Plain-language brief · Updated
Why injury rates at small workplaces jump around
A Spanish translation of this brief is available. English
The short answerAt a small workplace, one injury more or less changes the injury rate a lot. A rate from one year is a weak way to say whether a small workplace is safe or unsafe.
What the numbers show
- Many small workplaces report no injuries at all. In OSHA's public injury files, 75.11% of the workplace-years in the smallest size group (001-019 employees) reported zero cases. In the largest group (1000+ employees) it was 8.13%. A zero can mean a safe year. It can also mean a few lucky months.
- One injury moves the rate a lot. Take chemical plants with Under 50 employees. The middle plant works 56,687 hours in a year. For that plant, one more injury raises the rate by 3.53 points. For a plant with 1,000+ employees, one more injury raises it by only 0.06 points.
- The order changes from year to year. Researchers measure this with a score. A score of one means plants keep the same order every year. A score of zero means last year's order tells you nothing. For chemical plants with Under 50 employees the score is 0.314. For plants with 1,000+ employees it is 0.742.
What this means for you
- Workers: a low or high rate at a small workplace is not proof about your own risk. Ask about the hazards you work near: machines, chemicals, lifting, heat. Ask what the employer fixed after past injuries or near misses.
- Small employers: do not take comfort from a year with no injuries, and do not panic over a bad year. Look at several years together. Count near misses and hazards reported by workers. These show problems before someone is hurt.
- Anyone comparing workplaces: do not rank small workplaces by one year's rate.
Limits
The zero-case shares come from OSHA establishment filings in all industries. The one-injury swing and year-to-year scores come from chemical manufacturing plants only. The data are reported by employers and have errors. These results describe patterns in groups of workplaces, not any one workplace.
Where each number comes from
Each number is copied from a file in a public repository. The link opens that file.
| Number | Source file | Field |
|---|---|---|
| 75.11% | ehs-osha-analysis/outputs/tables/size_band_effects.csv | zero_share |
| 001-019 | ehs-osha-analysis/outputs/tables/size_band_effects.csv | size_band |
| 1000+ | ehs-osha-analysis/outputs/tables/size_band_effects.csv | size_band |
| 8.13% | ehs-osha-analysis/outputs/tables/size_band_effects.csv | zero_share |
| Under 50 | ehs-chemical-peers/outputs/summary.json | size_noise / band |
| 56,687 | ehs-chemical-peers/outputs/summary.json | size_noise / median_hours |
| 3.53 | ehs-chemical-peers/outputs/summary.json | size_noise / one_case_swing |
| 1,000+ | ehs-chemical-peers/outputs/summary.json | size_noise / band |
| 0.06 | ehs-chemical-peers/outputs/summary.json | size_noise / one_case_swing |
| 0.314 | ehs-chemical-peers/outputs/summary.json | stability / rho |
| 0.742 | ehs-chemical-peers/outputs/summary.json | stability / rho |