this year.
House: we don’t track polls for the 435 House districts. A state takes the color of the party that has won more of its called seats.
Results: after polls close on Election Day, races show as won once STTPN marks them called, based on major news organizations’ race calls. Until a race is called it shows as “Not called yet”; its state panel keeps the final pre-election polling average for reference.

Limits

These are qualitative judgments made by a small, founder-run publication. Human judgment and sampling limits can’t be eliminated. STTPN controls its own methodology and classifications and doesn’t reuse other organizations’ bias ratings.