CMQ/OE common mistakes
CMQ/OE · Understand the traps and how to avoid them.
On a control-chart or capability item, the candidate declares the process 'out of control' because a plotted point falls outside the specification limits, or calculates Cp/Cpk using the control limits as the tolerance.
Spec limits and control limits are both drawn as horizontal lines on the same chart, so under time pressure the candidate collapses them into one concept and reads whichever line is crossed as the trigger.
What to do instead
Say the definition out loud before answering: control limits come from the process itself (mean +/- 3 sigma of subgroup statistics) and answer 'is the process stable?'; specification limits come from the customer/design and answer 'is the process capable?'. Then check which question the stem actually asks and compute Cp/Cpk with USL-LSL in the numerator, sigma from the process.
When categorizing a cost-of-quality item, the candidate files rework, scrap and warranty claims under appraisal, or files inspection and test under prevention.
Both prevention and appraisal feel like 'quality activities the company chooses to fund', so the candidate sorts by whether the activity sounds proactive rather than by whether a defect already exists.
What to do instead
Apply one test: prevention = money spent so a defect never occurs (training, design review, SPC setup, supplier partnering); appraisal = money spent to find defects that may already exist (inspection, test, audit, calibration); internal failure = defect found before shipment (scrap, rework, re-test); external failure = defect found after shipment (warranty, returns, recalls, liability). Rework is always failure, never appraisal.
The candidate labels alpha as the consumer's risk and beta as the producer's risk, or picks the wrong risk when a sampling plan's acceptance number is raised or lowered.
The Greek letters carry no mnemonic and the two error types are mirror images, so candidates memorize a pairing without an anchor and then invert it under pressure.
What to do instead
Anchor on consequences: alpha (Type I) rejects a good lot, which hurts the producer, so alpha = producer's risk; beta (Type II) accepts a bad lot, which hurts the consumer, so beta = consumer's risk. Then reason on the operating characteristic curve: raising the acceptance number makes acceptance easier, so producer's risk falls and consumer's risk rises.
Presented with a scenario about turning strategy into annual operating actions with cross-functional negotiation of targets, the candidate selects SWOT analysis or the balanced scorecard instead of Hoshin planning/catchball.
All strategic-planning tools live in one memorized list, and the candidate pattern-matches on the word 'strategy' rather than on the specific verb in the stem (analyze, deploy, measure, negotiate).
What to do instead
Match tool to verb: 'assess internal/external environment' = SWOT; 'translate strategy into four perspectives of measures' = balanced scorecard; 'deploy breakthrough objectives down through the hierarchy with dialogue to align targets' = Hoshin kanri with catchball; 'identify which few tasks must be done perfectly' = QFD/prioritization. Read the last clause of the stem first, since it names the verb.
On situational-leadership items the candidate chooses the style that matches their own workplace preference (usually democratic or coaching) instead of the style the stem's follower-readiness and task-clarity cues call for.
Leadership questions feel like values questions, so the candidate answers what a good leader 'should' be rather than diagnosing the situation the stem describes.
What to do instead
Extract two facts from the stem before looking at options: how competent and committed the follower(s) are on this specific task, and how clear/structured the task is. Low readiness plus ambiguous task points to telling/directing; high readiness plus complex task points to delegating. Eliminate any option that ignores the stated readiness level, even if it is the more attractive style.
The candidate treats 'met the specification' as equivalent to 'satisfied the customer', and on Kano-model items classifies a must-be (dissatisfier) attribute as a delighter.
The exam's language of conformance and nonconformance trains attention on specs, and candidates carry that frame into customer-focused items where the customer's perception, not the drawing, is the criterion.
What to do instead
Ask two questions: does the stem describe an internal measure or a customer perception? And is the attribute one whose absence causes complaints but whose presence is taken for granted (must-be), one proportional to investment (performance), or one that surprises and delights (attractive)? Absence of complaints is not evidence of satisfaction; measure it directly.
The candidate orders Kirkpatrick's levels incorrectly, or places a question about trainee satisfaction or exam scores at the 'results' level.
The four levels are short, abstract labels with no intrinsic sequence, so they get recalled as a set rather than as a ladder, and 'results' sounds like any outcome including a test score.
What to do instead
Fix the ladder as reaction (did they like/engage with it) -> learning (knowledge, skill, attitude gain, i.e., test scores) -> behavior (on-the-job application) -> results (business metrics: defects, cycle time, cost, retention). Note that level 3 requires a follow-up after the class and level 4 requires a metric the training did not itself produce.
Given a team scenario, the candidate recommends 'have the team vote' when the stem describes disagreement in a storming team, or prescribes norming-stage facilitation for a team that is still forming.
Voting and team-building sound like constructive moves, so candidates choose them without first labeling the team's developmental stage or noticing that the stem is asking how consensus, not majority rule, is reached.
What to do instead
Label the stage from the stem's symptoms before choosing an intervention: forming = polite, dependent, needs direction and clear roles; storming = conflict over goals/roles/power, needs facilitated conflict resolution and ground rules; norming = cohesion emerging, needs reinforcement of agreements; performing = self-directed, needs autonomy and recognition. Also distinguish consensus (everyone can support and commit) from majority vote (a minority is outvoted), and pick consensus whenever the stem emphasizes commitment to the decision.
On acceptance-sampling items the candidate reverses the AQL/LTPD relationship to risk, or reads an OC curve as if lower probability of acceptance means a better lot.
AQL and LTPD are both 'percent defective' numbers, so the candidate confuses which end of the curve they sit on and which risk attaches to each.
What to do instead
Sketch the OC curve on scratch paper: x-axis is lot percent defective increasing left to right, y-axis is probability of acceptance. AQL is the good-lot quality at the left, where Pa is high (e.g., 0.95) and producer's risk (alpha = 1 - Pa) is low; LTPD is the bad-lot quality at the right, where Pa is low (e.g., 0.10) and consumer's risk (beta) is low. Steeper curves discriminate better; a tightened plan shifts the whole curve left.
On a corrective-action scenario that starts with a customer complaint or a failed audit, the candidate's chosen first step is to convene a root-cause team (5 Whys, fishbone) without first containing the problem, or jumps to training as the corrective action.
Root-cause analysis is the most rehearsed tool in the candidate's repertoire, so it becomes the reflexive answer to any 'what should be done' stem, and the word 'first' in the stem gets skimmed.
What to do instead
Memorize the sequence as containment -> verification of the problem and its scope -> root cause (occurrence and escape) -> corrective action -> verification of effectiveness -> standardization/read-across. Containment (quarantine, sort, notify the customer, correct the immediate condition) always precedes investigation because it stops the defect from escaping again while the cause is unknown. Training alone is a corrective action only when the root cause is genuinely a knowledge/skill gap, not when the process allowed the error.