CQIA common mistakes
CQIA · Understand the traps and how to avoid them.
When a question names a quality concept without naming the person, the candidate assigns it to whichever guru they memorized most recently (e.g., answering 'Deming' for 'quality trilogy: planning, control, improvement' or for 'zero defects').
The gurus share vocabulary (quality planning, improvement, control, cost of quality), and candidates study the names as isolated flashcards rather than anchoring each name to its distinctive deliverable.
What to do instead
Build a one-row-per-guru table with the distinctive artifact, not the philosophy: Juran = quality trilogy and Pareto emphasis; Deming = 14 points, PDCA, reduction of fear, no slogans; Crosby = zero defects, cost of nonconformance, 'quality is free'; Feigenbaum = total quality control, systems view; Ishikawa = fishbone diagram, company-wide quality. Before answering, ask 'what artifact is being described?' and map only from that column.
Classifying cost-of-quality items by the activity performed rather than by when the defect was caught, e.g., calling rework 'appraisal' because someone inspects it, or calling supplier audits 'internal failure' because problems were found.
Categories feel intuitive by department (inspection department = appraisal) instead of by purpose (preventing defects vs. evaluating vs. failing).
What to do instead
Apply two questions in order: (1) Is the cost incurred to stop defects from happening? -> prevention. (2) Is it incurred to judge conformance of product already made? -> appraisal. If product was already nonconforming: internal failure before delivery, external failure after. Drill the boundary pairs that get tested: rework before shipment = internal failure; field service/warranty = external failure; training and process design = prevention; test equipment calibration and final inspection = appraisal.
Treating the team sponsor/champion as the day-to-day team leader, or assuming the facilitator sets the agenda and owns results.
Real-world titles vary between organizations, so candidates answer from their own employer's usage instead of the exam's role definitions.
What to do instead
Memorize responsibilities, not titles: sponsor/champion = authority, resources, removes organizational barriers, receives results; team leader = runs meetings, owns schedule and deliverable; facilitator = process and meeting-effectiveness expert, often neutral, does not own content; recorder/scribe captures data and minutes; team member = content. If an option describes 'approving budget' it is the sponsor even if the word facilitator appears.
Mixing up team development stages, e.g., placing conflict ('storming') after norms are set, or labeling polite dependence as 'norming'.
the stages are learned as a list of five words rather than as observable behaviours, and polite silence feels like harmony.
What to do instead
Attach one behaviour to each stage and answer from behaviour: forming = politeness, dependence on leader, unclear purpose; storming = disagreement over goals/roles, resistance; norming = rules of conduct agreed, trust grows; performing = self-directed, focus on results; adjourning = disbanding and recognition. Any scenario containing conflict or challenge to the leader is storming, regardless of how far the team 'feels' along.
Answering improvement-methodology questions with the wrong step sequence or vocabulary, such as describing DMAIC as beginning with 'define the problem' but then listing 'plan-do-check-act' phases, or claiming kaizen blitz includes months of measurement.
PDCA, DMAIC and kaizen events all start with 'understand the problem', so candidates blur them into one generic loop under time pressure.
What to do instead
Fix three reference frames: PDCA = Plan/Do/Check/Act, the universal cycle; DMAIC = Define/Measure/Analyze/Improve/Control, project-based, data-heavy; kaizen blitz/event = short, focused, action-oriented team improvement over days. Check the answer option for a step that belongs to another framework (e.g., 'Measure' inside PDCA, or 'Control' as a kaizen phase) and eliminate it.
Reacting to every data point or every process fluctuation as a special cause, e.g., adjusting a process after a single point inside control limits, or declaring a supplier 'bad' because of one high reading.
Candidates know the words 'common' and 'special' cause but not the operational test: variation within limits has a stable system cause, so reacting is tampering and increases variation.
What to do instead
Use the control-chart test as the decision rule: points within limits = common cause, act on the system (management); points beyond limits or non-random patterns (runs, trends, alternation) = special cause, act on the event. When an exam option says 'investigate and adjust the machine after each high reading', it describes tampering and is wrong.
Building a Pareto chart by plotting raw frequencies and stopping there, or asserting that 80% of problems always come from 20% of causes; also sorting categories by cost when the data were counts.
The 80/20 line is quoted as a law, and the chart is drawn as a simple bar chart without cumulative percent or a ranked 'vital few' interpretation.
What to do instead
Remember the Pareto chart is bars in descending order plus a cumulative-percentage line, and the output is a prioritized short list of vital few causes. The 80/20 split is an observation, not a guarantee - read the actual cumulative curve. Never combine incompatible units on one axis; convert to a common measure (cost, counts) before ranking.
Choosing the wrong control chart type for the data, e.g., selecting an X-bar and R chart for single measurements, or a p chart for defect counts per unit.
Chart selection is remembered by name recognition rather than by two questions: attribute vs. variable, and subgroup size/constant sample size.
What to do instead
Answer with a decision tree: measurement data (variables) with rational subgroups -> X-bar and R (or s); single readings -> individuals and moving range (I-MR); attribute data with a constant sample size -> np or c; attribute data with a varying sample size -> p or u. Then separate limits from specifications: control limits come from the process, specification limits come from the customer, and never plot spec limits as control limits.
Drawing cause-and-effect diagrams with causes that are actually solutions or restated effects ('the defect is caused by defects'), or running 5 Whys and stopping at the first plausible-sounding answer or at a person's name.
Teams brainstorm in the language of fixes, and the five-whys chain is ended as soon as the answer feels actionable rather than when it reaches a controllable system cause.
What to do instead
Apply two filters on every branch: it must be a cause, not an effect or countermeasure, and it must be verifiable with data. Use the standard 5M1E or 4M1E skeleton (manpower, machine, material, method, measurement, environment) to force breadth instead of a single chain. In 5 Whys, keep asking until you reach something the team controls (a process condition), and reject 'operator error' as terminal unless it is traced to a process or training condition.
Using accuracy and precision interchangeably, or answering measurement-system questions with the wrong term for repeatability vs. reproducibility.
Everyday speech uses 'accurate' to mean 'good', and the two MSA terms sound alike.
What to do instead
Anchor with the target analogy and a source: accuracy = closeness to the true/reference value (bias, on average); precision = closeness of repeated readings to each other (spread). Within the measurement system, repeatability = same appraiser, same instrument, repeated trials (equipment variation); reproducibility = different appraisers using the same instrument (appraiser variation). If the option mentions two or more operators, it is reproducibility.