Data representation in the life cycle
The data work maps the system's data to the identified stakeholders and records provenance, governance, collection and the biases the data can cause.
What clause 5.2.3 asks
This part ties the data to the people. The first question is whether the data needed exists, or can be gathered or generated. The record then covers where each set came from, who governs it, how it was gathered, what it is for and what bias it could introduce. This is not a schema design. It is a test of whether the data can stand for the people the system will affect.
Why it matters
The clause makes the data work answerable to the stakeholder set. A data set is not representative in the abstract; it is representative of particular people with particular attributes. A team that starts from the data will describe its columns. A team that starts from the stakeholders will ask which people are missing and what that means for the output. The provenance record is what lets a later reader judge whether the data can carry the conclusion placed on it.
What good looks like
Build the map from the stakeholder reference set outward. For each stakeholder group, record whether the data can identify it, how many records represent it, and what is missing. Then record the source and the governance for each data set, the collection method and the biases the data can carry. Keep the map beside the profile version it belongs to, and redo it after retraining.
Evidence an assessor asks for
The assessor will compare the stakeholder set with the data map. A group in the set with no data row is an unrepresented group, and the burden is on the team to explain why that is acceptable.
- Stakeholder to data map Each stakeholder group mapped to the data that represents it, with coverage and gaps
- Provenance record Source, governance, collection method and first use for each data set
- Bias note The biases each data set can cause, with the evidence
- Availability record What exists, what must be collected and what must be created
Where it commonly fails
- The map lists data columns, not stakeholder groups
- Provenance stops at the vendor name
- The data is assumed to be representative because it is large
What to do before publication
Start from the stakeholder reference set, not from the data catalogue. The map is a test of representation, not an inventory.
Read across
- 5.2.2 Stakeholder identification in the life cycle
- 6.3 Outputs
- 7.1 Purpose
- 7.4.2.3 Data map
- 9.2.2 Evaluation of design and outputs: inputs
← 5.2.2 Stakeholder identification in the life cycle5.2.4 Risk and impact assessment in the life cycle →