Transparency

Methodology

How we select, measure, aggregate, and present government performance metrics. We believe transparency about our own methodology is as important as the data we present.

Core Principles

Objectivity

KPIs are selected based on their ability to measure government performance objectively, not on whether they reflect positively or negatively on any administration, party, or ideology.

Public data only

All data comes from official government sources (federal statistical agencies, public APIs, official reports) or from established, peer-reviewed research organizations. No proprietary data.

Reproducibility

All formulas, data sources, and aggregation methods are documented so that any informed person can reproduce our numbers from the original sources.

Honest uncertainty

Where data quality is limited, methodology is contested, or coverage is incomplete, we say so explicitly with caveat banners and data quality indicators.

Trend over snapshot

Single data points are typically misleading. We prioritize multi-year historical trend data and year-over-year comparisons over current-period snapshots.

Non-advocacy

We present metrics and trends without prescribing policy solutions. A rising or falling metric is presented as an observation, not a political verdict.

KPI Selection Criteria

KPIs are selected based on the following criteria, applied in order:

1
Data availability: A reliable public data source must exist with at least 3 years of historical data. Planned KPIs for which no current public source exists are marked "Coming Soon."
2
Outcome orientation: We prefer metrics that measure outcomes (what actually happened) over inputs (what was spent) or outputs (what was produced), though all three can be valuable.
3
Constitutional relevance: Each KPI must be directly relevant to the constitutional or statutory functions of the branch it is placed under.
4
Measurability: The metric must have a clear, documented formula with defined numerator, denominator, and measurement period.
5
Non-redundancy: We avoid duplicating information across KPIs unless the same underlying phenomenon is measured from usefully different angles.
6
Public interest: KPIs must be of genuine civic interest — metrics that policy-aware citizens, journalists, researchers, or public officials would find actionable.

Data Quality Ratings

Each KPI displays a data quality rating based on the following scale:

HighData is published by a federal statistical agency (e.g., BLS, BEA, U.S. Courts) with documented, peer-reviewed methodology. Historical series is complete and machine-readable.

MediumData is generally reliable but has known gaps, definitional inconsistencies, or a methodology that has changed over time. Self-reported data from agencies also falls here.

LowData coverage is incomplete, depends on voluntary reporting, or lacks an authoritative source. These KPIs are included for completeness but should be interpreted with significant caution.

EstimatedValues are model-based estimates, interpolated from partial sources, or are third-party calculations not directly published by the primary agency.

Update Frequency

Each KPI is tagged with the cadence at which the underlying source data is published. This dashboard is currently in a static-data MVP phase — the sample historical data is pre-loaded from official published sources as of mid-2024. In a production deployment with an ETL pipeline, KPIs with API-available sources would update automatically.

Daily

Debt-to-the-penny (Treasury), USAspending contract data, Federal Register

Monthly

BLS employment, CPI, FRED series, FEMA declarations, USCIS processing times

Quarterly

GDP, FRED fiscal data, U.S. Courts clearance rates, lobbying disclosure

Annual

Public trust surveys, U.S. Courts statistics, USSC sentencing data, FJC data

Per Congress

Bill passage rates, bipartisan voting rates, budget timeliness

Per Term / Event

Judicial vacancy data (updated weekly), SCOTUS confidence polls

Known Limitations

Sample data in MVP

This dashboard is currently an MVP with pre-loaded historical data. The data is drawn from official published reports but is not yet connected to live APIs. Figures should be verified against primary sources before being cited.

Federal scope only

This dashboard covers federal government institutions. State, county, and municipal government performance is not yet included, though the KPI framework is designed to support drill-downs to state and local levels where data is available.

Polling data limitations

Public trust and approval metrics depend on polling data, which has known limitations including declining response rates, weighting methodology disputes, and question wording effects. Polling trends are useful directionally but should not be treated as precise measurements.

Operational vs. policy performance

This dashboard measures operational and institutional performance (how government functions), not policy outcomes (whether specific policies achieve their goals). The distinction matters: a government can execute a budget efficiently while pursuing policies that many citizens view as harmful, and vice versa.

No composite scores

We intentionally do not publish composite scores, letter grades, or rankings across administrations. Such aggregations require normative weighting choices that would introduce political bias. We present individual metrics and let users draw their own conclusions.

Roadmap: Automated ETL Pipeline

The architecture for automated data ingestion is described in the deployment documentation. Priority APIs for scheduled ingestion (via cron jobs or a workflow orchestrator like Apache Airflow or Prefect) are:

→BLS Public Data API — CPI, employment data (monthly)
→FRED API — PCE, GDP, debt/GDP ratio (monthly/quarterly)
→USAspending.gov API — outlay execution, contract competition rates (monthly)
→Congress.gov API — bill counts, vote records (daily during session)
→FEMA OpenFEMA API — disaster declarations (as declared)
→U.S. Courts public data tables — clearance rates, caseload (quarterly)
→Senate LDA API — lobbying disclosures (quarterly)