Are You Measuring the Right Things in Safety Training?
- Megan Weber
- 2 days ago
- 3 min read
Most organizations measure safety performance the same way: Total Recordable Injury Rate, Lost Time Incident Rate, and fatality counts. These numbers go into board reports, bid submissions, and supply chain audits. They are used to evaluate vendors, reward teams, and demonstrate progress.
The research says we are putting too much weight on numbers that, at the level most organizations use them, are statistically too thin to tell us much at all.
The Problem with Lagging Indicators
Recordable injuries are rare events. A typical construction or manufacturing operation with 2,000 workers might record 30 to 60 OSHA-recordable injuries in a year. That sounds like a lot until you try to use it as data.
To reliably detect a real 20 percent improvement in underlying safety at standard statistical confidence levels, you need either a much larger workforce, a much longer observation window, or both. Most organizations have neither. Which means a genuine improvement can fail to show up in the numbers simply because the sample is too small. And a 30 percent drop in TRIR can happen in a good year with no real change at all.
The statistics don’t care what the press release says. Single-site, single-year recordable rates are mostly noise. Multi-year trends across a portfolio are meaningful. The industry consistently rewards the wrong one.
What Safety Training Actually Moves
The research is clear about what safety training reliably changes. Knowledge gains, hazard identification rates, safe work practices, and behavioral compliance show up consistently across the literature with effect sizes large enough to be confident in.
These are leading indicators. The conditions that precede incidents, not the incidents themselves. They are what the evidence primarily measures, and they are the outcomes an honest evaluation of safety training should focus on.
Leading indicators worth tracking include:
Comprehension rates, not just completion rates. Did workers demonstrate understanding, or just finish the module?
Hazard identification scores. Can workers identify the actual risks in the work they’re about to do?
Near-miss reporting volume. A rise in near-miss reports is usually a sign of a healthier safety culture, not a worse one.
Briefing frequency and quality. How often are supervisors delivering scope-specific safety briefings, and how are those briefings being assessed?
Why This Matters for Vendor Claims
Vendors that market 30, 40, or 50 percent TRIR reductions on the back of a single customer’s twelve-month deployment are either misreading the statistics or getting lucky on a small sample. Either way, the claim does not survive scrutiny.
A vendor that says “we reliably move leading indicators, and we report on lagging indicators honestly, including the limits of what your data can actually prove” is more useful than one promising headline injury rate reductions. The first is the kind that produces case studies that don’t hold up when someone checks the sample size.. The second is the kind of partner regulators, insurers, and serious clients increasingly want.
What To Do Instead
Lagging indicators still matter. They should be tracked, reported honestly, and aggregated over multi-year, multi-site samples large enough to bear the statistical weight being asked of them. A five-year trend across a major program tells you something real. One site’s TRIR over twelve months mostly does not.
The shift the evidence supports is to lead with what can actually be measured reliably: comprehension, hazard awareness, briefing quality, and pre-work verification. Treat lagging indicators as a long-run check, not a short-run scorecard.
Organizations that make that shift will not just report more honestly. They will manage more effectively because they will be acting on signals that actually tell them something about readiness before an incident happens, not after.
This is the third and final principle explored in The Three Principles, an independent industry study reviewing 25 years of peer-reviewed research on Safety Orientation and toolbox talks. Download the full study below:
Comments