PMI PMP Quality Planning and Assurance Practice Test

 

Topic 16 covers quality planning and assurance for the PMI Project Management Professional (PMP) certification. These original scenarios apply the July 2026 exam objectives across predictive, agile and hybrid projects. Use the stated constraints to select one answer unless the question specifies otherwise. For broader preparation, visit the PMP Exam Dumps page. Each option has an explanation of its role in the decision.

Question 1

A team is planning quality checks for a customer portal. The approved requirement is that at least 95% of searches finish within two seconds at 500 concurrent users. Which planned measure most directly verifies this requirement?

1. Measure the 95th percentile with one user before load testing.
2. Measure the fastest response observed during a full-load test.
3. Measure the 95th-percentile response time at the required load.
4. Measure the number of searches completed during each test hour.
5. Measure the mean response time with 500 concurrent users.

Correct Answer: C

 

Correct Answer

Answer C is correct because the requirement combines a response-time threshold, a percentile and an operating load. Measuring that percentile at 500 concurrent users tests all three conditions instead of substituting a convenient average.

Incorrect Answers

Answer A is incorrect because a single-user test can help establish a baseline, but contention at 500 users may change response times. The stated operating condition is essential to verification.

Answer B is incorrect because the best observed result describes only one favorable search. It provides no evidence about how many other searches exceed the two-second limit.

Answer D is incorrect because throughput can be useful for capacity planning, but a high completion count does not establish the proportion completed within two seconds.

Answer E is incorrect because a low mean can conceal a slow tail affecting more than 5% of searches. It preserves the load condition but does not establish the required percentile.

 

Question 2

A hybrid data-conversion project passes final sample inspections only after repeated rework. Managers want evidence that the conversion process can consistently produce acceptable output before the next release. Which activity best addresses that request?

1. Count the staff hours spent fixing the last release.
2. Ask the customer to approve the already corrected sample.
3. Audit the mapping controls and evidence of their use.
4. Compare the next release date with the contractual milestone.
5. Increase the number of records checked at the final inspection.

Correct Answer: C

 

Correct Answer

Answer C is correct because the recurring rework points to how work is performed. Examining the planned controls and evidence of their use provides assurance about the process that produces each release.

Incorrect Answers

Answer A is incorrect because rework effort indicates the burden of failure. It does not demonstrate that the process controls are designed well and actually followed.

Answer B is incorrect because customer acceptance of that sample addresses a particular output. It does not explain why defects repeatedly enter the conversion process.

Answer D is incorrect because schedule conformance is relevant to delivery, but the requested confidence concerns consistent quality. An on-time release can still require repeated defect correction.

Answer E is incorrect because a larger inspection may detect more bad output, but it does not establish whether the mapping and review process prevents the recurring errors.

 

Question 3

A stable machining process has individual-value control limits of 9.70 and 10.30 mm. Customer specifications are 9.90 to 10.10 mm. A measured unit is 10.20 mm and the chart has no special-cause signal. What is the correct interpretation?

1. The unit is acceptable because it lies within the control limits.
2. The upper control limit should be reduced to the upper specification immediately.
3. The specification should be widened to match the observed process limits.
4. The unit proves that the process has a special cause of variation.
5. The unit is nonconforming despite the absence of a process signal.

Correct Answer: E

 

Correct Answer

Answer E is correct because control limits describe expected process behavior; specification limits define acceptable output. A value of 10.20 is within the former and outside the latter, so stability does not imply conformance.

Incorrect Answers

Answer A is incorrect because this substitutes process variation limits for customer acceptance limits. The customer permits no value above 10.10 mm, regardless of the chart signal.

Answer B is incorrect because changing a chart limit to match a requirement does not change the process distribution. Control limits should be derived using an appropriate statistical method.

Answer C is incorrect because a wider tolerance would change the customer requirement. Process capability problems do not authorize the project to redefine acceptance unilaterally.

Answer D is incorrect because an unacceptable unit can arise from the ordinary variation of a stable but incapable process. The stated chart evidence does not identify a special cause.

 

Question 4

An iterative service-improvement team uses a control chart with an agreed signal rule of eight consecutive points above the center line. The last eight points are above that line but inside the control limits. What should the team do?

1. Recenter the chart now so that the latest points are evenly distributed.
2. Continue unchanged until a point exceeds the upper control limit.
3. Remove the highest point and recalculate the chart from the remaining seven.
4. Declare the service defective because the center line was exceeded.
5. Investigate a possible process shift and annotate relevant process changes.

Correct Answer: E

 

Correct Answer

Answer E is correct because the agreed run rule detects a nonrandom pattern even when individual points remain within limits. Investigation should establish the cause before deciding whether the shift is beneficial or harmful.

Incorrect Answers

Answer A is incorrect because new limits may eventually be appropriate after a verified process change. Immediate recentering would hide the unexplained shift and destroy the comparison needed for investigation.

Answer B is incorrect because that would ignore the explicitly adopted run rule. A sequence can signal a shift before any individual observation crosses a limit.

Answer C is incorrect because there is no evidence that the highest observation is invalid. Selective deletion would suppress the signal instead of explaining the process behavior.

Answer D is incorrect because the center line is a process reference, not a customer acceptance threshold. A shift warrants investigation but does not by itself prove nonconforming service.

 

Question 5

A packaging process is stable, with no signals under its agreed control-chart rules, but its ordinary variation produces too many damaged packages. Operators propose adjusting machine settings after every high or low observation. What is the better response?

1. Investigate every observation as a separate special-cause incident.
2. Replace the control limits with the customer damage threshold.
3. Keep the current process because statistical control establishes acceptable quality.
4. Improve the underlying process and verify the resulting performance.
5. Approve individual adjustments whenever an observation is above average.

Correct Answer: D

 

Correct Answer

Answer D is correct because the evidence points to common-cause variation in a stable process that is not good enough. Systematic improvement addresses that capability gap; reacting to each random fluctuation can increase variation.

Incorrect Answers

Answer A is incorrect because the chart provides no special-cause signal. Treating all ordinary variation as isolated incidents directs attention away from the process design that generates it.

Answer B is incorrect because an acceptance requirement and a statistical control limit serve different functions. Replacing the limits does not improve packaging performance or diagnose its variability.

Answer C is incorrect because stability means the process is predictable under the current conditions. A predictable level of damage can still violate the required quality level.

Answer E is incorrect because above-average observations are expected in a stable process. Changing settings in response to noise can create additional variability without removing the underlying source of damage.

 

Question 6

A normally distributed, stable production process has a mean of 11, standard deviation 0.5, and specification limits of 8 and 12. The quality lead reports Cp = 1.33. Using Cpk = min[(USL – mean)/(3 sigma), (mean – LSL)/(3 sigma)], what does Cpk show?

2. Cpk is 2.00; the larger distance from a specification is controlling.
3. Cpk is 0.67; the mean approaches the upper limit.
4. Cpk is 1.00; the mean remains between the specification limits.
5. Cpk is about 0.33; the denominator must contain six standard deviations.
6. Cpk is about 1.33; Cp already incorporates the process mean.

Correct Answer: B

 

Correct Answer

Answer B is correct because the upper-side ratio is (12 – 11)/1.5 = 0.67; the lower-side ratio is (11 – 8)/1.5 = 2.00. The smaller value reveals the centering problem that Cp alone misses.

Incorrect Answers

Answer A is incorrect because the lower-side ratio is 2.00, but capability is limited by the nearer specification. Taking the larger ratio hides the upper-tail conformance risk.

Answer C is incorrect because having an in-specification mean does not determine capability. The distribution spreads around that mean, and its distance from each limit must be considered.

Answer D is incorrect because the stated Cpk formula uses three standard deviations for each one-sided distance. Six standard deviations belong to the full-width Cp calculation.

Answer E is incorrect because the Cp index uses specification width divided by six standard deviations and does not locate the mean. Cpk is lower here because the process is shifted toward the upper specification.

 

Question 7

Testing a battery to destruction is the only validated way to measure its failure temperature. The approved quality plan permits statistically designed acceptance sampling for a large production lot. What is the most appropriate inspection approach?

1. Double the planned sample and accept on its mean failure temperature.
2. Replace temperature testing with a visual inspection of all batteries.
3. Test the first conveniently available batteries and generalize to the lot.
4. Accept the lot if the supplier provides its previous lot results.
5. Use the approved random sample and apply its acceptance and rejection rules.

Correct Answer: E

 

Correct Answer

Answer E is correct because destructive testing makes inspecting every saleable unit impractical. The approved sampling design provides a controlled decision method with stated risks rather than pretending to prove that every unit is good.

Incorrect Answers

Answer A is incorrect because a larger sample is not a substitute for the approved decision rule. An acceptable mean could conceal units failing the temperature criterion, and changing sample size or acceptance logic requires a justified plan revision.

Answer B is incorrect because visual inspection has not been validated for this characteristic. Complete coverage of an unrelated attribute cannot establish the required failure-temperature performance.

Answer C is incorrect because convenience selection may miss variation across production periods or positions. The approved random design is intended to support a defensible lot-level inference.

Answer D is incorrect because previous results may inform risk but do not constitute the required evidence for the present lot. The approved plan calls for sampling this production lot.

 

Question 8

A hybrid rollout reports twice as many installation defects on night shifts. Night shifts also perform three times as many installations and use a different device model. Before assigning the cause to the shift, which analyses are needed? Choose TWO.

1. Rank shifts using total defect counts because all defects matter.
2. Treat the shift difference as causal because it appears in two releases.
3. Average both shifts together before judging the device models.
4. Stratify results by device model as well as shift.
5. Use only day-shift records to establish a stable quality baseline.
6. Compare defect rates per installation, not just defect counts.

Correct Answers: D, F

 

Correct Answers

Answer D is correct because device model and shift vary together. Separating these factors can reveal whether a product-specific problem is being mistakenly attributed to the time of work.

Answer F is correct because the exposure differs by a factor of three. A higher raw count can coexist with a lower rate, so the denominator is necessary before comparing quality performance.

Incorrect Answers

Answer A is incorrect because every defect matters operationally, but raw totals do not isolate the source of differing performance when workload and device mix are unequal.

Answer B is incorrect because repetition strengthens the reason to investigate, but it does not remove the recurring workload and device-model differences that could explain the association.

Answer C is incorrect because pooling the observations can conceal the subgroup difference under investigation. Stratification is needed before a combined result can be interpreted safely.

Answer E is incorrect because excluding the questioned population avoids the comparison. It cannot explain whether the night-shift pattern is caused by exposure, device mix or another factor.

 

Question 9

An agile product team has 70 cosmetic defects, 20 report-formatting defects and one defect that can expose another customer’s records. Its release policy blocks any confirmed cross-customer disclosure. Which conclusions should guide quality prioritization? Choose TWO.

1. Downgrade the disclosure because its observed frequency is less than 2%.
2. Resolve cosmetic defects first because they dominate the Pareto count.
3. Address the disclosure defect as a release blocker under the stated policy.
4. Use the category counts to target recurring rework after applying the severity constraint.
5. Discard the frequent categories once a release blocker has been found.
6. Combine all categories into one count to make the priority objective.

Correct Answers: C, D

 

Correct Answers

Answer C is correct because the policy makes this consequence decisive regardless of frequency. A count-based ranking cannot override the explicit release condition for cross-customer disclosure.

Answer D is correct because the distribution still helps identify where frequent defects consume effort. It is useful alongside severity and release rules, rather than as a substitute for them.

Incorrect Answers

Answer A is incorrect because the policy is triggered by a confirmed defect, not by a frequency threshold. Rare occurrence does not negate the specified confidentiality consequence.

Answer B is incorrect because a Pareto view by frequency identifies volume, not necessarily consequence. The single disclosure defect violates an explicit release condition and cannot wait for cosmetic cleanup.

Answer E is incorrect because blocking release addresses immediate eligibility, but recurring cosmetic and formatting defects still represent quality losses. Their frequency remains relevant to subsequent process improvement.

Answer F is incorrect because the total would erase the distinction that matters most: one category prevents release. Combining unlike consequences does not make prioritization more informative.

 

Question 10

A team draws a cause-and-effect diagram for intermittent data loss. The diagram lists network congestion, retry logic and batch size as possible causes. What should the project manager request before approving a permanent corrective action?

1. Implement the least expensive suspected fix and close the causal investigation.
2. Ask participants to vote on the cause and treat the majority as validation.
3. Collect evidence that tests the most plausible causal hypotheses.
4. Approve enhanced monitoring as the permanent response without testing any cause.
5. Approve a separate fix for every branch in the diagram.

Correct Answer: C

 

Correct Answer

Answer C is correct because the diagram organizes possible causes but does not establish which causes actually produce the loss. Targeted logs or experiments can distinguish them before a permanent change is selected.

Incorrect Answers

Answer A is incorrect because low implementation cost may influence the choice among validated responses, but it does not show that this change addresses the data-loss mechanism. Evidence is needed before declaring a permanent correction.

Answer B is incorrect because a vote can help prioritize investigation, but agreement is not a test of causation. A popular explanation may be inconsistent with the actual failure evidence.

Answer D is incorrect because monitoring can support investigation and reveal recurrence, but it does not by itself remove the cause. The requested permanent corrective action still needs a causal and effectiveness basis.

Answer E is incorrect because changing every suspected factor increases effort and may create new defects. The diagram provides hypotheses that should be investigated, not a verified list of required fixes.

 

Question 11

A scatter plot shows that releases with more test hours also have more defects found. The sponsor concludes that testing causes poor quality and proposes reducing it. Which interpretation is best supported?

1. The association does not establish causation; release size and detection effort need investigation.
2. Reducing test hours will lower the underlying defect rate by the same proportion.
3. The relationship proves that every release requires more testing.
4. Compare raw defect totals across releases without accounting for release size.
5. The positive relationship proves that the test process is defective.

Correct Answer: A

 

Correct Answer

Answer A is correct because larger releases may require more testing and contain more defects, while additional testing can simply reveal existing defects. The plot alone cannot show that testing creates them.

Incorrect Answers

Answer B is incorrect because the data show association, not a response to controlled changes in test effort. Fewer detected defects can mean weaker detection rather than better product quality.

Answer C is incorrect because the pattern does not establish an optimal test level for each release. Risk, size, coverage and diminishing returns must be considered before changing the plan.

Answer D is incorrect because larger releases can contain more defects and require more test effort. Keeping the raw association without controlling relevant size differences leaves the causal interpretation unresolved.

Answer E is incorrect because a productive test process may find more defects when given more time. The sign of the association does not distinguish successful detection from faulty work.

 

Question 12

A service team classifies every processed claim as either conforming or nonconforming. Daily sample sizes vary from 80 to 200. It wants to monitor the proportion of nonconforming claims over time. Which chart best fits these data?

1. A c chart of the number of defects across all daily claims.
2. A histogram combining all daily sample results.
3. An individuals chart of the claim payment amounts.
4. An np chart using one fixed sample size for all days.
5. A p chart with limits that account for the daily sample sizes.

Correct Answer: E

 

Correct Answer

Answer E is correct because the observation is a binary classification and the target statistic is a proportion. A p chart can accommodate changing sample sizes through appropriately calculated limits.

Incorrect Answers

Answer A is incorrect because a c chart addresses defect counts for a constant opportunity basis, and a unit may have multiple defects. Here the measure is binary nonconformance with varying sample sizes.

Answer B is incorrect because a histogram can summarize a distribution but discards time order. The team needs ongoing monitoring of changes in a daily proportion.

Answer C is incorrect because payment amount is a continuous variable, but it is not the requested quality statistic. Its variation does not establish the fraction of claims classified as nonconforming.

Answer D is incorrect because an np chart tracks counts of nonconforming units with a constant sample size. Pretending the varying daily samples are equal would distort the comparison.

 

Question 13

A release-quality log contains four activities: preparing error-prevention training, inspecting completed modules, correcting a defect before delivery, and repairing a defect reported after delivery. Which classifications are correct? Choose TWO.

1. Inspecting completed modules is an appraisal activity.
2. Inspecting completed modules is an external failure activity.
3. Preparing error-prevention training is an internal failure activity.
4. Correcting the pre-delivery defect is an appraisal activity.
5. Repairing the customer-reported defect is an external failure activity.
6. Repairing the delivered defect is a prevention activity.

Correct Answers: A, E

 

Correct Answers

Answer A is correct because inspection evaluates completed work against requirements. It incurs a cost to detect nonconformance rather than to prevent its creation or repair an already identified failure.

Answer E is correct because the defect was discovered after delivery to the customer. That delivery boundary distinguishes this failure category from internal correction before release.

Incorrect Answers

Answer B is incorrect because the inspection is a planned evaluation, not a loss arising from a defect discovered after delivery. The activity described does not cross the customer-delivery boundary.

Answer C is incorrect because planned training is intended to avoid defects before they occur. It is prevention, even if its content draws on lessons from previous failures.

Answer D is incorrect because the inspection that discovered the defect is appraisal, but fixing the defective output is internal failure work. Detection and correction should not be merged.

Answer F is incorrect because repair restores output that has already failed in the customer environment. A separate process change to avoid recurrence could be preventive, but the repair itself is not.

 

Question 14

A project wants to reduce the creation of defects, while tracking inspection and repair separately. Which activities belong in its prevention category? Choose THREE.

1. Design an input control that rejects invalid combinations before records are created.
2. Rework a component rejected during final testing.
3. Develop a quality plan with agreed process controls before work starts.
4. Train the team on the validated assembly method before production begins.
5. Inspect a completed batch against the acceptance checklist.
6. Replace units returned under warranty.

Correct Answers: A, C, D

 

Correct Answers

Answer A is correct because the control prevents a known error from entering the product. Its purpose is to avoid nonconforming output, rather than inspect completed records for existing defects.

Answer C is correct because quality planning establishes how requirements will be achieved and controlled. This planned system design is prevention, distinct from the later execution of inspection activities.

Answer D is correct because the training builds the capability to perform the work correctly at its source. It is planned before production to reduce the likelihood of assembly errors.

Incorrect Answers

Answer B is incorrect because the component has already failed to conform and is being corrected before delivery. That is an internal failure cost, even though correction avoids a later customer problem.

Answer E is incorrect because this activity appraises output that already exists. It can prevent defective shipment, but it does not make the inspection itself a prevention-cost activity.

Answer F is incorrect because replacement responds to failures discovered after delivery. The action belongs to external failure, while any separate redesign intended to stop recurrence could be prevention.

 

Question 15

Two pilots delivered equal volumes with identical acceptance criteria. Pilot A used 20 hours of prevention, 30 of appraisal and 50 of failure work. Pilot B used 35, 30 and 15 hours respectively. What conclusion about quality effort is supported?

1. Both pilots used the same quality effort because appraisal was unchanged.
2. Pilot A was more efficient because its prevention effort was 15 hours lower.
3. Pilot B used 20 fewer quality hours, with less failure work.
4. Pilot B proves a permanent 20% reduction under every future workload.
5. Pilot B eliminated the need for appraisal because failure hours declined.

Correct Answer: C

 

Correct Answer

Answer C is correct because the totals are 100 hours for A and 80 for B. The shift is consistent with a useful prevention investment, although two pilots alone do not prove that prevention caused the entire reduction.

Incorrect Answers

Answer A is incorrect because equal appraisal hours do not imply equal total quality effort. Prevention and failure hours differ, producing totals of 100 and 80.

Answer B is incorrect because prevention is only one component. Ignoring appraisal and failure omits the additional 35 hours of failure effort that more than offsets A’s lower prevention effort.

Answer D is incorrect because the observed total reduction is 20%, but the comparison covers only these pilots. Different work or conditions could change the result, so permanence is not established.

Answer E is incorrect because the pilot still used 30 appraisal hours and 15 failure hours. The evidence does not demonstrate that removing verification would preserve the same results.

 

Question 16

A Scrum Team has completed the visible behavior of a Product Backlog item, but the required security tests in its Definition of Done have failed. A stakeholder wants it counted in the Increment for the Sprint Review. What should the team do?

1. Extend the Sprint until the security tests pass.
2. Count it as Done because the Product Owner can accept the visible behavior.
3. Change the Definition of Done during the review to exclude the failed tests.
4. Return the unfinished item to the Product Backlog.
5. Count the implementation as Done and move only testing into the next Sprint.

Correct Answer: D

 

Correct Answer

Answer D is correct because an item that does not meet the Definition of Done is not part of the Increment. Visible functionality cannot replace the agreed quality measures, and unfinished work returns for future consideration.

Incorrect Answers

Answer A is incorrect because in Scrum, Sprints have fixed lengths. Unfinished work does not justify extending the timebox; it should be made transparent and reconsidered in the Product Backlog.

Answer B is incorrect because agreement by the Product Owner does not override the Definition of Done. The failed security tests mean the item does not satisfy the shared completion standard.

Answer C is incorrect because removing a requirement to make this item appear complete would weaken transparency about its quality. The known failure needs resolution under the applicable standard.

Answer E is incorrect because splitting the accounting this way would claim an Increment that fails the required quality measures. Progress on implementation can be discussed without labeling the item Done.

 

Question 17

Three Scrum Teams develop one product. Each uses a different completion checklist, and integrated increments repeatedly fail shared compatibility tests. The organization has no mandatory Definition of Done. What is the appropriate quality agreement?

1. Each team keeps its checklist and lets integration staff reconcile quality later.
2. Agree and comply with one shared Definition of Done.
3. The Product Owner chooses a different quality threshold for each Sprint.
4. Let an integration group define the completion standard after the teams finish their local work.
5. Use the least demanding checklist to maximize the amount counted as complete.

Correct Answer: B

 

Correct Answer

Answer B is correct because multiple Scrum Teams working on the same product need a common Definition of Done. A shared quality baseline supports compatible Increments and makes completion transparent across teams.

Incorrect Answers

Answer A is incorrect because deferred reconciliation perpetuates incompatible meanings of completion. Local checklists can add practices, but they cannot replace the common Definition of Done needed for the product.

Answer C is incorrect because frequent unilateral threshold changes would undermine a shared understanding of a usable Increment. The teams must establish a common product standard where no organizational one exists.

Answer D is incorrect because a standard introduced only at integration cannot give the teams a shared basis for producing compatible Increments. Multiple Scrum Teams on one product need the common Definition of Done while they do the work.

Answer E is incorrect because a low threshold selected for accounting convenience does not solve compatibility failures. The shared standard must reflect the quality needed for an integrated usable product.

 

Question 18

A hybrid release meets its feature checklist, but the required recovery test cannot restore data within the approved four-hour limit. The release policy requires all mandatory quality tests to pass. What should the project manager recommend?

1. Repeat only the feature tests because they have already shown reliable results.
2. Release with a note that recovery performance will improve later.
3. Release because all user-visible features pass their checks.
4. Count a successful backup as evidence that the recovery test has passed.
5. Hold release while the recovery failure is corrected and retested.

Correct Answer: E

 

Correct Answer

Answer E is correct because recovery is a mandatory quality requirement, not an optional enhancement. Passing feature checks does not satisfy the explicit release policy when a required test has failed.

Incorrect Answers

Answer A is incorrect because retesting unaffected passing features does not resolve the known release blocker. The corrective work and verification must address the failed recovery requirement.

Answer B is incorrect because disclosure does not itself authorize an exception to the stated policy. The project needs a passing test or a legitimately changed requirement before release.

Answer C is incorrect because feature completion addresses only part of the quality requirements. The recovery limit remains mandatory even if users do not see it during normal operation.

Answer D is incorrect because creating a backup and restoring service within four hours are different capabilities. A backup result cannot substitute for the failed end-to-end recovery test.

 

Question 19

A quality audit finds that several approved reporting requirements have no associated test evidence. Existing functional tests all pass. Which two actions close the evidence gap? Choose TWO.

1. Execute the missing verification and retain results linked to the requirements.
2. Use the number of completed user stories as evidence of reporting quality.
3. Use stakeholder walkthrough notes without checking their coverage of the reporting requirements.
4. Repeat the existing functional suite and use additional passing runs to close the finding.
5. Mark the uncovered requirements as passed because the overall test suite is green.
6. Map the missing requirements to suitable verification methods and acceptance criteria.

Correct Answers: A, F

 

Correct Answers

Answer A is correct because a planned test alone is not evidence of a passing result. Executed, traceable verification closes the gap between an approved requirement and demonstrated conformance.

Answer F is correct because a requirement-to-test relationship identifies what must be demonstrated and by which method. Passing unrelated tests cannot establish conformance for the uncovered reporting requirements.

Incorrect Answers

Answer B is incorrect because story counts do not show that the specific reporting conditions were verified. Completion evidence must address the uncovered requirements, not only aggregate activity.

Answer C is incorrect because a walkthrough can be a valid verification method when it actually demonstrates the specified conditions. Unmapped notes do not establish coverage of the requirements identified by the audit.

Answer D is incorrect because repetition increases evidence for covered functions but leaves the same reporting requirements unverified. The missing coverage must be designed and executed rather than inferred from more unrelated passes.

Answer E is incorrect because the suite does not cover those requirements. Its success is evidence only for what it actually verifies, so the proposed inference would be unsupported.

 

Question 20

A team tests an automated document classifier on clean English documents only. The intended deployment includes scanned forms and two other languages. Before judging deployment quality, which actions are needed? Choose TWO.

1. Treat unreadable scans as out of scope without changing the approved requirements.
2. Increase the English sample size while retaining the same document conditions.
3. Accept the classifier if the overall accuracy exceeds the training-set score.
4. Check performance separately against agreed thresholds for the relevant document groups.
5. Add representative samples for the planned languages and scan conditions.
6. Use the model supplier’s generic accuracy claim as the deployment result.

Correct Answers: D, E

 

Correct Answers

Answer D is correct because a combined score can hide weak performance in a smaller group. Group-level evidence tests whether the intended quality requirement holds across the stated deployment conditions.

Answer E is correct because the test set omits known operating conditions. Including them helps expose failures that a clean English-only set cannot reveal before deployment.

Incorrect Answers

Answer A is incorrect because scanned forms are explicitly part of the intended deployment. Excluding them silently changes the tested scope and overstates readiness.

Answer B is incorrect because more observations can reduce sampling uncertainty within that population, but they do not cover missing languages or degraded scans. The coverage gap would remain.

Answer C is incorrect because a comparison with training performance does not establish suitability for the deployment population. The omitted operating conditions remain untested.

Answer F is incorrect because a generic claim may use different documents and thresholds. It cannot replace verification against this project’s actual input conditions and acceptance requirements.

 

Question 21

An agile team’s regression test fails intermittently because its shared test server runs out of storage. Developers repeatedly rerun the suite until it passes and then mark the build green. Which action best restores trustworthy quality evidence?

1. Count every intermittent failure as a confirmed product defect.
2. Quarantine the failing tests and use the reduced suite as sufficient release evidence.
3. Run only storage-independent checks and rely on the previous build’s full pass for the rest.
4. Correct the test-environment failure and rerun the suite under controlled conditions.
5. Allow unlimited reruns because any eventual pass demonstrates acceptable quality.

Correct Answer: D

 

Correct Answer

Answer D is correct because the known environment problem makes the results unreliable. Resolving it and preserving reproducible test conditions creates usable evidence instead of selecting a favorable rerun.

Incorrect Answers

Answer A is incorrect because the evidence identifies a test-server issue. Product defects should still be investigated when supported, but attributing all these failures to the product misclassifies the known cause.

Answer B is incorrect because quarantine may help manage a flaky test during investigation, but excluding required checks does not prove the build meets its quality conditions. The known storage fault should be corrected and the full required verification completed under reliable conditions.

Answer C is incorrect because a past build’s result does not verify the current build, and the reduced suite leaves current behavior untested. Restoring the environment and rerunning the required suite addresses the identified source of unreliable evidence.

Answer E is incorrect because an eventual pass can reflect a transient environment state and does not explain the failures. Repeating until success masks uncertainty about what was actually verified.

 

Question 22

A project compares two support packages. The basic package offers business-hours support and consistently meets every contracted target. The premium package offers round-the-clock support but repeatedly misses its contracted response times. What is the most accurate distinction?

1. Quality can be compared only by counting the number of support features.
2. The premium package has higher quality because it offers more support hours.
3. The basic package has a quality defect because it lacks overnight coverage.
4. The basic package has fewer features but better demonstrated conformance to its requirements.
5. Both packages have the same quality because each serves the same type of customer.

Correct Answer: D

 

Correct Answer

Answer D is correct because feature level and quality conformance are different dimensions. The basic offering can be lower grade while meeting its promises; a richer offering can still perform poorly against its own requirements.

Incorrect Answers

Answer A is incorrect because feature counts omit whether commitments are consistently achieved. The scenario provides direct conformance evidence that is more relevant to quality than the number of features.

Answer B is incorrect because a larger feature set describes the offering level. Repeatedly missing its response commitments is contrary evidence about conformance, regardless of the advertised coverage.

Answer C is incorrect because overnight coverage is not part of its contracted requirements. A missing premium feature is not automatically a defect in the basic package.

Answer E is incorrect because customer category does not establish fulfillment of requirements. Their actual response performance differs, so equal quality cannot be inferred from the audience.

 

Question 23

A predictive project requires an independent process audit before a major release. The only proposed auditor designed the process and approves its exceptions. Which arrangement best meets the independence requirement?

1. Have the process designer select the evidence and ask an independent auditor only to sign the summary.
2. Choose a competent auditor independent of the process responsibilities.
3. Use the process designer because expertise eliminates the need for independence.
4. Replace the process audit with a final product inspection by the same designer.
5. Let the designer audit the process if the sponsor attends the closing meeting.

Correct Answer: B

 

Correct Answer

Answer B is correct because the reviewer must be able to evaluate the controls without assessing their own design or exception decisions. Competence and freedom from the stated operational responsibility satisfy the audit purpose together.

Incorrect Answers

Answer A is incorrect because an independent signature does not make the underlying evaluation independent. The competent auditor needs control of the examination and conclusions rather than endorsing the process owner’s untested selection.

Answer C is incorrect because expertise is valuable but does not remove the self-review conflict. The scenario explicitly requires an independent audit, so technical knowledge alone is insufficient.

Answer D is incorrect because product inspection evaluates specific output and does not fulfill the required independent process audit. Both the audit objective and independence issue would remain.

Answer E is incorrect because attendance at the closing meeting does not change who collected and evaluated the evidence. The auditor would still be reviewing their own process and approvals.

 

Question 24

After a phased deployment, the team confirms that one release corrupted some customer records. Policy requires immediate containment and a verified action to prevent recurrence. Which actions meet these two needs? Choose TWO.

1. Wait for the full root-cause report before isolating the affected release.
2. Isolate the affected release and identify records needing controlled recovery.
3. Close the incident when the corrupted records have been repaired.
4. Continue rollout to obtain a larger defect sample before containing it.
5. Apply the first proposed fix everywhere and verify it after deployment.
6. Investigate the causal path and validate a preventive change before wider rollout.

Correct Answers: B, F

 

Correct Answers

Answer B is correct because containment limits further exposure and defines the population that may require correction. It addresses the immediate consequences while preserving evidence for the investigation.

Answer F is correct because a verified process or product change addresses recurrence. Validation is needed because implementing a plausible fix alone does not demonstrate that the failure mechanism has been removed.

Incorrect Answers

Answer A is incorrect because the policy requires immediate containment of a confirmed failure. Root-cause analysis is also necessary, but it should not delay limiting exposure and identifying the records that need recovery.

Answer C is incorrect because repair corrects affected output but does not establish why the corruption happened. The policy also requires a verified action against recurrence.

Answer D is incorrect because the failure is already confirmed and the policy calls for immediate containment. Deliberately expanding exposure is unnecessary to satisfy that instruction.

Answer E is incorrect because the policy requires a verified preventive action. Broad deployment before validation risks repeating the failure and bypasses the specified sequence.

 

Question 25

A team uses an agreed failure-mode score of severity multiplied by occurrence multiplied by detection difficulty, with higher scores reviewed first unless a mandatory severity override applies. Mode X scores 8, 2 and 3; mode Y scores 5, 4 and 4. Neither has an override. Which review order follows the method?

1. Review Y before X because their scores are 80 and 48 respectively.
2. Review X before Y because severity alone determines this method.
3. Review X first because better detection always increases its score.
4. Review both equally because their scores each sum to 13.
5. Defer both reviews until a mode scores at least 100.

Correct Answer: A

 

Correct Answer

Answer A is correct because multiplication gives X = 8 x 2 x 3 = 48 and Y = 5 x 4 x 4 = 80. Under the stated method and absent an override, Y has the higher priority score.

Incorrect Answers

Answer B is incorrect because severity can be decisive under an override, but none applies here. The explicitly agreed method multiplies all three factors rather than ranking by severity alone.

Answer C is incorrect because the factor is detection difficulty, so harder detection raises the product. X has a lower detection-difficulty value than Y and therefore does not gain priority on that basis.

Answer D is incorrect because the sums are equal, but addition is not the specified scoring rule. The multiplicative method produces different priorities because the factor combinations differ.

Answer E is incorrect because the scenario provides a relative review order and no score-of-100 exemption. Introducing an unstated threshold would bypass the specified method, under which Y is reviewed before X.

 

Question 26

A supplier’s lot passed the agreed acceptance-sampling plan, yet a customer later finds one defective unit. Which statement is technically sound?

1. A passed sample proves that the customer must have damaged the unit.
2. The acceptance criteria should be changed retrospectively to reject that lot.
3. One defective unit proves that the sampling procedure was not followed.
4. Sampling acceptance does not guarantee that every unit in the lot conforms.
5. Conclude that a larger sample would certainly have detected this unit.

Correct Answer: D

 

Correct Answer

Answer D is correct because a sampling plan makes a lot decision with defined statistical risks. A passing sample is compatible with some defective units remaining, so the finding requires investigation rather than a claim of impossibility.

Incorrect Answers

Answer A is incorrect because sampling does not inspect every unit and cannot establish that conclusion. The timing and cause of the defect need evidence from the actual unit and production records.

Answer B is incorrect because a retrospective rule change does not explain the original decision or prevent recurrence. The team must investigate using the recorded plan and manage any prospective revisions openly.

Answer C is incorrect because a correctly executed plan can accept a lot containing defects. The team should check execution, but the single finding does not by itself prove a procedural breach.

Answer E is incorrect because increasing sample size may change sampling risk, but it still need not select the defective unit. The finding calls for investigation of the lot and plan rather than certainty about an unperformed sample.

 

Question 27

Five measurements of a certified 100.0 mm reference are 101.9, 102.0, 102.0, 102.1 and 102.0 mm. Before using this instrument for acceptance testing, what is the primary concern?

1. The readings demonstrate high accuracy because they are close to one another.
2. Widen the product acceptance tolerance by two millimeters to accommodate the instrument.
3. The instrument is imprecise because none of its readings equals 100.0 mm.
4. More readings of production parts will eliminate the two-millimeter offset.
5. The readings are repeatable but biased; investigate calibration and correction.

Correct Answer: E

 

Correct Answer

Answer E is correct because the small spread indicates precision, while their consistent offset from the certified reference indicates poor accuracy. Acceptance decisions would be systematically shifted unless the measurement system is corrected.

Incorrect Answers

Answer A is incorrect because clustering describes precision. Accuracy requires agreement with the certified value, and every reading is about two millimeters above that reference.

Answer B is incorrect because the offset is evidence about the measurement system, not permission to change the product requirement. Altering acceptance tolerance would leave the instrument bias unresolved and could misclassify units.

Answer C is incorrect because failure to match the reference concerns accuracy. The narrow range of 101.9 to 102.1 mm shows the readings are closely repeatable under these conditions.

Answer D is incorrect because additional observations may estimate the bias more precisely, but averaging production readings does not remove a systematic instrument offset.

 

Question 28

Before deployment, an automated weld screen is tested on 100 known defective welds and 900 known conforming welds. It misses 10 of the defective welds and rejects 45 conforming welds. The quality plan requires both a missed-defect rate no greater than 5% of defective welds and a false-reject rate no greater than 6% of conforming welds. Which release assessment applies these criteria correctly?

1. Withhold deployment: the missed-defect rate is 10%, while the false-reject rate of 5% passes.
2. Approve deployment: only about 1.2% of the 865 accepted welds are actually defective.
3. Approve deployment using the 94.5% overall accuracy because conforming welds dominate production.
4. Withhold deployment for false rejection alone; reducing the 45 false rejects will satisfy the plan.
5. Approve deployment: the rates are 1% missed and 4.5% falsely rejected across all 1,000 welds.

Correct Answer: A

 

Correct Answer

Answer A is correct because the criteria have different denominators: 10/100 = 10% of defective welds are missed, whereas 45/900 = 5% of conforming welds are rejected. Both limits are mandatory, so the missed-defect failure prevents deployment despite the acceptable false-reject result.

Incorrect Answers

Answer B is incorrect because 10/(855+10) describes the defective share among accepted welds. It is not the specified missed-defect rate, whose denominator is all 100 known defective welds. A different conditional measure cannot establish compliance with the stated five-percent limit.

Answer C is incorrect because the screen correctly classifies 90 defective plus 855 conforming welds, giving 945/1,000 = 94.5% accuracy. That weighted total can conceal the unacceptable missed-defect rate and does not replace either mandatory conditional criterion.

Answer D is incorrect because forty-five rejected conforming welds represent 5% of the 900 conforming welds, already within the six-percent limit. The failing condition is missed defects; reducing false rejects alone would not correct the 10% missed-defect rate.

Answer E is incorrect because dividing each error count by the entire sample answers a different question. The quality plan conditions each rate on its relevant population: known defective welds for misses and known conforming welds for false rejection.

 

Question 29

An audit found that required peer reviews were skipped in six sampled designs. The team has written a revised review procedure. Closure policy requires both correction of the six original nonconformances and evidence that the revised control is operating on new work. Which evidence is needed before closure? Choose TWO.

1. An approved schedule for the next internal audit.
2. A written commitment from the review lead to apply the procedure on future designs.
3. Independent review results for the six affected designs, including resolution of identified defects.
4. Attendance records showing that every designer received the revised procedure briefing.
5. An assurance that final inspection will detect any design mistakes.
6. Records showing that the revised review control has been used on subsequent designs.

Correct Answers: C, F

 

Correct Answers

Answer C is correct because the original skipped reviews remain nonconformances even if new designs follow the revised procedure. Retrospective review and disposition of those six designs provide separate evidence that the original finding has been corrected.

Answer F is correct because a written procedure establishes intent but does not establish implementation. Actual review records demonstrate that the control is operating in the work it is meant to govern.

Incorrect Answers

Answer A is incorrect because a planned follow-up can organize verification, but it is not completed evidence. The stated closure conditions require correction and observed use of the control before the finding closes.

Answer B is incorrect because the commitment states intent but does not demonstrate implementation or correction of the six skipped reviews. Closure needs the evidence specified by the policy, not only a future promise.

Answer D is incorrect because training records demonstrate communication of the change, but do not prove that the six affected designs were reviewed or that the revised control is operating on subsequent work.

Answer E is incorrect because downstream inspection does not demonstrate compliance with the required peer-review control. It leaves the skipped-review finding unresolved.

 

Question 30

A hybrid project is finalizing a quality plan for a new sensor. Requirements include specified measurement accuracy and a mandatory calibration record for every delivered unit. Which three plan elements make these requirements executable? Choose THREE.

1. Assign responsibility and timing for verification and retention of calibration evidence.
2. Use the batch’s average accuracy as the sole release criterion.
3. Define measurable acceptance limits and the verification method for each requirement.
4. Retain only aggregate pass percentages after shipment.
5. Set the acceptance tolerance from the first batch’s observed spread.
6. Specify the response to failed verification and the conditions for retesting.

Correct Answers: A, C, F

 

Correct Answers

Answer A is correct because mandatory records can be lost even when tests are performed. Naming who verifies and retains evidence, and when, makes the required unit-level proof operational.

Answer C is correct because the team needs to know what result counts as conformance and how to demonstrate it. Linking limits to methods turns the accuracy and record requirements into verifiable conditions.

Answer F is correct because a quality plan must guide what happens when a sensor or record fails. Defined disposition and retest rules prevent inconsistent release decisions under schedule pressure.

Incorrect Answers

Answer B is incorrect because a batch average may conceal individual nonconforming sensors and says nothing about the required calibration record for each unit. It does not make both stated requirements executable.

Answer D is incorrect because an aggregate rate cannot supply the required calibration record for each delivered unit. The plan needs traceable evidence at the unit level.

Answer E is incorrect because the specification already defines required accuracy. Observed process variation may inform capability analysis, but it cannot substitute for approved acceptance limits or unit-level calibration evidence.