CDMP DMF Practice Test 40
Question 390
A governance review considers the following: The customer's tier is gold in billing but silver in the loyalty platform on the same date. Which proposal would create an avoidable risk?
- Separate short-term issue containment from durable root-cause correction.
- Treat disagreement as harmless because both values are allowed codes.
- Protect relevant audit records while investigating data problems.
- Retain a rollback or recovery plan before making destructive data changes.
Explanation
Correct Answer
B:Treat disagreement as harmless because both values are allowed codes. Consistency is agreement across representations under comparable meaning and time. A sound response instead: Reconcile tier definitions and synchronization logic using a defined authoritative source.
Incorrect Answers
A:Separate short-term issue containment from durable root-cause correction. An immediate workaround rarely proves the defect cannot recur. The case-specific response is: Reconcile tier definitions and synchronization logic using a defined authoritative source.
C:Protect relevant audit records while investigating data problems. Evidence handling supports retrospective control testing. The case-specific response is: Reconcile tier definitions and synchronization logic using a defined authoritative source.
D:Retain a rollback or recovery plan before making destructive data changes. Recoverability reduces the consequences of incorrect remediation. The case-specific response is: Reconcile tier definitions and synchronization logic using a defined authoritative source.
Question 391
A business owner identifies this issue: A team proposes proprietary event fields that cannot be interpreted by other platforms. What is the MOST appropriate data management response?
- Assess patterns against consistency, latency, autonomy, cost, governance and security needs.
- Publish target enterprise data architecture with shared identity principles and integration boundaries.
- Agree enterprise-level business concepts and relationships before mapping physical application schemas.
- Define canonical field semantics and versioned interface contracts with consumers.
Explanation
Correct Answer
D:Define canonical field semantics and versioned interface contracts with consumers. Standards reduce integration ambiguity when implementations use different technologies. Evidence to obtain: Conformance tests show field meanings and changes remain compatible with consuming systems.
Incorrect Answers
A:Assess patterns against consistency, latency, autonomy, cost, governance and security needs. This response fits a separate situation (An organization compares central warehouse, decentralized data products and a virtualization layer for analytics.), but does not resolve the stated case. The targeted response is: Define canonical field semantics and versioned interface contracts with consumers.
B:Publish target enterprise data architecture with shared identity principles and integration boundaries. This response fits a separate situation (Customer identifiers differ across sales, billing and support, preventing consistent reporting.), but does not resolve the stated case. The targeted response is: Define canonical field semantics and versioned interface contracts with consumers.
C:Agree enterprise-level business concepts and relationships before mapping physical application schemas. This response fits a separate situation (Two departments use the term account for unrelated concepts and merge records incorrectly.), but does not resolve the stated case. The targeted response is: Define canonical field semantics and versioned interface contracts with consumers.
Question 392
A data analyst must confirm the cause of this issue: A model incorrectly allows each order to have many unrelated customer owners. Which check provides the strongest diagnostic evidence?
- Review model annotations, glossary bindings and conflict-resolution decisions.
- Inspect subtype constraints, shared keys and invalid category combinations.
- Trace model-field lineage, interface consumers and staged migration tests.
- Test model examples for zero, one and multiple permitted owners.
Explanation
Correct Answer
D:Test model examples for zero, one and multiple permitted owners. This test checks the relevant business and technical failure rather than assuming its cause. Cardinality and optionality encode business relationship constraints. Expected control result: Invalid multi-owner orders are prevented and valid customer orders remain supported.
Incorrect Answers
A:Review model annotations, glossary bindings and conflict-resolution decisions. This check would investigate another issue (Teams use inconsistent definitions for status and state across logical models.), not the immediate case. The direct diagnostic is: Test model examples for zero, one and multiple permitted owners.
B:Inspect subtype constraints, shared keys and invalid category combinations. This check would investigate another issue (A party table mixes person birthdate with organization registration number, leaving ambiguous rules.), not the immediate case. The direct diagnostic is: Test model examples for zero, one and multiple permitted owners.
C:Trace model-field lineage, interface consumers and staged migration tests. This check would investigate another issue (An enterprise data model removes a field still used in multiple integration contracts.), not the immediate case. The direct diagnostic is: Test model examples for zero, one and multiple permitted owners.
Question 393
A data analyst must confirm the cause of this issue: A nightly billion-event aggregation exceeds single-machine memory and now misses its SLA. Which check provides the strongest diagnostic evidence?
- Check representative identifier mappings, exception cases and contract tests.
- Inspect raw landing zones, transformation versions and reconciliation checks.
- Test source outages, federated query plans, data permissions and cache behavior.
- Inspect partition sizes, shuffle metrics, retries and end-to-end output reconciliation.
Explanation
Correct Answer
D:Inspect partition sizes, shuffle metrics, retries and end-to-end output reconciliation. This test checks the relevant business and technical failure rather than assuming its cause. Big-data computation relies on parallelism and careful data partitioning rather than unlimited node memory. Expected control result: The distributed job completes correctly within the required window and reconciles counts.
Incorrect Answers
A:Check representative identifier mappings, exception cases and contract tests. This check would investigate another issue (One interface calls a field supplier_id while another calls it vendor_code with different identifier rules.), not the immediate case. The direct diagnostic is: Inspect partition sizes, shuffle metrics, retries and end-to-end output reconciliation.
B:Inspect raw landing zones, transformation versions and reconciliation checks. This check would investigate another issue (An analytics platform loads source extracts into a warehouse then performs transformations there.), not the immediate case. The direct diagnostic is: Inspect partition sizes, shuffle metrics, retries and end-to-end output reconciliation.
C:Test source outages, federated query plans, data permissions and cache behavior. This check would investigate another issue (An organization wants live access to distributed records without copying every source table.), not the immediate case. The direct diagnostic is: Inspect partition sizes, shuffle metrics, retries and end-to-end output reconciliation.
Question 394
The data program reviews the following failure: A postal address field is 100 percent complete but several entries are no longer current. Which change should be prioritized?
- Add input validation with clear user feedback and exception override governance.
- Identify the upstream source-process failure and implement a preventive correction.
- Assess completeness and timeliness separately, then choose remediation based on use.
- Measure required-field completeness and introduce appropriate collection checkpoints.
Explanation
Correct Answer
C:Assess completeness and timeliness separately, then choose remediation based on use. High performance on one data-quality dimension does not establish overall fitness. Evidence to obtain: The scorecard shows distinct dimension measures and decision-relevant targets.
Incorrect Answers
A:This response fits a separate situation (Bad dates are entered during onboarding and corrected only after monthly reporting failures.), but does not resolve the stated case. Add input validation with clear user feedback and exception override governance.
B:Identify the upstream source-process failure and implement a preventive correction. This response fits a separate situation (An analyst manually fixes the same missing supplier IDs every week.), but does not resolve the stated case. The targeted response is: Assess completeness and timeliness separately, then choose remediation based on use.
D:Measure required-field completeness and introduce appropriate collection checkpoints. This response fits a separate situation (Loan applications often omit annual income even though policy requires it for underwriting.), but does not resolve the stated case. The targeted response is: Assess completeness and timeliness separately, then choose remediation based on use.
Question 395
For this situation, which combined intervention and validation MOST directly addresses the issue? An analytics platform loads source extracts into a warehouse then performs transformations there.
- Action: Govern landing, transformation logic, lineage and compute costs in a controlled ELT workflow. Verification: Raw inputs and reproducible transformation outputs are traceable and validated.
- Action: Use change data capture with durable offsets, ordering assumptions and replay handling. Verification: Changed rows propagate with acceptable lag and no missing or duplicate business effects.
- Action: Choose integration methods based on latency, consistency, volume and operational risk. Verification: The fraud service receives timely events while tax reporting reconciles its complete batch.
- Action: Use idempotent processing and durable identifiers with explicit delivery semantics. Verification: Replayed events do not create duplicate postings and genuine events still post.
Explanation
Correct Answer
A:Action: Govern landing, transformation logic, lineage and compute costs in a controlled ELT workflow. Verification: Raw inputs and reproducible transformation outputs are traceable and validated. Both the intervention and measurable outcome are directly relevant. ELT moves transformation into the target platform but does not remove data quality obligations.
Incorrect Answers
B:This plan addresses a different situation (A customer platform needs low-latency updates without repeatedly scanning entire operational tables.), not the targeted problem. Here, action and evidence must match: Action: Govern landing, transformation logic, lineage and compute costs in a controlled ELT workflow.
C:This plan addresses a different situation (A daily tax report and a live fraud decision service share one nightly transfer mechanism.), not the targeted problem.
D:This plan addresses a different situation (An event consumer processes a payment notification twice after retry and creates duplicate ledger postings.), not the targeted problem. Here, action and evidence must match: Action: Govern landing, transformation logic, lineage and compute costs in a controlled ELT workflow.
Question 396
A data management auditor evaluates this situation: A scorecard reports 99 percent valid rows but excludes an entire failed daily feed. Which observable finding supports successful remediation?
- Regulatory users can reproduce lineage and assess permitted use of each observation.
- Quality scores include expected records and flag unavailable populations.
- Every accepted line references a valid invoice or has an owned exception.
- Equivalent datasets are evaluated against the same approved country-code version.
Explanation
Correct Answer
B:It tests the action: Combine validity metrics with completeness of data delivery and denominator controls.
Incorrect Answers
A:Regulatory users can reproduce lineage and assess permitted use of each observation. This would verify a different control (A scientific dataset has fine precision but lacks metadata required for the regulator's audit.), not the one in the question. The required evidence is: Quality scores include expected records and flag unavailable populations.
C:Every accepted line references a valid invoice or has an owned exception. This would verify a different control (An invoice line appears without an invoice header after a partial load.), not the one in the question. The required evidence is: Quality scores include expected records and flag unavailable populations.
D:Equivalent datasets are evaluated against the same approved country-code version. This would verify a different control (Different pipelines enforce different allowable values for the same governed country field.), not the one in the question. The required evidence is: Quality scores include expected records and flag unavailable populations.
Question 397
A reviewer asks for proof of improvement: Bad dates are entered during onboarding and corrected only after monthly reporting failures. Which result MOST convincingly demonstrates that the control works?
- Invalid dates are blocked at entry and legitimate exceptions are traceable.
- Quality scores include expected records and flag unavailable populations.
- Every accepted line references a valid invoice or has an owned exception.
- Regulatory users can reproduce lineage and assess permitted use of each observation.
Explanation
Correct Answer
A:It tests the action: Add input validation with clear user feedback and exception override governance.
Incorrect Answers
B:Quality scores include expected records and flag unavailable populations. This would verify a different control (A scorecard reports 99 percent valid rows but excludes an entire failed daily feed.), not the one in the question. The required evidence is: Invalid dates are blocked at entry and legitimate exceptions are traceable.
C:Every accepted line references a valid invoice or has an owned exception. This would verify a different control (An invoice line appears without an invoice header after a partial load.), not the one in the question. The required evidence is: Invalid dates are blocked at entry and legitimate exceptions are traceable.
D:Regulatory users can reproduce lineage and assess permitted use of each observation. This would verify a different control (A scientific dataset has fine precision but lacks metadata required for the regulator's audit.), not the one in the question. The required evidence is: Invalid dates are blocked at entry and legitimate exceptions are traceable.
Question 398
Before changing procedures, the team investigates: Contract PDFs are retained indefinitely after business use with no destruction rules. Which review or test MOST directly locates the relevant failure?
- Review content classes, retention triggers, holds and disposal authorizations.
- Compare sampled extraction results with source documents and exception logs.
- Examine record metadata, authenticity checks and immutable version history.
- Sample tagging accuracy and compare searches against curated representative records.
Explanation
Correct Answer
A:Review content classes, retention triggers, holds and disposal authorizations. This test checks the relevant business and technical failure rather than assuming its cause. Document and content management governs unstructured information across its lifecycle. Expected control result: Records have retention schedules and disposal evidence while active holds are respected.
Incorrect Answers
B:Compare sampled extraction results with source documents and exception logs. This check would investigate another issue (A team runs OCR on handwritten forms and loads extracted values directly into an account ledger.), not the immediate case. The direct diagnostic is: Review content classes, retention triggers, holds and disposal authorizations.
C:Examine record metadata, authenticity checks and immutable version history. This check would investigate another issue (An archived approval email is needed as evidence in a dispute but its author and date are missing.), not the immediate case. The direct diagnostic is: Review content classes, retention triggers, holds and disposal authorizations.
D:Sample tagging accuracy and compare searches against curated representative records. This check would investigate another issue (A repository mixes invoices, procedures and complaints under inconsistent free-text tags.), not the immediate case. The direct diagnostic is: Review content classes, retention triggers, holds and disposal authorizations.
Question 399
A business owner identifies this issue: Business stakeholders cannot agree which things should be treated as customers, accounts or products. What is the MOST appropriate data management response?
- Represent missingness according to explicit semantic rules and validate downstream behavior.
- Model phone numbers as atomic repeating child records with appropriate identifiers.
- Consider controlled denormalization with refresh ownership and reconciliation.
- Create a conceptual model of core business entities and relationships without implementation details.
Explanation
Correct Answer
D:Create a conceptual model of core business entities and relationships without implementation details. Conceptual models communicate business understanding before logical and physical detail. Evidence to obtain: Stakeholders approve shared concepts and relationship definitions.
Incorrect Answers
A:This response fits a separate situation (A registration field is blank, but downstream calculations treat unknown age as zero.), but does not resolve the stated case.
B:This response fits a separate situation (A contact table stores three comma-separated telephone numbers in one column.), but does not resolve the stated case.
C:Consider controlled denormalization with refresh ownership and reconciliation. This response fits a separate situation (A read-heavy dashboard suffers repeated expensive joins but operational source integrity is sound.), but does not resolve the stated case. The targeted response is: Create a conceptual model of core business entities and relationships without implementation details.
Question 400
In reviewing this data risk, select the LEAST defensible practice: A customer table identifies rows by mutable email address, causing broken relationships after updates.
- Inspect lineage and source context before changing a derived report.
- Retain a rollback or recovery plan before making destructive data changes.
- Use a frequently changing descriptive attribute as the permanent relational key.
- Document which dataset version was used in a material decision.
Explanation
Correct Answer
C:Use a frequently changing descriptive attribute as the permanent relational key. A stable primary key should preserve record identity across ordinary attribute changes. A sound response instead: Introduce a stable key and manage mutable natural identifiers as governed attributes.
Incorrect Answers
A:Inspect lineage and source context before changing a derived report. Provenance analysis distinguishes source errors from transformations. The case-specific response is: Introduce a stable key and manage mutable natural identifiers as governed attributes.
B:Retain a rollback or recovery plan before making destructive data changes. Recoverability reduces the consequences of incorrect remediation. The case-specific response is: Introduce a stable key and manage mutable natural identifiers as governed attributes.
D:Document which dataset version was used in a material decision. Version records support reproducible decisions. The case-specific response is: Introduce a stable key and manage mutable natural identifiers as governed attributes.