23-Ind-B5 Ergonomics · December 2017
Nivaar worked solution (AI-drafted; not reviewed by a licensed engineer)
National Exams — Dec. 2017 — 98-Ind-B5 Ergonomics. Three-hour, open-book exam (all notes, books and any non-communicating calculator permitted); the paper's own instructions state "a total of five (5) questions" while the printed marking scheme and all eight pages contain exactly four (Part A, Questions 1–2, mandatory; Part B, Questions 3–4, choose one) — a minor editorial inconsistency in the source, not a dropped question. The paper requires 4 of its own marked total; all four are solved below for completeness.
Reference texts: Sanders & McCormick, Human Factors in Engineering and Design (7th ed.) — error taxonomy, task analysis, mental workload measurement, macroergonomic assessment process; Waters, Putz-Anderson & Garg, NIOSH Applications Manual for the Revised NIOSH Lifting Equation (1994) — the RWL/LI formula and multiplier tables reproduced on the exam's own pages 7–8; NIOSH, Elements of Ergonomics Programs (1997) and CSA Z1004 (Canada) — workplace musculoskeletal-disorder (MSD) prevention programs; CSA Z1002 — hazard identification, elimination and risk assessment; ISO 11228 series — manual handling limits.
Question text not reproduced: the examination questions are © Engineers and Geoscientists BC. Open the official past paper (linked at the top of this page) to read the question, then follow the worked solution below.
Norman's dictum reflects a design principle, not a personnel one: error is a predictable output of the interaction between a task, an interface and a novice, part-time workforce, so the fix belongs in the system, not in blaming the operator. In this shop the principal sources of error are (1) high staff turnover and on-the-job learning, which means many operators are always partway up the learning curve; (2) inconsistent control/display design across the print, photocopy and binding equipment (different manufacturers, different button layouts and labelling conventions), forcing workers to re-learn mappings machine to machine; (3) dual, incompatible job-intake channels (email files vs. walk-in requests), which creates opportunities for a request to be lost, duplicated, or matched to the wrong customer; and (4) time pressure at peak periods (printing deadlines), which pushes operators toward faster, less-checked action sequences.
Using Norman's/Rasmussen's taxonomy, the errors that follow fall into two broad classes. Slips (execution errors – the operator has the right intention but the action goes wrong) are most likely on skill-based, highly practised sub-tasks: capture errors (starting a familiar sequence, e.g. "select paper size", but sliding into a more habitual one from a different machine); description errors (pressing the button that is spatially or visually similar to the intended one among the shop's inconsistent control sets); and mode errors (acting on a machine as though it were in a mode it is not in – e.g. sending a job to the machine that is still finishing a binding cycle). Mistakes (planning errors – the intention itself is wrong) dominate for a partly-trained workforce: rule-based mistakes (correctly applying a rule learned on machine A to machine B, where it does not apply – the classic negative-transfer error of a multi-machine, inconsistent-interface shop); and knowledge-based mistakes (a genuinely unfamiliar job, e.g. an unusual binding request, solved by ad-hoc reasoning that is wrong because the worker lacks the underlying knowledge).
Each error type has a distinct, matched mitigation rather than a single blanket fix:
Given. A shop with part-time, on-the-job-trained student workers, several distinct task types (printing, photocopying, binding, customer intake), and a management need to quantify how error affects performance, workload, efficiency and cost.
Approach. A mixed-methods field study is appropriate: error frequency/type cannot be inferred from cost records alone, and workload cannot be inferred from error logs alone, so direct observation/logging is paired with a validated subjective workload instrument and the shop's own cost data.
Timeline (approx. 6 weeks). Week 1: instrument design and staff briefing (informed-consent, non-punitive framing so errors are reported honestly). Weeks 2–3: baseline data collection across all shifts and task types (to capture the full range of workers, from newly trained to experienced). Week 4: preliminary analysis and identification of the highest-impact error categories. Weeks 5–6: a short validation window re-measuring the same tasks after any quick-win fix identified in Part (a) is trialled, to confirm the measurement approach is sensitive enough to detect a change.
Human resources. One human-factors analyst/observer (design, run and analyze the study); the shift supervisor (co-ordinates logging compliance, provides existing production/cost records); 2–3 student workers per shift as the observed population, spanning the training-tenure range (first week through experienced); and short-term IT/admin support to extract point-of-sale and job-queue timestamp data already generated by daily operations.
Tasks. (1) A structured error-incident log completed at the point of occurrence (or immediately after, to avoid recall loss), recording error type per the Part (a) taxonomy, the task/machine involved, time lost, and material wasted; (2) unobtrusive direct observation/video sampling of a subset of shifts to capture near-misses and errors workers do not self-report; (3) a NASA Task Load Index (NASA-TLX) survey administered at the end of each observed shift, giving a validated multidimensional workload score (mental, physical and temporal demand, effort, frustration, perceived performance); (4) extraction of existing cost data – wasted material, reprints, void transactions and job turnaround time – from the shop's own point-of-sale/production records for the same shifts.
Anticipated outcomes. An error rate per operator-hour broken down by task type and error class; a correlation between NASA-TLX workload score and error rate (testing whether error is workload-driven, training-driven, or interface-driven); a dollar cost of errors (wasted material + reprint labour + lost throughput); and a ranked list of the highest-cost task/error combinations to prioritize for the redesign recommended in Part (a) and Part (c).
Approach. The system is built directly around the two dominant error classes identified in Part (a): it teaches consistent control mappings (to reduce slips) and gives a safe space to make and recover from knowledge-based mistakes before doing so on a live customer job (to reduce mistakes), with progress visibility for the manager who requested it.
Each stage is tied to a specific principle: the task-module selector mirrors the shop's real task set (print/photocopy/binding/intake) rather than a generic "orientation" course, so a worker only needs to complete the modules relevant to the station they will actually run – consistent with training being targeted to the transfer task. The interactive scenario video presents the SAME control layout and labelling convention recommended in Part (a) for every machine type, so the training itself reinforces standardized mappings rather than teaching machine-specific quirks. The error-recovery simulation is the core mitigation for knowledge-based mistakes: rather than only describing the correct procedure, it lets the trainee deliberately choose a wrong action (e.g. the wrong binding method for a given page count) and see the realistic consequence (wasted material, a customer complaint) in a zero-cost environment, building the "recognize-then-recover" skill before it is needed on a live job. The knowledge-check quiz is a gate, not just a record, so a worker cannot be assigned to a station until competence on that module is demonstrated. The manager dashboard closes the loop with Part (b)'s measurement study – low quiz scores or repeated simulation failures on a specific module flag exactly which task/worker pairing needs supervised practice before that worker is trusted with it unsupervised.
