Ask a trenching crew what a trench box does, and most will tell you it stops the cave-in. It doesn’t. And our data shows most workers genuinely believe it does. We asked whether a trench box is designed to prevent cave-ins by strengthening the trench walls. Only 39% got it right. Roughly 6 in 10 workers think the box stops the collapse.
Here’s why that matters. A trench box protects the people inside it while the soil fails around it. A worker who believes the walls are reinforced will stand outside the box, reach beyond it, or delay repositioning it. Every one of those is how people get hurt.
A 2024 Carnegie Mellon study published by CPWR looked at OSHA trenching fatality cases and found 80% were cited for protective systems, 60% for training and education, and 50% for inspections. The same research found a trench box was involved in 17% of severe trenching injuries. The equipment we trust most keeps showing up in injury reports.
So, we asked why? Protective systems were the single overconfident topic from our data. This means over 60% of employees answered the question wrong but were confident they had it right.
A worker who’s unsure and knows it will ask, double-check, or defer to a competent person. A worker who’s wrong and confident - won’t. That’s the most dangerous person on the site, and nobody can pick him out of the crew. Including him.
Certified, confident, and wrong.
At Training Pulse, we measure what workers actually retain after the completion certificate gets issued. Across nearly 1,500 assessment responses on trenching and excavation, protective systems stands out as the weakest topic, both initially and after reinforcement. Post-training scores average 59%; even with reinforcement, they remain at 55%. The confidence gap held throughout. Continuous measurement revealed a pattern that workers are dangerously misaligned between what they know and what they think they know.
They know the soil. They don’t know the slope.
But overconfidence isn’t the only gap. In our data, 83% of workers knew that Type C soil can’t be reclassified as Type A just because it looks dry and firm. But only 36% knew the maximum allowable slope ratio for that same soil. They can spot the most dangerous ground on the site. They just don’t know what to do about it.
Easy questions hide the problem.
Question format matters more than we expected. True/false questions came back 84% correct, with confidence sitting right where it should. They make everyone feel competent. But “select all that apply” questions, where you need the complete picture, came back 45% correct against confidence of 8.3 out of 10. The worst one? Which site conditions must be checked before workers enter an excavation. Only 36% got it right, but confidence measured at 8.8. When we added longer, scenario-based questions that ask for real field judgment, scores went up and confidence dropped - demonstrating real learning. Harder questions do more than test. They make overconfidence visible.
Your gap isn’t your neighbor’s gap.
The blind spot moves. For one contractor it was protective systems. For another, competent-person duties. You won’t find your gap with a generic annual refresher. You find it by measuring your own crews. And it’s fixable: our lowest initial scorers improved by nearly 4x after retention boosters, from 1.98 to 7.71.
Completion doesn’t equal competence. Somewhere on your next dig, there’s a worker in the wrong spot, confident the box will protect him. Find him before the soil does.
Jay Stephens is Co-Founder & CEO of Training Pulse (trainingpulse.io), a learning analytics and retention platform for high-risk industries.