Ethical Data Collection and Use
What the exam tests
Ethical Data Collection and Use focuses on the front end of the lifecycle: whether data should be collected, how notice and consent work, and whether secondary use is compatible with original expectations. Questions test purpose limitation, minimization, lawful basis, vulnerable populations, provenance, and use-stage safeguards.
Core rules and concepts
- Collect only data that is adequate, relevant, and necessary for a specified purpose; define the purpose before collection, not after.
- Consent must be freely given, specific, informed, unambiguous, and withdrawable. Blanket consent, pre-ticked boxes, and bundled terms weaken validity.
- Secondary use requires a compatibility test: link to original purpose, relationship, data type, consequences, and safeguards. 'Compatible' is not 'convenient.'
- Legitimate-interest balancing requires necessity, proportionality, and an assessment of reasonable expectations; it is not a free pass.
- Third-party and scraped data need provenance checks, contractual restrictions, and notices the individual can actually see.
- Vulnerable people, children, employees, and patients need extra safeguards because consent may be pressured or capacity limited.
- Minimize retention and access; de-identification lowers but does not eliminate risk.
- Document the purpose, compatibility analysis, and safeguards; an undocumented decision is hard to defend.
Worked example
A weather app collects precise location to show local forecasts, then sells audience segments to advertisers. Collection for forecasts may be necessary, but sale is a new purpose. The exam answer should test compatibility: users expect forecast use, not advertising sale; precise location is sensitive and can reveal home, work, and habits; the power asymmetry is high. Safeguards: separate opt-in for advertising, coarse location by default, no sale of precise data, clear notice at collection, easy withdrawal, and a DPIA. If the business cannot meet those conditions, do not collect or sell.
Traps that cost marks
Assuming notice equals consent. Treating aggregate or pseudonymized data as risk-free. Using 'research' or 'innovation' to justify unlimited reuse. Ignoring collection bias that becomes a use-stage harm. Forgetting withdrawal and deletion.