01
Verified performance grade
A 0–100 job-performance score. A letter appears only after the same evidence gates clear for comparable officials.
Performance grade • method v1.0
Grade the job. Never grade the party.
RepWatchr's overall grade measures documented execution of public duties. Ideology, party loyalty, fundraising, follower counts, press volume, and community popularity do not change it.
First principle
One profile. Three outputs that never contaminate one another.
This separation prevents popularity or editorial preference from masquerading as job performance.
02
Vote and issue alignment
The sourced vote plus an optional comparison to a member's own priorities. Policy direction never changes the performance grade.
03
Participant sentiment
A separate, self-selected community signal. It is not a scientific poll, endorsement, or performance input.
The five dimensions
Every point answers a neutral job-performance question.
A yea and a nay are not scored as morally correct or incorrect. Missing evidence is never converted into a zero, a 50, or a guess.
20%
Voting accountability
Did the official take a documented position when eligible, regardless of whether the vote was yea or nay?
25%
Legislative effectiveness
Did the official turn comparable opportunities into substantive, documented outcomes?
25%
Ethics and integrity
What do final authoritative findings and required filings establish?
20%
Constituent service and transparency
Is the office objectively accessible, responsive, and compliant with transparency duties?
10%
Attendance and duty fulfillment
Did the official attend eligible duties after approved leave and official conflicts are excluded?
Publication gates
A grade is withheld until the evidence can defend it.
NR is not a favor or a punishment. It means the record has not cleared the same published gate required for everyone else.
Full number + letter
80%+
At least 80% of planned weight scoreable, 70% weighted coverage, 65% confidence, four categories, and ethics scoreable.
Provisional number
60%+
At least 60% of planned weight scoreable and 45% confidence. The number is labeled provisional and receives no letter.
Not rated
NR
Below the provisional gate: no numeric score, no letter, and a visible checklist of the evidence that is still missing.
Missing data rule
Confidence explains uncertainty. It does not punish the official.
Category confidence combines coverage, source quality, freshness, and reliability. It is displayed beside the score and never multiplied into the official's performance.
confidence = coverage ×
(0.50 source quality +
0.25 freshness +
0.25 reliability)
(0.50 source quality +
0.25 freshness +
0.25 reliability)
Evidence and due process
Allegations may be reported. They never become deductions by repetition.
Only final court orders, authorized ethics findings, audits, or equivalent authoritative dispositions can reduce the ethics score.
Tier 1
1.0
Official roll calls, journals, filings, audits, final ethics orders, and court records.
Tier 2
0.9
Official transcripts, meeting video, agency datasets, and direct official statements.
Tier 3
0.7
Named independent reporting tied to primary records or independently corroborated.
Tier 4
0.4
Campaign, advocacy, and social material—proof of the speaker's statement, not proof that an accusation is true.
Unverified
0
A research lead only. It cannot move a score.
Community signal
Verified participant sentiment stays visible—and separate.
One current response per verified member. In-district results are primary; other locations are labeled separately. Public results unlock at 25 eligible responses.
Uncertainty shrinks toward neutral—not toward disapproval.
RepWatchr uses a neutral prior of 50 with a starting weight of 50 responses. Small coordinated samples therefore remain visibly uncertain instead of looking like a zero.
sentiment = (n × raw average + 50 × 50) ÷ (n + 50)
Public label
Verified participant sentiment
Not an election. Not a scientific poll. Never part of the performance grade.
Letter scale
One scale everywhere. No invented precision.
Version 1 uses whole-letter grades only. The evidence does not justify plus/minus theater.
A
90–100
B
80–89
C
70–79
D
60–69
F
0–59
See a factual error or a missing source? Corrections create a new version instead of silently erasing the old calculation.
Submit a source or correction