Scenario Lab — Scenarios 4–6
Scenario 4 — Accurate Numbers, Dishonest Chart
Difficulty: Standard
Decisive clue: graphic scale creates the distortion; values are unchanged.
A KPI improves from 98.1% to 98.4%. A manager asks for a chart whose vertical axis starts at 98.0 so the small improvement looks dramatic. Every plotted value is correct; the slide will support a major funding decision.
Model answer
This is misleading visualization. Use an honest scale and context that communicates the true magnitude of change.
Why: ethical handling includes interpretation and communication; numeric correctness does not excuse visual manipulation.
Weaker: “The chart is acceptable because the points are correct.”
Changed fact: if the manager instead selected only a tiny favorable observation window, Timing becomes the primary category.
Operational mapping: Leading KA Data Warehousing & Business Intelligence; supporting Data Governance. Roles: analyst/BI developer, report owner, executive consumer, Governance/quality reviewer.
Source: §3.4.2, pp. 59–60.
Scenario 5 — The Hunch-and-Search Analysis
Difficulty: Standard
Decisive clue: a preferred conclusion drives filtering and interpretation.
A manager believes remote employees are less productive. An analyst repeatedly changes filters, excludes teams and tries date ranges until one subset supports the belief. Contrary results are omitted.
Model answer
The primary issue is bias — hunch and search / biased use of data.
A more ethical analysis should define the question/method transparently, use representative data, examine contrary evidence, document exclusions and test alternatives rather than shape the method around the desired answer.
Weaker: “It is only Timing because date ranges changed.” Timing appears, but the stronger clue is the systematic search for confirmation.
Changed fact: if the analysis method were sound but the sample omitted a major region, biased sampling methodology would be the more specific bias mechanism.
Operational mapping: Leading KA DW/BI; supporting Data Governance and Data Quality where evidence fitness is questioned. Roles: analyst/data scientist, business sponsor, peer/review function and Governance.
Source: §3.4.4, pp. 60–61.
Scenario 6 — The Dataset with No Trustworthy History
Difficulty: Difficult
Decisive clue: unknown origin, meaning and change history after transformation/integration.
Four legacy systems are integrated for an eligibility model. The final table has no reliable lineage, conflicting field definitions, uneven source Data Quality and no record of prior remediation/transformation changes. The team wants to proceed because the table loads successfully.
Model answer
Treat this as transformation/integration ethical risk. Do not use the dataset for consequential decisions until provenance, lineage, meaning, quality, ownership/protection and auditable remediation/change history are governed.
Technical loading proves movement, not preserved meaning or trustworthiness.
Weaker: “Run the model and monitor accuracy later.” That makes affected people the test environment despite known provenance/meaning risks.
Changed fact: if lineage, definitions, quality, ownership and change history were sound but the data merely arrived late, the primary issue would shift to the actual operational/timing problem.
Operational mapping: Leading KA Data Integration & Interoperability; supporting Metadata Management, Data Quality and Data Governance. Roles: integration owner, Metadata/lineage team, Data Stewards/Owners and model/business consumers.
Source: §3.4.5, pp. 61–62.