Comparison and scorecard methodology

The scorecard covers exactly six dimensions and never combines them. It is informational and is never an endorsement, a recommendation or an official result.

• Fiscal stewardship
• Delivery & resource efficiency
• Administrative discipline
• Education & social policy
• Manifesto credibility
• Integrity & accountability

• A score of 1 to 5 is authored by a person, never computed, and cannot be stored without an evidence summary, a rationale and at least one dated source.
• Not assessed is not zero. It is never averaged, never counted against a candidate and never quietly counted in their favour.
• Confidence is recorded separately and qualifies a score without adjusting it. Insufficient confidence means no number is shown.
• Public comments and reactions never affect any dimension, any confidence level or anything else the service reports as fact.
• A scorecard appears publicly only when the master scoring switch, the Beta switch and the methodology approval all permit it. With any one of the three off, none is shown.

Think an assessment is wrong? Send feedback.

A comparison places candidates for the same seat and the same election side by side, showing the same published information for each: party, candidacy stage, source coverage and the record we hold. Only direct opponents are compared; candidates from different contests are never presented as alternatives to one another.

Comparison does not require scoring. Most contests carry no scorecard, and those comparisons are complete rather than degraded: the evidence is the point, and the page states plainly where a structured assessment has not been published for a contest.

Where information is missing it is shown as missing, in the same words everywhere, rather than left blank for the reader to fill in with an assumption.

Every published claim traces to a source with a publisher, a title, a link where one exists and a publication date. Dates carry real weight here, because party affiliation, candidacy and officeholding all change, and a source is evidence only for the period it describes.

A higher source count means more independent material has been logged for that record. It does not mean the candidate is better, and it never raises a dimension score on its own. Reading it as a comparative judgement is the most common mistake, and the page is written to discourage it.

The scorecard assesses six dimensions and only these six: Fiscal stewardship, Delivery & resource efficiency, Administrative discipline, Education & social policy, Manifesto credibility, Integrity & accountability.

A dimension may carry a score of 1 to 5. That number is authored by a person against dated evidence, not calculated from inputs, and it cannot be stored at all without a short evidence summary, a full rationale and at least one dated source. Where the evidence does not support a number, the dimension is recorded as not assessed.

Not assessed is not zero. It is not averaged into anything, it is not counted against the candidate, and it is not quietly counted in their favour either. It means exactly what it says: we looked, and the available evidence does not support a score.

Confidence is recorded separately from the score, as high, moderate, limited or insufficient. Confidence describes how far the cited evidence actually goes; it qualifies a score and never adjusts it. A dimension marked insufficient carries no number at all.

The six dimensions are never combined. There is no overall score, no aggregate, no total, no weighted sum and no ranking, and the product exposes no mechanism to produce one. Dimensions carry no weights, so there is nothing to weight together.

• No candidate is ranked against another.
• No score is a prediction, an endorsement, a recommendation or an official result.
• A dimension score is a reading of documented evidence, not a verdict on character.
• Scores can change when corrections or new dated sources arrive.
• Comparison remains fully usable when no scorecard is shown.

Holding an office today does not mean standing for it again, so incumbency is recorded as its own fact with its own dates and never implies a future candidacy.

Candidacy is recorded at the exact stage the evidence supports, from reported aspirant, through declared aspirant and party nominee, to confirmed official candidate. A record is never promoted to a stronger stage than its sources establish.

Party affiliation is a third, independent question. Where the sources do not establish it, the record says the affiliation is under review rather than showing the party the person was last associated with, which may be precisely what changed.

Manifestos and campaign material are candidate-supplied. They are published as the candidate's own statement and are never presented as independently verified.

Manifesto credibility asks whether published commitments are specific, costed and consistent with the record. Delivery and resource efficiency asks what was actually delivered, at what cost, and with what independently verifiable output. Both require dated sources like any other dimension.

A project is attributed to a candidate only in the role the evidence supports. Occurring during someone's tenure, in their area, or at an event they attended is not attribution.

Public comments and reactions are shown separately from verified candidate information. They do not feed any dimension score, do not change confidence, and do not alter what the service reports as fact.

The separation is structural, not a matter of restraint. If public reaction could move an assessment, the record would become a popularity measure, and an organised campaign could change what the platform says about a person without producing a single document.

No aggregate sentiment figure is published anywhere, because a number of that kind is read as a score however it is labelled.

• Comments are moderated before they appear.
• Signing in is required to comment or react.
• Public opinion does not move search results, ballot listings or any editorial assessment.

Every published record and every scorecard dimension can be challenged. Corrections are reviewed by a person against the cited sources; nothing is published automatically and nothing changes at the moment a correction is sent.

Use the Suggest a correction page, reachable from any candidate record and from the footer, and say which claim is wrong.

• Say which record and which specific claim is wrong.
• Cite a source we can check, with its publication date.
• Corrections that arrive with a source can be acted on; those without usually cannot.
• Where sources genuinely conflict, the record shows the matter as contested rather than picking a side.
• Publication history is retained, so a change never erases what was previously shown.

Coverage varies by office and by state. Some contests are documented in depth and some are barely documented at all in public sources, so records differ in how much evidence stands behind them. A record with thin coverage says so and does not borrow confidence from the records beside it.

The scorecard is a Beta and covers only the contests it has been calibrated for. It appears publicly only when the master scoring switch, the Beta switch and the methodology approval all permit it; with any one of the three off, no scorecard is shown anywhere.

Nothing here determines, predicts or replaces an official election result. Official results are declared by the relevant electoral authority.

• An empty seat means we have not established who is standing, not that nobody is.
• This document is versioned; each edit bumps the version and the update date.
• Nothing on this site is an endorsement of, or opposition to, any candidate or party.