Methodology

How we calculate scores, detect conflicts, and ensure fairness. Every formula and threshold is documented here.

Open Methods. No Black Boxes.

Core Principles

Objectivity

All scores derive from verifiable data and transparent formulas. No editorial judgment influences individual scores.

Transparency

Every methodology detail is public. Users can see exactly how each score is calculated and what data sources are used.

Fairness

Politicians are compared within peer groups (same chamber, similar tenure). Balanced assessment considers both sides of every conflict.

Data-Driven

Scores update as new data arrives. No manual overrides. AI-generated content is clearly labeled with confidence levels.

Grading System

Scores are mapped to letter grades on a 0–100 scale. Colors indicate performance bands.

GradeScore RangeDescriptionColor
A+97100Exceptional
A9396Excellent
A-9092Very Good
B+8789Good
B8386Above Average
B-8082Slightly Above Average
C+7779Average
C7376Below Average
C-7072Slightly Below Average
D+6769Poor
D6366Very Poor
D-6062Extremely Poor
F059Failing

10 Scoring Dimensions

Each politician is evaluated across 10 dimensions. Dimensions with insufficient data are excluded from the composite score.

D1Promise Fulfillment

Weight: 15%

Rate at which a politician follows through on campaign promises and public commitments

Σ(status_weight × specificity_weight) / Σ(specificity_weight)

Sources: Campaign speeches, press releases, official statements, legislative records

D2Specificity & Clarity

Weight: 8%

How specific, measurable, and actionable the politician's commitments are

Σ(specificity_score × measurability) / count

Sources: Same as D1 — evaluated at extraction time

D3Financial Independence

Weight: 10%

Degree of concentration or diversification in funding sources

100 − (HHI_normalized × sector_risk_multiplier)

Sources: FEC filings, OpenSecrets, state disclosure databases

D4Conflict Exposure

Weight: 12%

Inverse of total conflict flag severity — fewer and less severe flags yield higher scores

100 − Σ(severity_weight × pattern_risk)

Sources: Cross-referencing votes, trades, donors, and committee assignments

D5Voting Consistency

Weight: 12%

Alignment between stated positions/promises and actual votes

matched_votes / total_linked_votes × 100

Sources: Congress.gov roll call data, promise-vote linkage analysis

D6Legislative Productivity

Weight: 10%

Bills introduced, cosponsored, and passed relative to chamber peers

percentile_rank(bills_passed + 0.3 × bills_cosponsored)

Sources: Congress.gov legislative data, GovTrack statistics

D7Constituent Alignment

Weight: 12%

Degree to which voting behavior matches district/state preferences

aligned_votes / district_issue_votes × cook_pvi_adjustment

Sources: District polling data, Cook PVI, constituent surveys

D8Attendance & Participation

Weight: 6%

Vote participation rate and committee hearing attendance

(votes_cast / votes_held) × 0.7 + (hearings_attended / hearings_held) × 0.3

Sources: Official chamber attendance records, committee logs

D9Transparency & Disclosure

Weight: 7%

Timeliness and completeness of financial disclosures

filing_completeness × on_time_rate × correction_penalty

Sources: STOCK Act filings, financial disclosure reports, OGE records

D10Bipartisan Effectiveness

Weight: 8%

Cross-party collaboration on bills and amendments

bipartisan_cosponsors / total_cosponsors × passage_bonus

Sources: Congressional cosponsorship data, bipartisan index scores

Composite Score

The composite score is a weighted average of all available dimensions. At least 5 dimensions must have data for a composite score to be generated.

composite = Σ(dimension_score × weight) / Σ(weight) where dimension_score ≠ null

Weight Presets

Users can switch between focus presets to weight dimensions differently.

PresetD1D2D3D4D5D6D7D8D9D10
Balanced15%8%10%12%12%10%12%6%7%8%
Anti-Corruption Focus8%5%20%25%10%5%5%5%15%2%
Effectiveness Focus20%5%5%5%10%20%10%10%5%10%
Representation Focus20%5%5%5%20%5%25%10%3%2%
Transparency Focus5%15%20%20%5%3%5%5%20%2%

Interactive Formula Explorer

68
D+
Composite Score
Weighted average of 10 dimensions
Dimension
Score
Weight
Promise Fulfillment
7215%
Specificity & Clarity
858%
Financial Independence
4510%
Conflict Exposure
3812%
Voting Consistency
9112%
Legislative Productivity
6810%
Constituent Alignment
7712%
Attendance & Participation
956%
Transparency & Disclosure
527%
Bipartisan Effectiveness
648%
Contribution to Composite
Promise10.8%
Specificity6.8%
Financial4.5%
Conflict4.6%
Voting10.9%
Legislative6.8%
Constituent9.2%
Attendance5.7%
Transparency3.6%
Bipartisan5.1%

Conflict Detection

Eight conflict patterns are monitored across five primary (P1–P5) and three secondary (S1–S3) categories. Each pattern has specific trigger conditions and mitigating factors.

P1Donor-Vote Alignment

significant

Trigger: Votes benefiting major donor sectors within 90 days of large contribution

Mitigating factors: Small donor ratio, party-line vote, broad bipartisan support

P2Stock-Committee Overlap

noteworthy

Trigger: Securities holdings in sectors under committee jurisdiction

Mitigating factors: Blind trust, index fund, pre-scheduled trade, broad market ETF

P3Stock-Vote Timing

critical

Trigger: Trades within 30 days before/after votes affecting the security's sector

Mitigating factors: Pre-scheduled, blind trust, loss-generating trade, broad index

P4Business-Legislation Connection

significant

Trigger: Introducing or championing bills benefiting businesses with financial ties

Mitigating factors: Broad economic benefit, constituent demand, bipartisan effort

P5Donor-Committee Jurisdiction

noteworthy

Trigger: Significant donations from sectors under the politician's committee jurisdiction

Mitigating factors: Industry-standard levels, constituent employers, small donor offset

S1Donor-Legislation Beneficiary

significant

Trigger: Legislation directly benefiting specific donor entities

Mitigating factors: Broad beneficiary pool, consumer benefit, prior advocacy record

S2Revolving Door

noteworthy

Trigger: Post-service employment patterns with previously regulated entities

Mitigating factors: Cooling off period compliance, different sector, public service track record

S3Bundled Influence

critical

Trigger: Multiple conflict patterns co-occurring for the same entity/sector

Mitigating factors: Independent verification of each sub-pattern, context assessment confirms routine

Severity Thresholds

informational
noteworthy
significant
critical

Flags de-escalate one severity level when 2+ mitigating factors are confirmed present.

Data Source Tiers

Data is classified into four reliability tiers. Higher-tier data takes precedence in conflicts.

Tier 1

Government Records

Congress.gov, FEC filings, OGE disclosures, STOCK Act reports, official press releases

Primary authoritative data — highest confidence

Tier 2

Regulatory & Watchdog

OpenSecrets, GovTrack, ProPublica Congress API, CRP sector classifications

Curated secondary data — high confidence with cross-reference

Tier 3

AI-Extracted

Campaign speech transcripts, press conference analysis, social media statements

Claude-extracted with confidence scoring — requires verification

Tier 4

Community

User submissions via moderated community contribution pipeline

Multi-stage review (automated → AI → peer → staff) before incorporation

Fairness Safeguards

Balanced Assessment

Every conflict flag includes both a "case against" and "case for" the politician, generated by AI with structured prompts that mandate both perspectives.

Prohibited Language

Scoring and assessment text is screened for partisan language, inflammatory terms, and editorializing. Only factual, neutral language is used in scores and summaries.

Peer Comparison

Individual metrics are contextualized against chamber percentiles. A politician isn't penalized for behaviors common across all members (e.g., industry-standard PAC contributions).

Mitigating Factors

Each conflict pattern has specific mitigating factors that are actively checked. When 2+ are present, severity is automatically de-escalated by one level.

Known Limitations

Data Gaps

Not all politicians have complete data across all 10 dimensions. State-level officials may have fewer federal data sources. The composite score adjusts by only averaging available dimensions.

AI Confidence

AI-extracted promises and context assessments carry confidence scores. Lower-confidence extractions are flagged and weighted accordingly. Community verification helps improve accuracy over time.

Timing Lag

Financial disclosure data can lag by 30–45 days. Vote data is typically available within 24 hours. Promise status updates depend on legislative calendar and news cycle coverage.

Causation vs. Correlation

Conflict flags identify patterns of potential concern, not proven wrongdoing. A donor-vote alignment flag means the pattern exists, not that the donation caused the vote.