The IQ Methodology

The Influence Score covers 643 of the 650 sitting MPs. The seven not scored are the abstentionist Members who do not take their Commons seats and sit in no party group — so the role, division and committee signals the score depends on never register for them. We exclude them rather than publish a number built on absent data.

A score is only as good as its sanity checks. Every time the model is rebuilt, it is tested against a hand-checked reference set of well-known figures — the Prime Minister and Chancellor, opposition leaders, long-serving backbenchers, and high-profile figures who hold no formal office. The test is not whether each lands on an exact number, but whether the ordering is defensible: senior office-holders should sit near the top, and a backbencher should not outrank the Cabinet. If the model ever inverts that, the check fails and we investigate before anything publishes.

Two things the score deliberately does not do. It does not reward visibility for its own sake: a figure with a large media presence but no formal position — Nigel Farage is the standing example — scores below where name recognition alone would put him, because the Influence Score measures structural power, not fame. And several dimensions, media reach among them, are measured as exposure relative to a typical baseline, so they can read above 100% for the most-covered figures; the headline score weights them, it does not let them run away with it.

We are candid about the edges. Positional weight currently leans toward government roles, and we are refining how opposition front-bench positions are credited. Attention is measured at party level rather than per individual. These limitations are documented, reviewed by analysts, and corrected over time — the algorithm is audited, not assumed.

The Influence Score is algorithmic: no figure is nudged by hand to flatter or to punish, and the weights are fixed in advance — not tuned to produce a preferred result. But an algorithm inherits the quality of its inputs. Source data can be stale, a role can be miscoded, a relationship can be missed. So every published cycle is reviewed by analysts who verify edge cases and flag anomalies before it goes out, and the model is tested against the calibration set above.

That review will not catch everything. If you believe a score, a relationship, or an underlying fact is wrong, tell us — we will check it, and correct it if you are right. Write to operations@the-iq.com with the figure and what looks off. Corrections to the data feed through at the next scheduled update.