Evidence, Data and Digital Competence
What the exam tests
Whether you can commission, interpret and challenge evidence, govern data lawfully, and assure digital services and algorithmic tools.
Core rules
- Evidence quality varies: systematic reviews and evaluations sit above single studies and anecdote. For causal claims, use experiments or credible quasi-experiments such as difference-in-differences, regression discontinuity or matching.
- Correlation is not causation; address selection bias, confounding and measurement error explicitly.
- Data quality dimensions: accuracy, completeness, timeliness, consistency, validity and uniqueness; document them with metadata and data dictionaries.
- Data-protection principles - lawfulness, purpose limitation, minimisation, accuracy, storage limitation, integrity and confidentiality, accountability - apply to analysis, not only to IT. Run a data protection impact assessment for high-risk processing.
- Digital services should rest on user research, reusable components, open standards, security by design and interoperability.
- AI and algorithms need risk classification, human oversight, transparency, bias testing and logging; high-risk uses demand stronger controls.
Worked example
Evaluating a training programme for caseworkers. Reasoning: a simple before-and-after comparison confounds training with seasonal workload. Instead compare trained and untrained offices, collect before-and-after outcome data (resolution time, error rate), and apply difference-in-differences with a parallel-trends check. Link payroll and case data only under a documented legal basis and minimisation; publish aggregate results and limitations. If the sample is small, report confidence intervals rather than a single number.
Traps
Citing one study as settled evidence; conflating statistical with practical significance; collecting personal data just in case; buying systems with no data-portability clause; deploying algorithms without bias or oversight testing.