Part 410 min

4.2 Collecting information and interviewing

Step 2 of OSHA's four-step method is collect information, and it is the step where investigations are won or lost. Memories fade within hours and reshape themselves within days. Documents that would have answered the question get filed, overwritten or thrown away. And the single most productive question in the whole process takes six words and is almost never asked.

1 Interviews come first, because memory does not wait

The guide is blunt about timing: “Since memories fade, interviews must be conducted as promptly as possible: preferably as soon as things have settled down a bit and the site is both secure and safe. The sooner a witness is interviewed, the more accurate and candid his/her statement will be.”

Expect to interview more than once. The guide notes that an investigation always involves interviewing and possibly re-interviewing some of the same or new witnesses as more information becomes available — up to and including the highest levels of management. That last clause matters: if the sequence of events runs back to a scheduling decision, the person who made it is a witness.

2 How to run the interview

The guide gives a list of instructions for interviewers, and every one of them is there because the opposite behavior has ruined interviews before.

Conduct it in the language of the person, using a translator if needed.
State clearly that the purpose is fact-finding, not fault-finding, and that the goal is to learn how to prevent the next one.
Establish a climate of cooperation. Avoid anything that could be perceived as intimidating or as looking for someone to blame.
Tell the person they may have an employee representative present where that is available or appropriate.
Ask them to recount their version, and do not interrupt.
Take notes, or record with permission — the interviewee must agree before being recorded.
Have blank paper or a sketch available for them to point at and draw on.
Ask clarifying questions afterwards, reflect the facts back, and correct inconsistencies there and then.

And then the last one on the list, which is the one worth building the whole interview around: ask what they think could have prevented it, focusing on the conditions and events preceding the injury. People who do a job every day usually know. Very few are ever asked.

On the job: interview at the scene if it is safe, not in an office. Standing where it happened produces detail that no conference room can: sight lines, noise, footing, what was in the way.

3 The documents, before they disappear

Interviews are one source; the guide lists the others explicitly, and most of them have a shelf life measured in weeks.

Other sources of information — and why each one goes missing
Source	Why it matters
Equipment manuals	What the manufacturer actually specifies — often different from site practice
Industry guidance documents	What the industry recognizes, which is what “recognized hazard” means
Company policies and records	What was supposed to happen, in writing
Maintenance schedules, records and logs	Whether the item was due, overdue, or never on the list
Training records, including communication to employees	Whether the person had been trained, when, and on what
Audit and follow-up reports	Whether this condition had been found before
Enforcement policies and records	Whether the rule was enforced or merely written
Previous corrective action recommendations	The single most damaging document if an earlier action was never completed

Add the ones a modern site generates and nobody thinks of: telematics from the machine, the pre-task plan or permit for the shift, the toolbox talk record for the week, the weather record, the delivery schedule, and any near miss reports on the same task.

4 What, when and how — the checklist's shape

Appendix E of the guide is a question checklist grouped as what, why, when and how. It is long, and its value is that it forces questions nobody asks under pressure.

The what group covers the injury, what the employee was doing, what they had been told to do, the tools and machine involved, the instructions and precautions given, and the protective equipment that should have been used against what was used. The when group is the one that catches scheduling and supervision: when did the employee start on that job, when were they assigned to it, when were the hazards pointed out, when did the supervisor last check on progress, and when did the employee first sense something was wrong.

The how group is deliberately forward-looking: how did the employee get injured, how could they have avoided it, how could fellow workers have avoided it, and how could the supervisor have prevented it. Those are the questions whose answers turn into corrective actions.

5 Facts and conclusions live in different columns

The discipline that keeps an investigation usable is separating what is observed from what is inferred. A photograph is a fact. “The guard had been removed” is a fact if the guard is on the bench and a conclusion if it is merely absent. “The operator removed the guard” is a conclusion that requires evidence, and “the operator was careless” is not a conclusion at all — it is a verdict, and the next chapter explains why it ends the investigation early.

Common mistake: writing the narrative from the first account you hear and then testing later accounts against it. Collect all of them first, then reconcile. The differences between accounts are information, not noise.

6 Building the timeline

1
Lay out every fact you have on a single line each, with a time against it where you can.
2
Include the events before the shift: the schedule, the delivery, the change of plan, the crew substitution.
3
Mark where accounts disagree, and what evidence would settle each disagreement.
4
Mark the points where something could have stopped the sequence — a control that was absent, present but bypassed, or present and ineffective.
5
Those marks are where the root-cause questions of the next chapter start.

A timeline is not a formality. It is the tool that shows how much of the sequence happened before anybody arrived on site — which is almost always where the changeable causes are.

Key takeaways

Interview promptly — memories fade — and expect to re-interview as more emerges
Say plainly that the purpose is fact-finding, not fault-finding, in the person’s own language
Let people tell it their way without interruption, then ask clarifying questions
Always ask what they think could have prevented it
Collect the documents fast: maintenance, training, permits, audits and earlier corrective actions
Keep facts and conclusions separate, and build a timeline that starts before the shift did

Free educational content — not OSHA-authorized training, no certificate or card issued. Follow your employer's program and the standards cited.