How the Bluff & Spin Index works

0 to 100. Lower is better. 0 means every checked claim held up and every question got a straight answer; the more bluff and spin, the higher the number. The weights are fixed and the same for every politician: a measure without enough data never hands its weight to another. Consistency and Substance are shown on each page but are not part of the number. Any change creates a new methodology version.

The same fixed weights for everyone. Consistency and Substance are shown alongside, never in the number.

Nobody picks which transcripts count. The same rule runs for every politician (selection rule v5, applied 8 Oct 2026):

1. Transcripts from the 24 months before that date in which they say at least 30 sentences.
2. Question-and-answer first. The newest journalist-led interview: the interviewer is a professional journalist (has worked as one for a news organisation), on any outlet including their own channel or podcast; not a conference, a book launch, Parliament, or a show hosted by a politician or campaigner. Then other transcripts where they are asked at least 5 questions, newest first, until there are at least 2 and 60 questions in total (at most 6).
3. Balance: a friendly interview counts only when there is at least one challenging one for it. Who asks decides: questions from opponents at PMQs, the press, committees and journalists are challenging; an event conversation or a show hosted by someone on the guest's side is friendly.
4. At most 2 question-and-answer transcripts from the same day: a media round repeats the same lines to several outlets in one morning, so the rule takes the newest 2 and moves on to other occasions.
5. Plus the newest transcript without questions (a speech, statement or debate), so prepared words are tested too.
6. The same event saved twice: the official record wins over a broadcaster, party or third-party transcript, and those win over video captions.

Approval only checks that a transcript is what it says it is (the right speaker, the complete text). Transcripts the rule does not choose stay in the record and on the site, marked as not counting.

• Provisional means fewer than 30 checked claims so far.
• Every politician's number covers both halves of the question: do the claims hold up, and do they answer what they are asked. Speeches alone cannot show the second, so a record without enough questions shows its findings but no number yet.
• Behind every number is a likely range (90%), from resampling the transcripts and the claims and questions inside them. It is in About this score on each politician's page; where two ranges overlap, our evidence cannot yet tell those two scores apart.
• Transcripts marked Awaiting scoring are saved but not yet counted.