These 12 CBDA data quality questions test how you validate data, trace its lineage, protect personal data, and keep bias out of results. Most come from the Source Data domain, with a few from Analyze Data, reporting and data strategy. Each is an original scenario with one best answer, and the answer key cites the Guide section behind it.
How these CBDA data quality questions are built
The 12 questions are original, not exam dumps. Like the real exam, each is a short business scenario asking what the analyst should do first, next or best. Most are medium difficulty, and two need a small calculation.
The key Guide ideas behind the set:
- 2.2.4 Validate Data. Business validation: stakeholders with authority approve sources and set accuracy criteria. Technical validation: testing accuracy, completeness, consistency, uniqueness and timeliness.
- 2.2.1 and 2.2.2. Privacy and security requirements can make data unavailable or unfit for use.
- 2.2.3 Collect Data. Data lineage: where data comes from, how it changed, where it is stored.
- 2.3.2 Prepare Data. Weighting corrects bias; statistical fixes can clash with business rules.
- Techniques 3.4, 3.10, 3.17, 3.2. Data dictionaries standardize definitions, ETL can mask attributes, surveys protect anonymity, visuals stay unbiased.
- 2.6.3 Data Strategy. Governance, architecture, security and metadata management are distinct.
For the wider picture of this domain, read our guide to sourcing and validating data and the companion set of CBDA Source Data practice questions.
| Sub-topic | Questions |
|---|---|
| Data quality characteristics | 1, 2 |
| Business vs technical validation | 3 |
| Data lineage and source choice | 4 |
| Data dictionary and definitions | 5 |
| Privacy, masking and data fitness | 6, 7 |
| Bias in collection and preparation | 8, 9, 10 |
| Ethical reporting | 11 |
| Organizational data strategy | 12 |
CBDA exam at a glance
| Fact | Detail |
|---|---|
| Questions | 75 multiple-choice, scenario-based |
| Time | 2 hours |
| Source Data domain weight | 15% |
| Exam fee | US$405 (IIBA members US$250) |
| Passing score | Not published by IIBA |
Last verified: October 2026
Practice Questions
Question 1
Freshcart, a grocery delivery company, has extracted order records to analyze weekly volumes by region. The analytics lead wants evidence that the extracted figures are accurate.
Which check BEST tests accuracy?
- A. Confirm every mandatory order field has a value
- B. Compare sample order totals on screen with database records
- C. Confirm the order date column holds only valid dates
- D. Confirm the extract covers exactly the requested weeks
Question 2
Northgrid Utilities requested meter readings for April to June 2026. Profiling the extract gave these results.
| Check | Result |
|---|---|
| Total rows | 48,000 |
| Distinct reading IDs | 46,800 |
| Rows dated outside April–June | 0 |
| Rows with a null meter ID | 0 |
| Non-date values in date field | 0 |
Which data quality issue should the analyst raise with the data owner?
- A. Timeliness: the extract may miss the requested quarter
- B. Completeness: some readings lack a meter ID
- C. Consistency: the date field mixes formats
- D. Uniqueness: 1,200 rows (2.5%) repeat a reading ID
Question 3
At Harbor Mutual, an insurer, technical quality tests on three claims data sources have passed. The claims director, who owns the sources, has not reviewed them, and no one has agreed how accuracy will be judged.
What should the analyst do NEXT?
- A. Ask the director to approve the sources and accuracy criteria
- B. Start the analysis, since the technical tests passed
- C. Ask the data scientist who ran the tests to sign off
- D. Add more sources so any single one matters less
Question 4
Pinecrest Home, a furniture retailer, needs revenue data for a promotion study. The finance warehouse and a marketing data mart, which reshapes data into a friendlier form, report last quarter's revenue 4% apart.
What should the analyst do FIRST?
- A. Average the two figures to balance the gap
- B. Use the data mart, since it is easier to analyze
- C. Trace each figure's origin and transformations
- D. Drop the question until the teams agree
Question 5
Bluefin Telecom combines billing and app data. Billing calls a customer active after a payment in 90 days; the app team, after a login in 30 days. Their reports show very different active-customer counts.
Which technique would BEST resolve this problem?
- A. A data dictionary entry with the rule and validation
- B. A data flow diagram of customer data between systems
- C. A survey of how often customers use the service
- D. A business simulation of revenue per definition
Question 6
Medway Health must link patient admissions across two systems to study readmissions. Analysts do not need patient identities, but both files contain national ID numbers.
What should the analyst recommend for the transform step?
- A. Load full records and restrict dashboard access
- B. Remove every record containing a national ID
- C. Group patients into age bands
- D. Mask the IDs so records still link
Halfway there. If you want a longer workout, create a free account and open the CBDA exam simulator. It holds a full 75-question set with exam and practice modes. certificate.tips is our product.
Question 7
Trailfit, a fitness app, stores users' GPS running routes. An insurance partner asks the analytics team to use those routes to build user risk profiles.
What should the analyst do FIRST?
- A. Use the data, since it was collected in normal app use
- B. Review privacy requirements with compliance and architecture
- C. Run the analysis, then strip names before sharing
- D. Let data scientists decide by predictive value
Question 8
Brightpath, an HR software firm, drafts an engagement survey that asks for employee numbers and lets managers see their team's answers. Several employees say they would not answer honestly.
Which change would BEST improve the quality of the responses?
- A. Add questions to cover each theme in depth
- B. Make every question open-ended free text
- C. Remove identifiers and make responses anonymous
- D. Survey only top-rated performers
Question 9
Lakeshore Bank surveyed customers about its mobile app. Urban customers are 60% of the base but gave 80% of responses.
| Segment | Share of customers | Share of responses | Satisfied |
|---|---|---|---|
| Urban | 60% | 80% | 80% |
| Rural | 40% | 20% | 50% |
Which overall satisfaction figure should the analyst report?
- A. 74%, reflecting the answers actually received
- B. 65%, the simple average of the two segments
- C. 80%, since urban customers are the majority
- D. 68%, weighting segments by share of customers
Question 10
Corvia Lending is analyzing approved loans; 6% lack income. The data scientist proposes mean imputation to keep the distribution intact. Policy forbids approval without verified income.
What should the analyst do?
- A. Flag the rule conflict and investigate how income was recorded
- B. Accept mean imputation as statistically sound
- C. Drop the records, since 6% will not change results
- D. Replace missing values with zero to keep all records
Question 11
At Solstice Couriers, on-time delivery rose from 96.1% to 97.0% this quarter. Before a board meeting, a vice president asks the analyst to start the chart's y-axis at 96% so the gain looks large.
What should the analyst do?
- A. Comply, since the vice president owns the message
- B. Drop the chart and mention the change verbally
- C. Use a fair scale and state the real size of the gain
- D. Add a second y-axis showing both scales
Question 12
Ridgeway Logistics streams data from thousands of truck sensors to a dozen analytics teams. Teams cannot tell which datasets exist, what fields mean, or who maintains them, so work is duplicated.
Which data strategy consideration addresses this problem MOST directly?
- A. Data security
- B. Metadata management
- C. Data architecture
- D. Shared sampling standards
Ready for the next round? The CBDA exam simulator has a full 75-question set across all six domains, in a timed exam mode or an explained practice mode. Create a free account to start.
Answer Key and Explanations
1. Correct answer: B. Accuracy means the data is correct and represents what the source intended. The Guide suggests checking it by comparing numbers in a front-end system with the same data pulled from the database. Source: Guide to Business Data Analytics, 2.2.4 Validate Data Why the others are wrong: A tests completeness; C tests consistency; D tests timeliness.
2. Correct answer: D. 48,000 rows minus 46,800 distinct IDs leaves 1,200 repeats, or 2.5%. The Guide assesses uniqueness by looking for duplicates, and every other check is clean. Source: Guide to Business Data Analytics, 2.2.4 Validate Data Why the others are wrong: A, no rows fall outside the period; B, no meter IDs are null; C, the date field is clean.
3. Correct answer: A. Business validation means stakeholders with authority over the sources approve them and set the acceptance criteria for judging accuracy. Passing technical tests does not replace that approval. Source: Guide to Business Data Analytics, 2.2.4 Validate Data Why the others are wrong: B skips business validation; C, the data scientist has no authority over the sources; D adds scope without approval.
4. Correct answer: C. Tracing origin, transformations and storage is data lineage, which the Guide ties to judging quality. It also notes that a secondary source that has already been reshaped may need a quality check against the central warehouse before you choose it. Source: Guide to Business Data Analytics, 2.2.3 Collect Data; 2.2.1 Plan Data Collection Why the others are wrong: A hides the cause; B, ease of use is not quality; D is premature.
5. Correct answer: A. A data dictionary standardizes what a data element means across sources. For analytics, the Guide extends its basic entries (name, aliases, meaning, description) to track business rules and validation for derived elements. Source: Guide to Business Data Analytics, 3.4 Data Dictionary Why the others are wrong: B shows data movement, not meaning; C collects new data; D models outcomes without fixing the term.
6. Correct answer: D. Masking attributes such as social security or credit card numbers is one of the transformation types the Guide lists for ETL. Masked keys still link records. EU guidance treats pseudonymisation as an example of data protection by design. Source: Guide to Business Data Analytics, 3.10 Extract, Transform, and Load (ETL) Why the others are wrong: A exposes IDs upstream; B breaks the linkage; C leaves the IDs untouched.
7. Correct answer: B. The Guide says privacy rules can make data unavailable or a dataset unfit for use, and that analysts work with the architecture team on legal and privacy compliance. Under GDPR purpose limitation, reuse for an unrelated purpose needs a compatibility check, and sharing with a third party may need explicit consent. Source: Guide to Business Data Analytics, 2.2.1 Plan Data Collection; 2.2.2 Determine the Data Sets Why the others are wrong: A, collection does not permit every use; C, removing names later does not make the use lawful; D, predictive value is not a privacy test.
8. Correct answer: C. Guide survey practices include respecting sensitivities and ensuring anonymity. People who fear exposure give guarded answers. Source: Guide to Business Data Analytics, 3.17 Survey and Questionnaire Why the others are wrong: A does not remove the fear; B keeps the identifiers and manager visibility in place; D adds sampling bias.
9. Correct answer: D. Weighted: 0.6 × 80% + 0.4 × 50% = 68%. Unweighted: 0.8 × 80% + 0.2 × 50% = 74%, inflated by urban responses. The Guide names weighting as a bias correction, with sample weights for unequal sampling. Source: Guide to Business Data Analytics, 2.3.2 Prepare Data Why the others are wrong: A keeps the bias; B ignores the customer mix; C drops rural customers.
10. Correct answer: A. The Guide's own example: mean or median imputation may be statistically sound yet conflict with business rules the analyst can highlight. Missing income on approved loans signals a recording problem to investigate with stakeholders. Source: Guide to Business Data Analytics, 2.3.2 Prepare Data Why the others are wrong: B ignores the rule; C assumes the gaps are random; D misstates income.
11. Correct answer: C. Under "present unbiased analysis", the Guide warns that a different y-axis scale on the same data can bias the audience. Visuals should help people see the insight, not steer them. Source: Guide to Business Data Analytics, 3.2 Business Visualizations Why the others are wrong: A puts preference above honest reporting; B avoids the issue; D adds confusion.
12. Correct answer: B. The Guide defines metadata management as administering information about the data assets an organization holds. It notes that tracking data gets harder as streaming volume and project count grow. Source: Guide to Business Data Analytics, 2.6.3 Data Strategy Why the others are wrong: A protects data but does not describe it; C sets storage and integration models; D is a sampling practice, not a catalog.
How to score this set
- 11–12 correct: Solid. Move on to mixed-domain practice.
- 8–10 correct: Reread 2.2.4 Validate Data and 2.3.2 Prepare Data; most misses come from mixing up quality characteristics.
- Below 8: Study the whole Source Data domain again, then retry. Our CBDA analytics strategy quiz covers the data strategy side.
To practise at full exam length, take the 75-question set in the CBDA exam simulator. You can create a free account to save your scores and review the explanations.
What candidates say about data quality on the CBDA
Candidates describe a test of judgement more than recall. The author of our CBDA certification overview flagged about 20 of 75 questions and found the real exam harder than question banks, with more analytical decisions than definitions. Mufaddal Haidermota, who earned the CBDA in December 2022, wrote on Medium that he relied on per-domain mock questions, full-length tests, mind maps and cheat sheets. A Techcanvass story about Ivan, a former developer, credits mind maps and two to three daily study hours. Trainer LN Mishra of Adaptive US advises practitioners to answer questions daily, take timed full mocks, and review every wrong answer.
FAQ
What data quality characteristics does the CBDA Guide name?
Section 2.2.4 Validate Data describes accuracy, completeness, consistency, uniqueness and timeliness, and gives an example of how to assess each. Exam questions usually describe a check and ask which characteristic it tests. Learn the example checks, not just the names.
Does the CBDA test GDPR or other privacy laws?
The Guide does not teach any specific law. It does expect analysts to treat privacy as a constraint when planning and choosing data, and to involve architecture and compliance teams. General principles such as purpose limitation and data minimisation help you reason through those scenarios.
How is data lineage different from data mapping?
Data lineage traces where data comes from, how it changes, and where it is stored, so you can judge its quality. Data mapping is a technique that defines, attribute by attribute, how source fields move into a target repository. A data map supports traceability, which helps when you investigate lineage issues.
Sources
- Certification in Business Data Analytics (CBDA) — IIBA, checked October 2026
- Certification Fees — IIBA, checked October 2026
- CBDA Certification Handbook — IIBA, checked October 2026
- Regulation (EU) 2016/679 (General Data Protection Regulation) — EUR-Lex, Publications Office of the EU, 27 April 2016
- Principles of the GDPR — European Commission, accessed October 2026
- A guide to the data protection principles — UK Information Commissioner's Office, updated March 2026
- Business Data Analytics Certification — Mufaddal Haidermota, Medium, January 2023
- CBDA Certification Journey: Ivan — Techcanvass, 2024
- Mastering the CBDA Exam — LN Mishra, Adaptive US, October 2024