CSP common mistakes
CSP · Understand the traps and how to avoid them.
When an item asks for the 'most effective' or 'best' way to control an identified hazard, the candidate selects training, procedure revision, or PPE because those are the controls they actually see used at their site, rather than selecting elimination, substitution, or engineering controls.
Daily exposure to administrative controls and PPE normalises them as the default answer, and the candidate reads 'most effective' as 'most practical' or 'most commonly implemented'. Time pressure rewards the first familiar-sounding option.
What to do instead
Before answering any control question, write the hierarchy out in order (elimination, substitution, engineering, administrative, PPE) and rank every option against it; if the stem says 'most effective', 'most reliable', or 'at the source', the answer must be the highest feasible level, and PPE is only correct when the stem specifically asks about last-resort or interim protection.
The candidate treats hazard identification and risk assessment as the same activity, so on an item asking what a risk assessment must produce, they choose 'a list of hazards' or 'a walkthrough of the work area' instead of options involving likelihood, severity, and a risk rating.
In practice the two steps are often done in one meeting, so the boundary blurs; the candidate also conflates a hazard (source of harm) with risk (combination of likelihood and consequence).
What to do instead
Anchor the distinction verbally: identification answers 'what can hurt us', risk assessment answers 'how bad and how likely'. On every risk item, check that the chosen option contains both a probability/likelihood element and a consequence/severity element before selecting it.
The candidate answers exposure-limit items using whichever limit value they remember from a workplace SDS, mixing up OSHA PELs, NIOSH RELs, and ACGIH TLVs, or treating a TWA as if it were a ceiling or short-term limit.
All three are expressed in ppm or mg/m3 and all appear on safety data sheets and in industrial hygiene reports, so they look interchangeable; the candidate also forgets that a PEL is a legal minimum, not the most protective value.
What to do instead
Memorise the identity of each: PEL = enforceable regulatory limit, REL = NIOSH recommendation, TLV = ACGIH consensus guideline; TWA = 8-hour average, STEL = 15-minute, ceiling = never to be exceeded. When a stem asks for the most protective or recommended value, REL (or the lowest number) is usually correct, and when it asks what is enforceable, the PEL is.
On management-system items the candidate misplaces an activity within the Plan-Do-Check-Act cycle, for example classifying management review as part of 'Check' or treating a written policy as the 'Do' phase, and confuses distinct elements such as management commitment, roles and responsibilities, and management review.
The elements overlap in real audit documents and the candidate studies them as a flat list of clause headings rather than as a cycle with inputs and outputs.
What to do instead
Build a one-page PDCA map and place each element on it explicitly: policy, planning, objectives and hazard identification in Plan; implementation, training, and operational control in Do; monitoring, audit, and incident review in Check; management review, corrective action, and revised objectives in Act. Test yourself by naming the phase before naming the element.
When asked for a measure that shows whether a safety process is working proactively, the candidate selects total recordable incident rate, lost-time rate, or number of injuries, i.e. lagging outcomes, instead of leading indicators such as inspection completion, near-miss reporting rate, or training currency.
Injury rates are the numbers most often reported to leadership and are the easiest to recall, so they dominate the candidate's mental library of 'safety metrics'.
What to do instead
Classify every metric you meet as leading (activity, exposure, or preventive process before harm) or lagging (outcome after harm). For any stem containing 'predictive', 'proactive', 'before an incident', or 'leading', eliminate every rate based on injuries, illnesses, or days away.
During incident investigation items the candidate stops the causal chain at the injured worker's action, choosing 'retrain the employee' or 'reissue the procedure' as the corrective action, and labels the immediate cause as the root cause.
The immediate pre-incident act is the most visible fact in the report, and retraining is the cheapest, fastest corrective action, so it feels like a complete answer.
What to do instead
Force at least three 'why' levels past the individual's action and ask what management-system condition allowed it (design, purchasing, scheduling, supervision, training adequacy, maintenance). Eliminate any option whose only corrective action is retraining or discipline, and prefer options that change the system so the error cannot recur.
On emergency preparedness items the candidate picks a generic response such as 'evacuate the building' for a release of a heavier-than-air toxic gas, an approaching tornado, or an external civil disturbance, instead of shelter-in-place or a specific protective action.
Evacuation is the most familiar and most rehearsed response, and the candidate does not link the hazard's physical behaviour (density, wind, warning time) to the correct protective action.
What to do instead
Before choosing, classify the event by warning time and by whether the hazard is inside or outside and whether movement increases exposure. Practise matching: short-warning/severe-weather or outdoor release = shelter-in-place and HVAC shutdown; structural fire or bomb threat = evacuation; then verify the option also names who authorises the action and how it is communicated.
The candidate treats consensus standards such as ANSI, NFPA, or ACGIH documents as legally binding on every employer, or assumes an OSHA standard applies to all employers regardless of the cited industry or the general-duty clause requirement, and so selects a legally incorrect 'requirement'.
Consensus standards are cited inside regulations and referenced in company policy, which makes them feel mandatory; the candidate also does not distinguish 'shall' from 'should' or incorporated-by-reference standards from guidance.
What to do instead
For every legal item, first ask 'who makes this enforceable and against whom'. Only statutes, regulations, and standards incorporated by reference carry legal force; a consensus standard binds only when adopted by the employer or incorporated. Watch for the words shall/must versus should/recommended as the discriminator in the options.
The candidate confuses inherent risk with residual risk, or assumes a risk is 'closed out' once a control is selected, without re-evaluating the risk with the control in place and verifying that the control actually reduced it.
Risk registers usually display one number, so the candidate never sees the two states separated and reads the post-control score as if it described the raw hazard.
What to do instead
Read every risk stem twice, once as 'before any controls' and once as 'after the listed controls'. If the stem asks about the untreated hazard, answer from inherent risk; if it asks whether further action is needed, answer from residual risk and require evidence that the control was implemented and verified effective.
The candidate answers 'first action' or 'initial step' emergency and investigation items with an action that is correct but comes later in the sequence, such as beginning the investigation with interviews rather than securing and preserving the scene, or calling for evacuation without first activating the alarm and notifying the coordinator.
All the listed actions are legitimate parts of the response, and the candidate evaluates each on whether it is helpful rather than on whether it is the earliest in the required order.
What to do instead
For any sequence item, write the required order in shorthand before looking at the options (secure/isolate, preserve scene, notify, assess, investigate, correct) and pick the option that occupies the earliest unfilled slot; never select an option simply because it is a good practice.