Example: The Removal of OSHA Injury/Illness Data. OSHA rolled back public access to establishment-level injury and illness logs that had been published annually. Suddenly, researchers and journalists couldn’t see workplace safety patterns. Establishment-level data became “restricted,” forcing researchers to apply for access—sometimes taking 6–12 months—or rely on partial aggregated summaries.
Play: Freedom of Information Act (FOIA)
File a FOIA request to recover missing or restricted data. If a federal dataset suddenly disappears or becomes restricted, you can submit a Freedom of Information Act (FOIA) request to obtain establishment-level logs, tables, or documentation. FOIA Portal: https://www.foia.gov/
Play: Crowdsource
Set up a crowdsourced data collection system through worker centers or partner organizations. If official reporting dries up, you can create a community-driven reporting channel by partnering with local worker centers, advocacy organizations, or unions to gather incident or outcome data directly from affected people. In this example, worker advocacy groups switched to crowdsourced injury reporting through worker centers.
Example model (Worker Centers): https://www.workercenters.org/
Play: Reconstruct
Reconstruct indicators using hospital discharge data and ICD codes. If federal injury or illness data disappears, you can estimate patterns using state hospital discharge databases and ICD-coded injury records, many of which remain stable and publicly accessible. State Hospital Discharge Databases Directory: https://www.hcup-us.ahrq.gov/
Play: Buy new
Private data vendors selling safety datasets saw a surge in nonprofit subscriptions (irony noted). Subscribe to private safety or labor datasets as a temporary stopgap. If you’re missing official data, you can use private vendors who maintain injury, labor, or safety datasets. (It’s ironic, but these datasets sometimes remain more stable across political cycles.) Example (Equifax Workforce Solutions / industry analytics): https://workforcesolutions.equifax.com/
Example (Verdantix EHS Market Data): https://www.verdantix.com/
**
Play: Synthesize**
Use synthetic microdata derived from older years. When newer microdata is missing or unreliable, you can create synthetic microdata that statistically mirrors older, higher-quality federal microdata using accepted privacy-preserving techniques. Synthetic Data Guidance (U.S. Census Bureau): https://www.census.gov/programs-surveys/sipp/guidance/users-guide.html
Tools (Synthpop R package): https://cran.r-project.org/web/packages/synthpop/index.html
**
Play: Administrative Data**
Replace missing federal data with stable administrative datasets (e.g., USPS, NHTSA). If the primary dataset has vanished, you can pivot to administrative datasets that remain insulated from political shifts, like postal address change data (USPS), traffic safety crash data (NHTSA), or state licensing databases.USPS Address Data Overview: https://about.usps.com/who/legal/privacy-policy/approved-shippers.htm
NHTSA Crash Data Portal: https://cdan.dot.gov/
Anything else we should add?