Provenance · Refresh cadence · QA

Where every number comes from

The tracker, the comparator, the calculators, the scorecards — every data point in this catalog cites its source, last-verified date, and refresh cadence. Documented here so state teams and reviewers can audit us.

Data dashboard on a screen
Cited, dated, refreshable

Our methodology in short

Principle 01
Authoritative sources only

Federal agencies, peer-reviewed research, TA-center data. No blog posts, no press releases without primary source backing.

Principle 02
Every record dated

Last-verified date attached to every data point. Stale records flagged after 12 months.

Principle 03
Automated + human QA

Automated pipeline flags anomalies. Subject matter experts sign off on every quarterly refresh.

Principle 04
Corrections public

Every correction is logged with date, prior value, corrected value, and reason.

Sources catalog

Source What it feeds Refresh Last verified
NCSL bill trackersState Policy Tracker · Model Language LibraryWeekly2026-07-14
ABLE NRC state comparisonFinancial Literacy simulator · ComparatorQuarterly2026-06-30
EEOC MD-715 + state/local reportsState-as-Employer scorecardAnnual2026-01-15
BLS state-year employmentModel Employer + ComparatorMonthly2026-07-05
JAN Cost/Benefit SurveyAccommodation Reality CheckAnnual2026-05-01
DOL RETAIN evaluation reportsRTW ROI CalculatorPer release2024-10-30
Kennedy Forum State Parity ScorecardMental Health snapshot + ComparatorAnnual2026-02-10
SAMHSA Recovery-Ready FrameworkMental Health Self-AuditPer release2025-11-14
RSA-911 outcomesEducation-to-Workforce WizardAnnual2026-03-01
SSA POMSABLE + SSI + Entrepreneurship simulatorsPer POMS update2026-06-01
DOL AI Literacy FrameworkAI Literacy QuizPer release2025-09-20
Yin (2026) RISEI Lab WP 2026-05Causal ABLE effect estimatesPer revision2026-05-15
State agency reports (50 states)State Policy Tracker (per-state records)RollingPer-state entry

Quarterly refresh process

Step 1

Automated pull

Every listed source pulled by scheduled job. Snapshots stored with timestamp.

Step 2

Anomaly detection

Quarter-over-quarter change thresholds flag suspect values for SME review.

Step 3

SME sign-off

Subject matter expert reviews every flagged record before publication.

Step 4

Public change log

Every change logged with date, prior value, new value, and rationale.

Known limitations

  • Territory coverage. American Samoa, Guam, Northern Mariana Islands, Puerto Rico, USVI are represented in the tracker but per-territory data density is lower than for the 50 states.
  • State legislative variation. Some states use bill-tracking systems not covered by NCSL. Manual entry closes this gap on a rolling basis.
  • Lag on federal data. RSA-911 and EEOC MD-715 data typically publish 12-18 months after the fiscal year. We label reporting lag explicitly.
  • Descriptive, not causal. The tracker is descriptive. Causal claims only appear where research supports them (e.g., Yin 2026 for ABLE, RETAIN Phase 2 for RTW).
  • Aggregation. Some state-year comparisons involve small denominators; confidence intervals available on request.
Corrections welcome. If you find a value that seems wrong, email us at info@inclusiveaiinstitute.org with the citation. We investigate every report and publish the outcome in our change log.